BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//EE - ECPv5.10.0//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:EE
X-ORIGINAL-URL:https://ee.iisc.ac.in
X-WR-CALDESC:Events for EE
BEGIN:VTIMEZONE
TZID:Asia/Kolkata
BEGIN:STANDARD
TZOFFSETFROM:+0530
TZOFFSETTO:+0530
TZNAME:IST
DTSTART:20220101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20230120T213000
DTEND;TZID=Asia/Kolkata:20230120T223000
DTSTAMP:20260404T023037
CREATED:20230117T002035Z
LAST-MODIFIED:20230117T002448Z
UID:240253-1674250200-1674253800@ee.iisc.ac.in
SUMMARY:EE Faculty Colloquium by Prof. Kaushik Basu
DESCRIPTION:Venue: MMCR\, EE \nTime: 4pm\, 20 January 2023 \nAbstract:Silicon Carbide MOSFETs (SiC MOSFETs) fall into wide band gap (WBG) power devices. These devices are commercially available in the voltage range of 600-3300V and compete with the state-of-the-art Si-insulated gate bipolar junction transistors (IGBTs). Superior material properties  of SiC MOSFET lead to smaller die sizes. This results in faster switching transients and lower switching loss. However\, it excites device and circuit parasitic that may lead to prolonged oscillation\, high device stress\, spurious turn-on and EMI-related issues. So\, the benefit of using SiC MOSFET as a power device comes with numerous design challenges resulting in slow commercial adaptation. It is predicted that the overall market share of WBG devices (SiC and GaN together) will be roughly 10% of the total market for power semiconductors by 2025. To overcome the design challenges and fully utilise the benefits of fast-switching SiC MOSFETs\, a better understanding of switching dynamics is essential. However\, the switching dynamics of SiC MOSFET are different compared to its Si counterpart due to the highly non-linear device characteristics and participation of circuit parasitic in the process. In this talk\, we will discuss our recent work on developing an analytical model of the switching dynamics for hard and soft transitions of SiC MOSFET by simplifying the complex non-linear dynamics predicted by the behavioural model. The developed model\, given the device-related parameters extracted from the data sheet\, estimated or measured circuit parasitic\, can predict lost switching energy\, rate of change of device voltage etc.\, necessary for a successful power converter design through computation with sufficient accuracy. Based on this model a Python based interactive software tool has been developed. The results of this work are applied to two WBG-based advanced power converter development: A 3kW onboard charger for 2 Wheelers and a 200kW SiC-based inverter for high-bandwidth power amplifiers.Speaker’s Bio:Kaushik Basu received the BE. Degree from the Bengal Engineering and Science University\, Shibpore\, India\, in 2003\, the M.Sc. degree in electrical engineering from the Indian Institute of Science (IISc)\, Bangalore\, India\, in 2005\, and the PhD degree in electrical engineering from the University of Minnesota\, Minneapolis\, in 2012\, respectively. He was a Design Engineer with Cold Watt India in 2006 and an Electronics and Control Engineer with Dynapower Corporation USA from 2013-to 15. He is an Associate Professor in the Department of Electrical Engineering at the IISc. He served as the Technical Program Committee Vice-Chair of IEEE ECCE 2019 and 2022. In 2019\, he received the Prof. Priti Shankar Teaching Award from IISc. As a co-author\, he received the Second Best Prize Paper Award from IEEE Transactions on Transportation Electrification in 2021.  He is IEEE senior member and is the founding chair of both IEEE PELS and IES Bangalore Chapter. He is an Associate Editor of IEEE Transactions on Power Electronics\, IEEE Transactions on Industrial Electronics\, and Springer Journal of Power Electronics. His research interests include most aspects of Power Electronic converter design from a few kW to a few MW for applications ranging from space\, grid integration of renewables and storage to fast charging of electric vehicles.
URL:https://ee.iisc.ac.in/event/ee-faculty-colloquium-by-prof-kaushik-basu/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20230112T203000
DTEND;TZID=Asia/Kolkata:20230112T213000
DTSTAMP:20260404T023037
CREATED:20230109T051231Z
LAST-MODIFIED:20230109T051231Z
UID:240242-1673555400-1673559000@ee.iisc.ac.in
SUMMARY:Thesis Colloquium of Sayan Paul
DESCRIPTION:Thesis Title: Pulse Width Modulation Techniques of Inverter Fed Split-Phase Machine Drive in Linear and Overmodulation Regions \nDegree registered: PhD \nGuide: Prof. Kaushik Basu \nDate and Time: 12th January 2023\, 03:00 PM \nPlace: MMCR EE\, \nMeeting Link: Teams Meeting Link \nMulti-phase machines (MPMs) have more than three windings in their stator\, rotor\, or both. With the broader adoption of power-electronic converters for efficient driving of the machines\, MPMs are gaining attention in different applications due to their certain advantages over three-phase machines. One such advantage is higher fault tolerance due to higher phase redundancy\, which makes it suitable for safety-critical applications like electric vehicles (EVs)\, ship propulsions\, electric aircraft\, etc. Another advantage is that MPMs allow power splitting across multiple phases. Hence\, the power rating per phase drive unit becomes low\, making it suitable for high-power applications like railway traction\, pumps\, compressors\, etc. Recent literature also proposes using the same multi-phase converter fed MPM\, otherwise used for propulsion\, as an onboard battery charger; it substantially reduces space\, weight\, and cost. During charging mode\, the leakage inductance of the machine provides the required inductance for the grid connection\, and MPM’s higher degrees of freedom are used to lock the rotor electronically. An asymmetrical six-phase machine (ASPM) or split-phase machine is one such MPMs and is very common in EVs. This thesis aims to devise the pulse-width modulation (PWM) techniques of a two-level six-phase inverter fed ASPM to improve the overall drive efficiency. \nASPM has two sets of balanced three-phase windings\, which are spatially shifted by 30 degrees (electrical angle). In one of the popular configurations\, the two three-phase winding sets are connected in star fashion with two isolated neutral points. This machine is conventionally analyzed in two two-dimensional (2D) orthogonal subspaces. One of these subspaces is associated with electromagnetic energy transfer and torque production. The other subspace doesn’t transfer energy through the air gap and the equivalent circuit in this plane\, consisting of winding resistance and leakage inductance\, provides a low impedance. Therefore\, excitation of this non-energy-transferring subspace causes a large current and associated copper loss. Any PWM technique of ASPM aims to synthesize the desired voltages in the energy-transferring plane and minimize the applied voltage in the non-energy-transferring subspace. \nLinear modulation techniques (LMTs) of ASPM apply zero average voltage in the non-energy-transferring subspace and synthesize the desired voltages in the energy-transferring plane on an average over a switching cycle. It is expected that these LMTs should avoid more than two switching transitions of an inverter leg within a carrier period to limit the instantaneous switching loss. Through an innovative approach\, our work finds a way to account for all possible infinitely many LMTs that follow the rule of at most two transitions per leg. But each of them results in a different current ripple performance. Ripple current is inevitable in PWM converters and should be minimized through modulation to reduce the associated copper loss. The total ripple current RMS of ASPM is contributed by both energy-transferring and non-transferring planes. One machine parameter also impacts this performance: the ratio of high-frequency inductances in these two subspaces. For all reference voltage vectors and the whole feasible range of the machine parameter\, our work finds the techniques with minimum current ripple (RMS) among the above infinite possible LMTs through numerical optimization. A hybrid PWM strategy is proposed with these optimal techniques\, which outperforms all existing techniques regarding current ripple performance. \nOvermodulation (OVM) techniques of ASPM attain higher voltage gain in energy-transferring subspace than LMTs by applying non-zero average voltage in the non-energy transferring subspace. This operation doesn’t cause any torque ripple\, but the applied voltage in non-energy transferring subspace should be minimised to reduce unwanted current and associated loss. The existing OVM technique in the literature minimizes this average voltage from the space-vector perspective with a pre-defined set of four active vectors. To find the best technique\, one needs to perform the above minimization problem with all possible sets of active vectors\, which can give higher voltage gain. So\, this requires the evaluation of a large number of cases. In this thesis\, we have formulated the above minimization problem in terms of average voltage vectors of two three-phase inverters\, where active vectors need not be specified beforehand. Thus\, the analysis is more general. Following the above analysis\, eight switching sequences in one part and two in another part of the OVM region are derived\, which attain the minimum average voltage injection in the non-energy transferring subspace. \nAlthough the above OVM sequences apply the same average voltages in the two subspaces\, they have different high-frequency ripple currents due to different switching strategies. The current ripple study of the OVM techniques of ASPM is missing in the literature. Hence\, one of our works in the thesis studies the current ripple performances of the above optimal PWM sequences in the OVM region\, which apply minimum average voltage in the non-energy-transferring subspace. We find the sequence with the best switching current ripple performance for a given reference vector in the OVM region and the machine parameter. After that\, a PWM technique is proposed\, which substantially improves the high-frequency current ripple performance (RMS) compared to two existing OVM techniques for a given machine parameter value. \nFinally\, simple carrier-comparison-based implementation methods of the proposed LMTs and OVM sequences are found. The six-phase inverter is split into two three-phase inverters\, and the proposed strategy implements the PWM sequences per three-phase inverter basis. In carrier-based implementations\, the duty signal of the top switch of an inverter leg is compared with a triangular carrier. The bottom switch’s gating pulse complements the top switch’s pulse with a fixed dead time. The duty signal of the top switch of any leg has two components- a modulation signal and a common-mode signal. Two 180-degree phase-shifted carrier signals are required to implement the proposed sequences. The energy-transferring plane of ASPM is divided into twenty-four equivalent sectors; the carrier signals and the expressions of modulation and common-mode signals differ from one sector to another. Henceforth\, a sector-independent algorithm is proposed in this thesis to derive these duty signals that substantially reduce the computational burden. \nThe proposed techniques are validated through simulation in MATLAB/Simulink and experiments on a hardware prototype at a power level of 4 kW.
URL:https://ee.iisc.ac.in/event/thesis-colloquium-of-sayan-paul/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20230111T223000
DTEND;TZID=Asia/Kolkata:20230111T233000
DTSTAMP:20260404T023037
CREATED:20230110T223726Z
LAST-MODIFIED:20230110T223914Z
UID:240244-1673476200-1673479800@ee.iisc.ac.in
SUMMARY:Lecture by Prof. Anubha Gupta
DESCRIPTION:Department of Electrical Engineering\, Indian Institute of Science  \n\n\nand  \n\n\nIEEE Signal Processing Society\, Bangalore Chapter  \n\n\n  \n\n\ncordially invite you to a lecture on  \n\n\nProper Definition of Dirichlet Conditions and Convergence of Fourier Representations  \n\n\n By  \n\n\n Prof. Anubha Gupta  \n\n\nIndraprastha Institute of Information Technology -Delhi  \n\n\n Date and time: January 11\, 2023\, 5 PM  \n\n\n Venue: Multimedia Classroom\, Department of Electrical Engineering (EE)\, IISc.  \n\n\n Coffee will be served during the talk.  \n\n\n Abstract:  Fourier theory is the backbone of Signal Processing (SP) and Communication Engineering. It has been widely used in almost all branches of science and engineering in numerous applications since its inception. However\, Fourier representations such as Fourier series (FS) and Fourier transform (FT) may not exist for some signals that fail to fulfil a predefined set of Dirichlet conditions (DCs). We note a subtle gap in explaining these conditions as available in the popular SP literature. For example\, the original DCs require a signal to have bounded variations over one time period for the convergence of FS\, where there can be at most countably infinite number of maxima and minima\, and at most countably infinite number of discontinuities of finite magnitude. However\, a large body of the literature replaces this statement with the requirements of a finite number of maxima and minima over one time period\, and a finite number of finite discontinuities over one time period. Due to the latter\, some signals fulfilling the original DCs are incorrectly perceived as not having convergent FS representation. A similar problem holds in the description of DCs for the FT. This talk is based on our recent lecture notes published in IEEE Signal Processing Magazine (Sep 2022 issue)\, wherein we provide the required clarifications and a lucid but precise description and explanation on the DCs along with a lot of suitable examples.  \n\n\n Brief Bio:  \n\n\n \n\n\nAnubha Gupta (anubha@iiitd.ac.in) received her B.Tech and M.Tech from Delhi University\, India in 1991 and 1997 in Electronics and Communication Engineering. She received her PhD. from Indian Institute of Technology (IIT)\, Delhi\, India in 2006 in Electrical Engineering. She did her second master’s as a full-time student from the University of Maryland\, College Park\, USA from 2008-2010 in Education. She worked as Assistant Director with the Ministry of Information and Broadcasting\, Govt. of India (through Indian Engineering Services) from 1993 to 1999 and\, as faculty at NSUT-Delhi (2000-2008) and IIIT-Hyderabad (2011-2013)\, India. Currently\, she is working as a Professor at IIIT-Delhi\, where she served as the Dean\, Academic Affairs from June 2019 to June 2020. She has authored/co-authored more than 100 technical papers in scientific journals and conferences. She received SERB POWER Fellowship\, 2021 from DST. Govt. of India. She has published research papers in both engineering and education. A lot of exciting work is being taken up in her lab: SBILab (Lab: http://sbilab.iiitd.edu.in/index.html \n\n\n\n\n\nSBILab – Indraprastha Institute of Information Technology\, Delhi\nSignal Processing and Biomedical Imaging Lab. Led By Prof. Anubha Gupta of IIIT-Delhi (Started in Jan 2014) SBILab focuses on Signal Processing areas including applications of Wavelet Transforms\, Machine (Deep) Learning\, and Compressed Sensing\, Sparse Reconstruction\, fMRI/EEG/MRI/DTI Signal and Image Processing\, Genomics Signal Processing\, Signal Processing for Communication Engineering\, and …\nsbilab.iiitd.edu.in\n\n\n\n\n\n). Her research interests include applications of machine learning in cancer genomics\, cancer imaging\, biomedical signal and image processing including fMRI\, MRI\, EEG\, ECG signal processing\, and Wavelets in deep learning. Dr. Gupta is a senior member of IEEE Signal Processing Society (SPS) and a member of IEEE Women in Engineering Society. She is a technical committee member of BISP committee of IEEE SPS Society for Jan 2022– Dec 2024.   \n\n\n  \n\n\nHost: Prof. Chandra Sekhar Seelamantula\, EE\, IISc.  \n\n\n  \n\n\nAll are invited. 
URL:https://ee.iisc.ac.in/event/lecture-by-prof-anubha-gupta/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20230106T153000
DTEND;TZID=Asia/Kolkata:20230106T163000
DTSTAMP:20260404T023037
CREATED:20221227T011631Z
LAST-MODIFIED:20221227T011659Z
UID:240210-1673019000-1673022600@ee.iisc.ac.in
SUMMARY:Seminar by Prof. Niladri Chakraborty
DESCRIPTION:Title: Emulator-based Microgrid studies vis a vis Some Indian Renewable interconnected Microgrids. \nTime: 10 am 6/1/23 \nVenue: EE MMCR \nAbstract: Mini / Micro Grid in Stand Alone or Grid integrated forms are coming up very fast in the conventional power system. Many a time\, these are renewable sources integrated. Design considerations\, Plant automation\, Remote monitoring and O & M strategies play important roles in their operational exploitation. However\, for economics and systematic\, technically sound energy businesses\, it is prudent to have emulator-based studies done before these systems are commercially realized. Attempts will be made to showcase an emulator-based microgrid available in the Department of Power Engineering\, Jadavpur University\, Kolkata. At the same time\, generalized Design considerations\, Plant automation\, Remote monitoring\, O & M strategies and System studies will be briefly highlighted.  Introduction of different types of renewable integrated Mini / Microgrids encountered during consultancy services or for technical studies available elsewhere in the pan-Indian domain will be made perceptible to the audience. \n Bio: Prof Niladri Chakraborty was born in Kolkata on 27/08/1964. He received his Bachelor’s (1986) and Masters’s Degree (1989) in Electrical Engineering from Jadavpur University\, Kolkata\, India. He was the recipient of the prestigious Commonwealth Scholarship in the United Kingdom and obtained the Diploma of Imperial College and Doctor of Philosophy from the University of London in 1999. He was elected as the youngest Dean of the Faculty of Engineering and Technology at Jadavpur University in 2010. He has served twice as Head of the Department of Power Engineering and as Joint Director of the School of Energy Studies at Jadavpur University. He was often elected to the Executive Council and The Court\, the highest policy-making academic body of Jadavpur University. He has also received Scholarships from the Royal Society (UK)-DST and Abdul Kalam Award\, besides many other awards from numerous academic activities. Dr Chakraborty has served as one of the Lead Scientists in India’s National Communication to United Nations on Climate Change (NATCOM\, India\, UNFCCC) and a Specialized Energy Expert in formulating the Futuristic Energy Action Plan for the Government of West Bengal\, India. He has also completed many research projects obtained from different funding agencies and earned revenue for Jadavpur University as a consultant. \n Prof. Chakraborty’s recent research interests include Energy Economics\, Renewable integrated Microgrid\, Environmental Measurement and Material Modelling. Dr Chakraborty has in his credit to be the author of about 75 referred journal publications and nearly 130 International Conference Publications. Prof. Chakraborty has guided 13 PhDs (4 more are in writing-up status) and 41 Masters of Engineering Thesis.
URL:https://ee.iisc.ac.in/event/seminar-by-prof-niladri-chakraborty/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20230102T213000
DTEND;TZID=Asia/Kolkata:20230102T223000
DTSTAMP:20260404T023037
CREATED:20221229T225034Z
LAST-MODIFIED:20221229T225125Z
UID:240213-1672695000-1672698600@ee.iisc.ac.in
SUMMARY:Lecture by Prof. Manoj Saranathan
DESCRIPTION:Department of Electrical Engineering\, Indian Institute of Science\nand\nIEEE Signal Processing Society\, Bangalore Chapter\ncordially invite you to a lecture on \nAdvances in imaging and segmentation of thalamic nuclei with applications \nby \nProf. Manoj Saranathan\, Professor of Radiology at UMass Chan Medical School in Worcester\, Massachusetts \nDate and time: January 2\, 2023\, 4 PM\nVenue: Multimedia Classroom\, Department of Electrical Engineering (EE)\, IISc. \nCoffee will be served at 3.45 PM. \nAbstract: The thalamus is a subcortical deep brain structure increasingly implicated in a number of neurodegenerative and neuropsychiatric conditions. It is subdivided into regions called nuclei which are linked to specific cortical and sensory regions of the brain. However\, thalamic nuclei have largely been ignored in most imaging studies as they are mostly invisible in conventional MRI. In this talk\, I will present our work on MRI methods to improve visualization of thalamic nuclei as well as cutting edge thalamic nuclei segmentation methods. I will then show examples of characterization of atrophy of specific thalamic nuclei in neurodegenerative diseases as well as for cutting edge neurosurgical treatment of epilepsy and essential tremor. \nBiography of the speaker: Manoj Saranathan is an MRI physicist with over twenty-five years of experience in MR physics\, pulse sequence development\, image reconstruction\, and image processing spanning industry and academia. His current research interests are focused on ultra high-resolution imaging and segmentation of deep brain structures like thalamus and hippocampus and the specificity of their involvement in pathology such as alcoholism\, multiple sclerosis\, Alzheimer’s disease\, and essential tremor. Another area of interest is high spatio-temporal resolution dynamic contrast enhanced MRI for quantification of physiologic function and cancer. One of his methods\, DISCO\, is now a product available on all GE MRI scanners since 2017\, and is widely used for prostate and breast imaging worldwide. He has a PhD in Bioengineering from the University of Washington\, Seattle and is currently a Professor of Radiology at UMass Chan Medical School in Worcester\, Massachusetts. \nHosts: Prof. P. S. Sastry and Prof. Chandra Sekhar Seelamantula\, EE\, IISc \nAll are invited.
URL:https://ee.iisc.ac.in/event/lecture-by-prof-manoj-saranathan/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20230102T163000
DTEND;TZID=Asia/Kolkata:20230102T173000
DTSTAMP:20260404T023037
CREATED:20221226T032757Z
LAST-MODIFIED:20221226T032848Z
UID:240207-1672677000-1672680600@ee.iisc.ac.in
SUMMARY:Thesis Defence of Mr. Ruturaj Gavaskar
DESCRIPTION:Title: On Plug-and-Play Regularization using Linear Denoisers.\nDegree Registered: PhD\nGuide: Prof Kunal Narayan Chaudhury\nDate: Jan 2\, 2023.\nTime: 11:00 am. \nVenue: Online.\nMS Teams: https://tinyurl.com/bdz95wmw\n(meeting ID: 483 231 041 151; Passcode: QDF8WS) \nAbstract: The problem of inverting a given measurement model comes up in several computational imaging applications. For example\, in CT and MRI\, we are required to reconstruct a high-resolution image from incomplete noisy measurements\, whereas in superresolution and deblurring\, we try to infer the ground truth from low-resolution or blurred images. Traditionally\, this is done by minimizing f + φ\, where f is a data-fidelity (or loss) function that is determined by the acquisition process\, and φ is a regularization (or penalty) function that is based on a subjective prior on the target image. The solution is obtained numerically using iterative algorithms such as ISTA or ADMM. \nWhile several forms of regularization and associated optimization methods have been proposed in the imaging literature over the last few decades\, the use of denoisers (aka denoising priors) for image regularization is a relatively recent phenomenon. This has partly been triggered by advances in image denoising in the last 20 years\, leading to the development of powerful image denoisers such as BM3D and DnCNN. In this thesis\, we look at a recent protocol called Plug-and-Play (PnP) regularization\, where image denoisers are deployed within iterative algorithms for image regularization. PnP consists of replacing the proximal map — an analytical operator at the core of ISTA and ADMM — associated with the regularizer φ with an image denoiser. This is motivated by the intuition that off-the-shelf denoisers such as BM3D and DnCNN offer better image priors than traditional hand-crafted regularizers such as total variation. While PnP does not use an explicit regularizer\, it still makes use of the data-fidelity function f. However\, since the replacement of the proximal map with a denoiser is ad-hoc\, the optimization perspective is lost — it is not clear if the PnP iterations can be interpreted as optimizing some objective function f + φ. Remarkably\, PnP reconstructions are of high quality and competitive with state-of-the-art methods. Following this\, researchers have tried explaining why plugging a denoiser within an inversion algorithm should work in the first place\, why it produces high-quality images\, and whether the final reconstruction is optimal in some sense.\nIn this thesis\, we try to answer such questions\, some of which have been the topic of active research in the imaging community in recent years. Specifically\, we consider the following questions. \n1. Fixed-point convergence: Under what conditions does the sequence of iterates generated by a PnP algorithm converge? Moreover\, are these conditions met by existing real-world denoisers? \n2. Optimality and objective convergence: Can we interpret PnP as an algorithm that minimizes f + φ for some appropriate φ? Moreover\, does the algorithm converge to a minimizer of this objective function? \n3. Exact and robust recovery: Under what conditions can we recover the ground truth exactly via PnP? And is the reconstruction robust to noise in the measurements? \nWhile early work on PnP has attempted to answer some of these questions\, many of the underlying assumptions are either strong or unverifiable. This is essentially because denoisers such as BM3D and DnCNN are mathematically complex\, nonlinear and difficult to characterize. A first step in understanding complex nonlinear phenomena is often to develop an understanding of some linear approximation. In this spirit\, we focus our attention on denoisers that are linear. In fact\, there exists a broad class of real-world denoisers that are linear and whose performance is quite decent; examples include kernel filters (e.g. NLM\, bilateral filter) and their symmetrized counterparts. This class has a simple characterization that helps to keep the analysis tractable and the assumptions verifiable. Our main contributions lie in resolving the aforementioned questions for PnP algorithms where the plugged denoiser belongs to this class. We summarize them below. \n• We prove fixed-point convergence of the PnP version of ISTA under mild assumptions on the measurement model. \n• Based on the theory of proximal maps\, we prove that a PnP algorithm in fact minimizes a convex objective function f + φ\, subject to some algorithmic modifications that arise from the algebraic properties of the denoiser. Notably\, unlike previous results\, our analysis applies to non-symmetric linear filters. \n• Under certain verifiable assumptions\, we prove that a signal can be recovered exactly (resp. robustly) from clean (resp. noisy) measurements using PnP regularization. As a more profound application\, in the spirit of classical compressed sensing\, we are able to derive probabilistic guarantees on exact and robust recovery for the compressed sensing problem where the sensing matrix is random. An implication of our analysis is that the range of the linear denoiser plays the role of a signal prior and its dimension essentially controls the size of the set of recoverable signals. In particular\, we are able to derive the sample complexity of compressed sensing as a function of distortion error and success rate. \nWe validate our theoretical findings numerically\, discuss their implications and mention possible future research directions.
URL:https://ee.iisc.ac.in/event/thesis-defence-of-mr-ruturaj-gavaskar/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20230102T133000
DTEND;TZID=Asia/Kolkata:20230104T223000
DTSTAMP:20260404T023037
CREATED:20221128T001914Z
LAST-MODIFIED:20221128T003251Z
UID:240127-1672666200-1672871400@ee.iisc.ac.in
SUMMARY:Workshop on Protection and Stability of Renewable Dominated Power Grids
DESCRIPTION:Click here for the poster.\nTopics that will be covered in the workshop:\n\nOverview of Photovoltaic and Wind Generations\nConverter Controls for Renewables\nGrid Connection Requirements\nImpact of Renewables on Fault Analysis and Protection\nImpact of Renewables on System Stability\nTraining on Renewable Modelling in PSCAD\nAC Microgrids\nDC Microgrids\nCase Studies\n\nThe list of speakers: (Click here for the schedule.)\n\nProf Sukumar Brahma\, Clemson University\, USA\nProf Prasad Enjeti\, Texas A&M University\, USA\nDr Ritwik Majumder\, Mathworks\nProf Vinod John\, IISc\nProf Kaushik Basu\, IISc\nProf Gurunath Gurrala\, IISc\nDr. Vishnu Mahadeva Iyer\nProf Sarasij Das\, IISc\n\n 
URL:https://ee.iisc.ac.in/event/workshop-on-protection-and-stability-of-renewable-dominated-power-grids/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221230T163000
DTEND;TZID=Asia/Kolkata:20221230T173000
DTSTAMP:20260404T023037
CREATED:20221205T053506Z
LAST-MODIFIED:20221205T053816Z
UID:240135-1672417800-1672421400@ee.iisc.ac.in
SUMMARY:Invited Talk by Prof Anjan Bose
DESCRIPTION:Title: Maintaining Reliability and Resiliency while Decarbonizing the Power Grid \nAbstract: The world has focused strongly\, through the Paris Accords\, on ridding the electricity generation mix of fossil fuels and replacing that with sustainable non-carbon resources. The emerging new generation mix\, however\, is changing the characteristics of the grid thus requiring the modification of the ways we plan\, design\, operate and control the grid. The new engineering tools and procedures must be ready as the penetration of renewable energy resources keeps increasing. These methods to provide highly reliable electricity to society were developed over decades but now must be updated more urgently. In this address we outline where these threats are coming from and how we maintain the highly reliable systems despite the changes. In addition\, the increase in extreme weather events is necessitating more efficient ways to recover from grid damage thus increasing the resiliency of the system in addition to maintaining reliability. \nAuthor’s Bio: Prof Anjan Bose (Life Fellow\, IEEE) received the B.Tech. degree from IIT Kharagpur\, Kharagpur\, India\, the M.S. degree from the University of California\, Berkeley\, CA\, USA\, and the Ph.D. degree from Iowa State University\, Ames\, IA\, USA. He has worked for industry\, academe\, and government for 40 years in electric power engineering. He is currently a Regents Professor and holds an Endowed Distinguished Professor in power engineering at Washington State University\, Pullman\, WA\, USA\, where he also served as the Dean for the College of Engineering and Architecture\, from 1998 to 2005. He is a member of the U.S. National Academy of Engineering and a Foreign Fellow of the Indian National Academy of Engineering. He received the Herman Halperin Award and the Millennium Medal from the IEEE and was recognized as a Distinguished Alumnus by IIT Kharagpur and Iowa State University
URL:https://ee.iisc.ac.in/event/invited-talk-by-prof-anjan-bose/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221220T163000
DTEND;TZID=Asia/Kolkata:20221220T173000
DTSTAMP:20260404T023037
CREATED:20221215T225256Z
LAST-MODIFIED:20221215T225552Z
UID:240183-1671553800-1671557400@ee.iisc.ac.in
SUMMARY:Thesis Colloquium of Mr. Debarpan Bhattacharya
DESCRIPTION:Degree Registered: MTech Resesarch\nTitle: A Learnable Distillation Approach For Model-agnostic Explainability With Multimodal Applications\nAdvisor: Prof. Sriram Ganapathy\nDate and Time: Tuesday\, Dec 20th \, 11am\nVenue:  MMCR\, Electrical Engineering\, IISc.\n\nAbstract: Deep neural networks are the most widely used examples of sophisticated mapping functions from feature space to class labels. In the recent years\, several high impact decisions in domains such as finance\, healthcare\, law and autonomous driving\, are made with deep models. In these tasks\, the model decisions lack interpretability\, and pose difficulties in making the models accountable. Hence\, there is a strong demand for developing explainable approaches which can elicit how the deep neural architecture\, despite the astounding performance improvements observed in all fields\, including computer vision\, natural language processing\, generates the output decisions.\n\nThe current frameworks for explainability of model learning are based on gradients (eg. GradCAM\, guided-gradCAM\, Integrated gradients etc) or based on locally linear assumptions (eg. LIME). Some of these approaches require the knowledge of the deep model architecture\, which may be restrictive in many applications. Further\, most of the prior works in the literature highlight the results on a set of small number of examples to illustrate the performance of these XAI methods\, often lacking statistical evaluation.\n\nThis talk proposes a new approach for explainability based on mask estimation approaches\, called the Distillation Approach for Model-agnostic Explainability (DAME). The DAME is a saliency-based explainability model that is post-hoc\, model-agnostic\, and applicable to any architecture/domain. The DAME is a student-teacher modeling approach\, where the teacher model is the original model for which the explainability is sought\, while the student model is the mask estimation model. The input sample is augmented with various data augmentation techniques to produce numerous samples in the immediate vicinity of the input. Using these samples\, the mask estimation model is learned to learn the saliency map of the input sample for predicting the labels. A distillation loss is used to train the DAME model\, and the student model tries to locally approximate the original model. Once the DAME model is trained\, the DAME generates a region of the input (either in space or in time-domain for images and audio samples\, respectively) that best explains the model predictions. \n\nWe also propose an evaluation framework\, for both image and audio tasks\, where the XAI models are evaluated in a statistical framework on a set of held-out of examples with the Intersection-over-Union (IoU) metric. We have validated the DAME model for vision\, audio and biomedical tasks. Firstly\, we deploy the DAME for explaining a ResNet-50 classifier pre-trained on ImageNet dataset for the object recognition task. Secondly\, we explain the predictions made by ResNet-50 classifier fine-tuned on Environmental Sound Classification (ESC-10) dataset for the audio event classification task. Finally\, we validate the DAME model on the COVID-19 classification task using cough audio recordings. In these tasks\, the DAME model is shown to outperform existing benchmarks for explainable modeling. \n\nThe talk will also illustrate results on various multimodal datasets and conclude with a discussion on the limitations of the DAME approach along with the potential future directions.\n\n######################################################################\n\nAll are welcome\,\n\n\n\n 
URL:https://ee.iisc.ac.in/event/thesis-colloquium-of-mr-debarpan-bhattacharya/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221213T203000
DTEND;TZID=Asia/Kolkata:20221213T213000
DTSTAMP:20260404T023037
CREATED:20221212T221650Z
LAST-MODIFIED:20221212T221650Z
UID:240170-1670963400-1670967000@ee.iisc.ac.in
SUMMARY:Lecture by Dr. Raveesh Magod
DESCRIPTION:Title: Powering the Future – Research and Current Trends Shaping the Next Generations of Power Management ICs \nDate and Time: 13th December 2022 at 3pm \nVenue: MMCR EE \nAbstract: Power management has been a key enabler for a variety of ever increasing modern day electronic applications like smartphones\, data center servers\, electric vehicles and smart grids to name a few. High-efficiency\, small form-factor\, reliability and lower cost will continue to be critical requirements of power regulation for such applications. In this talk\, details about what constitutes this broad area of power management along with research vectors and market trends defining the next-generation ICs in power conversion is presented. Specifically i) ultra-low power consumption converters ii) Gallium Nitride (GaN) and Silicon Carbide (SiC) based power converters and iii) low noise and low EMI power conversion\, are identified and described as three major focus areas. Personal research contributions\, insights into latest research and industry products\, and future performance trendlines are presented in detail. \nSpeaker Bio: Raveesh Magod received the M.S. and Ph.D. degree in Electrical engineering from Arizona State University\, Tempe\, AZ\, USA\, in 2014 and 2018 respectively. In 2017\, he joined Jack Kilby Labs\, Texas Instruments\, Dallas\, TX\, USA\, where he is Member of Technical Staff and has been involved in R&D of wide range of power converter products ranging from nanopower voltage regulators to high power-density DC-DC converters and recently has been focusing on GaN based power converters. From 2015 to 2016\, he was an analog design intern at Texas Instruments\, Tucson\, AZ\, USA\, where he designed low-power voltage supervisors and low quiescent current LDOs. From 2010 to 2012\, he was a Design Engineer at Sankalp Semiconductor (A HCL technologies company)\, Hubli\, India\, developing low-power CMOS interface solutions. He was the co-recipient of A.K. Chowdhary Best Paper award at the International conference on VLSI Design\, 2021 and has five granted/pending U.S. patents to his name. He also serves as a technical program committee member for the CICC and a reviewer for multiple reputed IEEE journals.
URL:https://ee.iisc.ac.in/event/lecture-by-dr-raveesh-magod/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221209T213000
DTEND;TZID=Asia/Kolkata:20221209T223000
DTSTAMP:20260404T023037
CREATED:20221207T000049Z
LAST-MODIFIED:20221207T000049Z
UID:240140-1670621400-1670625000@ee.iisc.ac.in
SUMMARY:Seminar by Dr. Subhash Lakshminarayana
DESCRIPTION:Title: IoT-Based Load-Altering Attacks Against Power Grids\nDate: 09-12-2022\, Friday\nTime: 4:00 PM – 5:00 PM\nVenue: MMCR of EE Department\, Room No: C-241 (First floor) \nAbstract: Large-scale Internet of Things (IoT)-based load-altering attacks can have a major impact on power grid operations such as causing unsafe frequency excursions and destabilizing the grid’s control loops. In this talk\, I will present my recent research on enhancing the resilience of power grids to IoT-based load-altering attacks. First\, I will present a novel analytical framework to investigate the impact of IoT-based static/dynamic load-altering attacks (S/DLAAs) on the power grid’s dynamic response using the theory of second-order dynamical systems. The results help identify the victim nodes from which that attacker can launch the most impactful attacks and offer insights into how the temporal fluctuations of load and renewable energy sources impact the grid’s vulnerabilities to LAAs. Finally\, I will present results on the detection and mitigation of such attacks. \nBiography: Dr. Subhash Lakshminarayana is an associate professor at the University of Warwick. His research interests include cyber-physical system security and wireless communications. He serves as an associate editor in the IET Smart Grid journal. His research is funded by Innovate UK\, EPSRC-PETRAS National Centre of Excellence for Cybersecurity of IoT Systems UK\, and the EUTOPIA European Alliance.
URL:https://ee.iisc.ac.in/event/seminar-by-dr-subhash-lakshminarayana/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221123T153000
DTEND;TZID=Asia/Kolkata:20221123T163000
DTSTAMP:20260404T023037
CREATED:20221115T032528Z
LAST-MODIFIED:20221115T032641Z
UID:240120-1669217400-1669221000@ee.iisc.ac.in
SUMMARY:Thesis Colloquium of Mr  Bidhan Biswas
DESCRIPTION:Degree  Registered:   Ph.DThesis Title:                Short Circuit and Open Circuit Natural Frequencies of 3-Φ Transformers: Derived Analytical Expressions and its ApplicationsGuide:                          Prof. L. Satish\nDate & Time:               Wednesday\, 23 November 2022\, at 10:00 AM\nVenue:                         Hybrid Mode\, on MS TEAMS and held in HV Seminar Hall\n\n\nAbstract: Frequency Response Analysis (FRA) method is perhaps the most sensitive tool that can detect even the slightest of winding/core movements. High sensitivity\, non-invasiveness\, non-destructiveness\, and on-site capability are some of its salient features – making it an ideal monitoring and detection tool. The existence of Standards (IEEE\, IEC\, and CIGRE) is ample testimony of its global acceptance and superior detection capabilities. The principle of detection is based on observance of a deviation between two measured FRAs which implies a possible fault. Naturally\, the next logical step is to analyse these deviations to determine the type of fault\, estimate the extent of damage and its severity\, and as a bonus\, predict its location\, if possible. However\, even after three decades of existence\, arriving at these inferences  is still at the research level. Even though there is a consensus among all the standards on FRA test/measurement procedures\, best-suited terminal connections\, cable layout\, grounding practices\, etc.\, they remain largely silent regarding interpretation and diagnostics.\n\n\n\nA detailed analysis of literature compiled in Chapter 1 reveals that perhaps lack of a mathematical foundation might be one reason for the present plight of FRA. So\, developing a generic mathematical-based approach for interpretation and location of  incipient mechanical winding damages in actual 3-Φ transformer windings\, using measured FRA\, is imperative. Development of a such generic method necessitates derivation of closed-form expressions which can provide a direct link between measured FRA quantities to the electrical parameters of the winding. For assessing damage severity\, the challenge is to identify a quantity which is not only extractable from measured FRA\, but also be sensitive\, monotonic\, and traceable to the fault. Driven by this philosophy this thesis aims to address the following –\n\n\n\n•   Propose a unified and general approach to derive closed-form analytical expressions (for each multiphase winding) to link the measured open and short circuit natural frequencies to electrical parameters of the winding\, and valid for any condition of the neutral•   Define a quantity calculable from the measured FRA’s peak/trough frequencies which is physically related to mechanical damage in the winding\, and perhaps yield some physical insight about damage•   Develop novel methods using the derived analytical expressions to identify an incipient\, discrete\, and localized axial and/or radial displacement in any multiphase winding\, and applicable for any condition of the neutral\n\n\n\nIn the second chapter\, a generic and unified analytical method is developed (applicable to any 1-Φ or 3-Φ winding) starting from the basic mutually coupled lossless ladder network model to derive equations which relate the harmonic sum of squares of short circuit natural frequencies (SCNF) and open circuit natural frequencies (OCNF) to the elemental winding inductances and capacitances. Complete details of the derivation are discussed\, and all the derived formulae were cross verified by extensive numerical circuit simulations.\n\n\n\nEach one of these derived expressions has a strikingly similar structure and possesses a unique property viz.\, the contribution of series capacitances and ground capacitances are decoupled. This important property paves way for estimating a physical quantity that is directly responsible for the winding resonances\, viz.\, the effective air-core inductance (Leff). This estimation requires multiple FRA measurements. Chapter 3 presents complete details of the concept\, its derivation\, measurements\, and experimental results are for all 1-Φ and 3-Φ windings.\n\n\n\nLoss of clamping pressure in a winding is not directly identifiable by any means\, other than an FRA measurement. But\, this damage cannot be judged by merely comparing two FRAs. So\, a clamping pressure measurement experiment was carried out on a single isolated winding to ascertain the sensitivity and monotonicity afforded by the quantity\, Leff\, to a change in clamping pressure. Driven by the promising results\, author proceeds to build a method based on Leff to find the location of a discrete and localized axial displacement (AD) in any 3-Φ winding configuration. Details of this method\, experimental results\, and measurement steps are presented in Chapter 4.\n\n\n\nProceeding further\, Chapter 5 discusses concept of a new method\,  measurement steps and experimental results to identify presence of a Radial Displacement (RD) in a 3-Φ star winding with neutral-open\, as well as\, in a delta connected winding. Driven by success\, the concept was extended to identify the simultaneous occurrence of a discrete and localized AD and RD in one phase of a 3-Φ star winding\, with neutral-open. Preliminary experimental results proved the method can successfully identify faulted phases that contained AD and RD.\n\n\n\nAll experiments reported in the thesis were carried out on transformer windings rated at 33 kV\, 3.5 MVA. The results are encouraging and the author believes that true potential of the proposed methods can be judged when implemented on actual transformers.\n\n\n\nIn summary\, this thesis presents\, perhaps for the first time\, a mathematical basis for identifying and diagnosing axial and radial displacements in 1-Φ and 3-Φ windings using the peak/trough frequency data from the measured FRA. The author believes that this is a small step forward in advancing FRA as a diagnostic tool.\n\n* * * * * * * * * * ALL ARE CORDIALLY INVITED * * * * * * * * *
URL:https://ee.iisc.ac.in/event/thesis-colloquium-of-mr-bidhan-biswas/
LOCATION:HV seminar Hall
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221118T200000
DTEND;TZID=Asia/Kolkata:20221118T210000
DTSTAMP:20260404T023037
CREATED:20221109T234238Z
LAST-MODIFIED:20221109T234314Z
UID:240117-1668801600-1668805200@ee.iisc.ac.in
SUMMARY:Thesis Defence of Mr. Unni V S
DESCRIPTION:Degree Registered: PhD (Engg). \nGuide: Prof. Kunal Narayan ChaudhuryDate: November 18\, 2022.Time: 2:30 pm.Venue: Online.Link: MS Teams link: https://tinyurl.com/yr96memmTitle: Efficient and Convergent Algorithms for High-Fidelity Hyperspectral Image Fusion.Abstract: Hyperspectral (HS) imaging refers to acquiring images with hundreds of bands corresponding to different wavelengths of light. HS imaging has a wide range of applications such as remote sensing\, industrial inspection\, environmental monitoring\, etc. A fundamental consideration with multiband sensors is that the amount of incident energy is limited and this creates an intrinsic tradeoff between spatial resolution and the number of bands—current optical sensors can either generate images with high resolution but a small number of bands or images with a large number of bands but reduced resolution. For example\, HS images have hundreds of bands but low spatial resolution\, whereas the opposite is true for multispectral (MS) images. An extreme case is a panchromatic (PAN) image with very high spatial resolution but just a single band. Image fusion refers to techniques where multiband images with high spatial resolution are synthetically generated using image processing algorithms. It includes pansharpening (MS+PAN)\, hyperspectral sharpening (HS+PAN)\, and HS-MS fusion (HS+MS). Reconstructing a fused image from the observed images is ill-posed and needs regularization. Diverse regularization methods have been proposed over the years for general imaging problems\, many of which perform very well for fusion. This includes vector total variation\, sparsity and dictionary-based penalties\, generalized Gaussian- and GMM-based priors\, etc. This thesis proposes novel regularization models and algorithms that can outperform state-of-the-art image fusion techniques. We can broadly group these into two classes—explicit and implicit regularization.Explicit regularization refers to the design of hand-crafted penalty functions that impose desirable properties (e.g.\, smoothness) on the reconstruction; this is used along with the observed data for fusion. We propose a convex regularizer that is motivated by nonlocal patch-based methods for image restoration. Our regularizer accounts for long-distance correlations in hyperspectral images\, considers patch variation for capturing texture information\, and uses the higher resolution image for guiding the fusion process. Unlike local pixel-based methods\, where variations along just horizontal and vertical directions are penalized\, we use a wider search window in terms of nonlocality and directionality. This is shown to yield state-of-the-art results. The catch is that the resulting optimization problem is non-differentiable and we cannot use simple gradient-based algorithms. However\, we show that by expressing patch variation as filtering operations and judiciously splitting the original variables and introducing latent variables\, we develop a provably convergent iterative algorithm\, where the subproblems can be solved efficiently using FFT-based convolution and soft-thresholding.In the implicit approach\, we rely on a recent paradigm known as plug-and-play (PnP) regularization\, where powerful off-the-shelf denoisers are used for regularization purposes. While this has been shown to give state-of-the-art results for general restoration tasks\, it has not so much been explored for fusion. In fact\, we faced few technical challenges in applying PnP for hyperspectral fusion. Firstly\, existing denoisers are slow when applied to multiband images and we need to apply such denoisers several times with the PnP framework. Secondly\, convergence is generally not guaranteed for PnP regularization since the mechanism is ad-hoc. Along with efficiency and good denoising performance\, we need to come up with a denoiser with specific properties that can guarantee convergence. We proposed a couple of approaches to solve this problem. In the first approach\, we have developed a high-dimensional kernel denoiser with low cost yet good denoising performance\, which can guarantee PnP convergence. The overall algorithm is fast and competitive with state-of-the-art methods. In the second approach\, we leverage the power of deep learning to develop a trained patch denoiser which has a couple of advantages over conventional end-to-end learning: (1) Unlike end-to-end networks which require excessive ground-truth data for training\, we can be trained the denoiser from patches extracted from the observed images. For example\, in HS+MS fusion\, the MS image captures the same scene and has the same spatial resolution as the target image. We train the denoiser by sampling clean patches from the MS image and corrupting them with noise.(2) Compared to end-to-end learning\, where the training is done with a fixed forward model\, our method can be deployed for different forward models. This is possible thanks to the decoupling of the inversion (of the forward model) and denoising steps in PnP.We use the trained denoiser for PnP regularization and establish convergence of the PnP iterations under a technical assumption that we verify numerically. As far as the reconstruction quality is concerned\, our method outperforms state-of-the-art variational and deep-learning fusion techniques.
URL:https://ee.iisc.ac.in/event/thesis-defence-of-mr-unni-v-s/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221114T160000
DTEND;TZID=Asia/Kolkata:20221114T170000
DTSTAMP:20260404T023037
CREATED:20221102T044553Z
LAST-MODIFIED:20221102T045520Z
UID:240090-1668441600-1668445200@ee.iisc.ac.in
SUMMARY:Thesis Defence of Meshineni Deepchand
DESCRIPTION:Degree Registered:              M.Tech(Res).\nGuide:                                    Dr. Rathna G N\nVenue:                                    MMCR EE\, C 241\nDate & Time:                         14th November 10:30am \nTeams meeting link: Click here to join the meeting \nTitle:    Non contact Breathing and Heartbeat signals monitoring using FMCW radar \nAbstract: Non-contact breathing and heartbeat signals monitoring are the tasks of extracting them without contact sensors. It became even more critical in COVID 19\, and hence it is crucial to estimate them correctly. FMCW (Frequency Modulated Continuous Wave) radar is employed to estimate these two signals without contact. Radar captures chest displacement and body movement. Because of this\, breathing and heartbeat signals are distorted. The reduction of false peaks and peak estimation is crucial for breathing rate calculation. So in this thesis\, firstly\, we propose a novel way for tracing body movement and eliminating the traced segment for breathing and heart rate calculation. In the second part\, we efficiently reduced false peaks using maximal overlap discrete wavelet transform (MODWT) to decompose and reconstruct the filtered breathing signal for estimating breathing rate. The heartbeat signal is estimated using bandpass filtering of unwrapped phase. We also compared our algorithm with the task force monitoring (TFM) device as a reference and discussed its performance. Also\, our proposed method for breathing rate estimation has an accuracy of 92.43% and heartrate estimation it is 85.16%. \n* * * * * * * * ALL ARE CORDIALLY INVITED * * * * * * * *
URL:https://ee.iisc.ac.in/event/thesis-defence-of-meshineni-deepchand/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221111T170000
DTEND;TZID=Asia/Kolkata:20221111T170000
DTSTAMP:20260404T023037
CREATED:20221107T014902Z
LAST-MODIFIED:20221107T025832Z
UID:240109-1668186000-1668186000@ee.iisc.ac.in
SUMMARY:Lecture by Dr. Ganesh Sivararaman
DESCRIPTION:Indian Institute of Science and\nThe IEEE Signal Processing Society\, Bangalore Chapter\nCordially invites you to the following talk on\n“Unsupervised adaptation in speech technologies”\n(Click here for the poster.) \nSpeaker: Dr. Ganesh Sivaraman\, Pindrop\, Atlanta\, USA\nDate and Time: 11th November 2022 at 11:30am to 12:30pm\nVenue: MMCR (Room No. C241)\, 1st Floor\, Dept. of Electrical Engineering \nAbstract: Unsupervised learning and adaptation techniques have taken center stage due to the exponential growth of unlabeled data. For many practical applications unsupervised learning helps in the adaptation of machine learning systems to mismatched train and test domains. Unsupervised adaptation can be performed by three broad approaches – 1) feature transformations in the test domain\, 2) model adaptation to test domain\, and 3) generation of synthetic test domain samples. This talk will outline these methods by showing three specific examples from speech processing. Unsupervised speaker adaptation for acoustic-to-articulatory speech inversion serves as an example of feature transformation-based adaptation. Adaptation of end-to-end ASR systems without manual transcriptions will be presented as an example of model adaptation. Finally\, children’s speech simulation for zero-shot child speech classification using X-vectors will be presented as an example of synthetic data generation for the test domain. \nBiography:  Ganesh Sivaraman is a Senior Research Scientist at Pindrop\, in Atlanta\, USA. He received his M.S. (2013) and Ph.D. (2017) in Electrical Engineering from the University of Maryland College Park. His research experience and publications span several speech technologies like acoustic-to-articulatory inversion\, ASR\, speaker recognition\, deepfake detection\, and speech enhancement. During his PhD at Maryland\, he was awarded the Future Faculty Fellowship\, and the International Graduate Research Fellowship by the A. James Clark School of Engineering. Along with his official work\, he is actively involved in teaching\, mentoring\, and collaborating with doctoral students at Maryland. Apart from research work\, Ganesh is a fluent speaker of Sanskrit actively learning and teaching the language as a volunteer of Samskrita Bharati USA. He is passionate about creating computational tools for learning Sanskrit pronunciation.
URL:https://ee.iisc.ac.in/event/lecture-by-dr-ganesh-sivararaman/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221111T150000
DTEND;TZID=Asia/Kolkata:20221111T160000
DTSTAMP:20260404T023037
CREATED:20221106T232123Z
LAST-MODIFIED:20221106T233337Z
UID:240104-1668178800-1668182400@ee.iisc.ac.in
SUMMARY:Thesis Colloquium of Sayantan Das
DESCRIPTION:Thesis Title:  Modeling of lightning attachment to aircraft and  quantification of the influencing parameters \nGuide: Prof. Udaya Kumar \nDegree registered:          Ph.D. \nDate and Time:              11th November 2022\, 9:30 AM\nMeeting link:                  https://teams.microsoft.com/l/meetup-join/19%3ameeting_MDg5YzdhYWUtMTc3Zi00Yjg0LWE1ZTktYjgyY2I5Y2MyNDI4%40thread.v2/0?context=%7b%22Tid%22%3a%226f15cd97-f6a7-41e3-b2c5-ad4193976476%22%2c%22Oid%22%3a%227ef4df52-6005-46aa-bff3-d96db1a85b71%22%7d \nAbstract: According to Air Transport Action Group (ATAG)\, 45 million aircraft took off worldwide in 2020\, which translates to 1.5 lakh per day. Statistically\, the aviation industry is found to double its fleet size every fifteen years. Lightning is considered one of the dreadful environmental threats to aircraft. Past incidents show that lightning strikes can lead to structural damage\, operational interruption\, and loss of lives. Field data suggest that\, on average\, an aircraft gets struck by lightning once or twice a year. Therefore\, the threat due to lightning is considered a crucial safety aspect of an aircraft. \nDesign of suitable lightning protection for aircraft involves Zoning of its skin. It is intended to differentiate lightning attachment points\, channel slipping regions\, and regions that carry just the stroke current. The first step of Aircraft Zoning is to identify the initial attachment points. For the same\, different methods like Laboratory experiments\, similarity principle\, Rolling Sphere Method (RSM)\, and Field-based approach are suggested in the standard ARP5414. In reality\, the lightning strikes to aircraft can be of two modes\, Aircraft-initiated and Aircraft-intercepted. In the former one\, under the influence of a thundercloud or descending lightning leader\, the aircraft initiates stable bipolar connecting leaders\, upward and downward leader toward the ground. These leaders are deemed to propagate hundreds of meters to complete the lightning strike. In Aircraft-intercepted strikes\, the aircraft intercepts a descending lightning leader and hence gets struck. The laboratory experiments on scaled aircraft models or isolated aircraft parts are inadequate to assess the initial attachment points. The similarity principle suggested in the standard is qualitative and can’t be extended to aircraft of any size and shape. The 25m Rolling Sphere Method (RSM) is routinely employed to determine the attachment points. This method doesn’t consider the connecting leader discharges from aircraft and therefore overestimates the possible attachment points. Most (90%) of the lightning strikes to aircraft are attributed to aircraft-initiated mode\, which involves significant connecting leader activities. Therefore\, it has to be traced accurately to assess attachment points. \nIn literature\, it is difficult to find a model for bipolar leader discharges from aircraft. However\, work on either negative or positive leader inception and propagation from laboratory gaps and their extension can be relatively found. Based on them\, the present work aims to develop a suitable model for simulating bipolar leader discharges from aircraft. Additionally\, the aircraft-intercepted mode of lightning strikes is also included. In summary\, a novel model adapting the pertinent physical aspects of the leader discharges has been developed to accurately assess initial lightning attachment points to aircraft.  \nUsing the model developed\, the following practically important questions are addressed:  \n\nDependency of the frequency of lightning strikes to aircraft on its shape and size.\nRate of lightning strikes to aircraft at different altitudes\nFor a given aircraft and its route\, the number of times it gets struck by lightning\nThe average number of strikes to an aircraft per year\n\nTo present a quantitative assessment\, two different aircraft models\, McDonnell Douglas DC-10 and Standard Dynamic Model are considered. \nIn summary\, a novel model based on physical grounds has been developed to assess the initial lightning attachment points on aircraft. Using the same\, further methodologies are constructed to quantify the dependency of the strike rate on aircraft size\, altitude\, and possible average strike rate. \nALL ARE WELCOME
URL:https://ee.iisc.ac.in/event/thesis-colloquium-of-sayantan-das/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221109T193000
DTEND;TZID=Asia/Kolkata:20221109T203000
DTSTAMP:20260404T023037
CREATED:20221103T232724Z
LAST-MODIFIED:20221103T233801Z
UID:240097-1668022200-1668025800@ee.iisc.ac.in
SUMMARY:Lecture by Prof. Carlos Busso
DESCRIPTION:Organised by\nIndian Institute of Science and\nThe IEEE Signal Processing Society\, Bangalore Chapter\nTitle: Robust Emotion Recognition (Click here for the poster)\nDate and Time: 9th November 2022 at 2:00pm\, Refreshments: 3:00pm\nVenue: MMCR (Room No. C241\, 1st Floor\, Dept. of Electrical Engineering \nAbstract of the talk: It is challenging to achieve robust and well-generalized models for tasks involving subjective concepts such as emotion. This tech talk will describe novel approaches to effectively develop robust speech emotion recognition (SER) systems. At the resource level\, we will describe our effort to collect the MSP-Podcast corpus\, which is a large\, naturalistic emotional database. The data collection protocol combines machine-learning algorithms to retrieve recordings conveying balanced emotional content annotated with a cost-effective crowdsourcing protocol. To improve the temporal modeling of SER systems\, this seminar will also discuss a novel dynamic chunking approach that maps the sequences of different lengths into a fixed number of chunks that have the same duration by adjusting their overlap. This simple chunking procedure creates a flexible framework\, facilitating parallel computing. The approach can incorporate different feature extractions and sentence-level temporal aggregation approaches to cope\, in a principled way\, with a sequence-to-one SER task. Likewise\, the seminar will discuss multimodal pre-text tasks that are carefully designed to learn better representations for predicting emotional cues from speech\, leveraging the relationship between acoustic and facial features. Finally\, the seminar will discuss our current effort to design multimodal emotion recognition strategies that effectively combine auxiliary networks\, a transformer architecture\, and an optimized training mechanism for aligning modalities\, capturing temporal information\, and handling missing features. These models offer principled solutions to increase the generalization and robustness of emotion recognitions  systems. \nSpeaker Biography: Carlos Busso received his PhD degree (2008) in electrical engineering from the University of Southern California (USC)\, Los Angeles\, in 2008. He is a professor at the Electrical Engineering Department of The University of Texas at Dallas (UTD). At UTD\, he leads the Multimodal Signal Processing (MSP) laboratory [http://msp.utdallas.edu]. He is a recipient of an NSF CAREER Award. In 2014\, he received the ICMI Ten-Year Technical Impact Award. In 2015\, his student received the third prize IEEE ITSS Best Dissertation Award (N. Li). He also received the Hewlett Packard Best Paper Award at the IEEE ICME 2011 (with J. Jain)\, and the Best Paper Award at the AAAC ACII 2017 (with Yannakakis and Cowie). He received the Best of IEEE Transactions on Affective Computing Paper Collection in 2021 (with R. Lotfian) and in 2022 (with Yannakakis and Cowie). He is the co-author of the winner paper of the Classifier Sub-Challenge event at the Interspeech 2009 emotion challenge. His research interest is in human-centered multimodal machine intelligence and applications. His current research includes the broad areas of affective computing\, multimodal human-machine interfaces\, nonverbal behaviors for conversational agents\, in-vehicle active safety system\, and machine learning methods for multimodal processing. His work has direct implication in many practical domains\, including national security\, health care\, entertainment\, transportation systems\, and education. He was the general chair of ACII 2017 and ICMI 2021. He is a member of ISCA\, AAAC\, and a senior member of ACM and IEEE.
URL:https://ee.iisc.ac.in/event/lecture-by-prof-carlos-busso/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221107T200000
DTEND;TZID=Asia/Kolkata:20221107T220000
DTSTAMP:20260404T023037
CREATED:20221102T034842Z
LAST-MODIFIED:20221102T042355Z
UID:240082-1667851200-1667858400@ee.iisc.ac.in
SUMMARY:Online Thesis Defence of Ashiq Muhammed P E
DESCRIPTION:Degree Registered:              Ph.D.\nGuide:               Prof. Satish L and Prof. Udaya Kumar\nThesis Title:     Improved Understanding of Standing Waves in Single Layer Coil and Elegant Methods to Estimate Transformer Winding Parameters \nClick here to join the Meeting \nAbstract: Analyzing the effect of impulse voltages (like lightning\, switching) on transformer winding has occupied centerstage in core electrical engineering research for over a century. These investigations gather great significance and relevance as it eventually governs the design of insulation in the winding. Notwithstanding the colossal contribution this domain has witnessed from stalwarts in the past century\, a closer scrutiny surprisingly reveals that there still exists tiny grey areas that demands attention. Pursuing this line of thought\, the first part of this thesis aims to clearly describe what this grey area is and resolving it would provide a deeper insight about fundamental understanding of surge response in transformer windings – with special emphasis on its standing wave phenomenon. Following this\, in the latter part\, elegant procedures are stitched together to determine a few electrical parameters of the transformer winding equivalent circuit that have the potential to help in assessing mechanical status of windings. Objectives of the thesis are – \n1.    Formulate an analytical method to determine the exact shape of standing waves for all modes in a uniform single layer coil as a solution of its governing partial differential equation2.    Estimate series capacitance of a uniform transformer winding from its measured driving point impedance3.    Determine effective air-core inductance of an iron-core uniform winding as a function of its axial length from measured driving point impedance \nFirst part of the thesis revisits a century-old classical theory of standing waves on uniform single layer coils. Accurate information about natural frequencies and shapes of the corresponding standing waves are essential for gaining a deeper understanding of the response of coils to impulse excitations. Analytical studies on coils have largely been based on the assumption that standing waves are sinusoids in both space and time. However\, this contradicts the results from numerical circuit analysis and practical measurements. So\, this thesis attempts to bridge this discrepancy by revisiting the classical standing wave phenomena in coils. It not only assesses the reason for the aforementioned inconsistency\, but also makes a contribution by analytically deriving the exact mode shape of standing waves for both neutral open/short conditions. For this\, the coil is modelled as a distributed network of elemental inductances and capacitances\, while an exponential function describes the spatial variation of mutual inductance between turns. Initially\, an elegant derivation of the governing partial differential equation (in terms of voltage as the variable instead of flux) for surge distribution is presented and to the best of our knowledge\, for the first time\, an analytical solution for the same has been found by the variable-separable method to find the complete solution (sum of time and spatial terms). Hyperbolic terms in the spatial part of the solution have always been neglected but are included here\, thus\, yielding the exact mode shapes. For verification\, both voltage and current standing waves computed from the analytical solution were plotted and compared with PSPICE simulation results on a 100-section ladder network representing a uniform single-layer coil. Then\, practical measurements were made on a tailor-made large-sized single layer coil with a length of 2.02 m\, diameter of ~1 m and having 640 turns. It turns out that even in such simple single layer coils\, the shape of standing waves of all modes deviates considerably from being sinusoidal. It was further observed that this deviation depends on spatial variation of mutual inductance\, capacitive coupling\, and order of the standing waves. \nIn the second part\, an elegant method for determining the series capacitance (Cs) and air-core equivalent inductance of a uniform winding as a function of its axial length (termed as M0x in this thesis) of a uniform transformer winding\, from its measured DPI magnitude\, is discussed. Knowledge about the series capacitance of the winding is essential\, which along with shunt capacitance\, determines the initial impulse voltage distribution when a surge impinges on the winding. Unlike previously published approaches\, the proposed method does not involve any cumbersome and time-consuming curve-fitting or running of optimization/search algorithms. Neither does it require winding geometry data. The proposed procedure for finding series capacitance relies on a property that is observable in the driving point impedance (DPI) function of a lossless winding with an open neutral condition\, viz.\, the ratio of the product of squares of open circuit natural frequencies to the product of squares of short circuit natural frequencies bears a particular relation to the coefficients of the DPI function. A simple procedure involving a deft manipulation and combination of a few well-known properties that correlate the roots of a polynomial to its coefficients are then utilized for determining series capacitance. \nKnowledge about equivalent air-core inductance distribution as a function of its axial length (i.e.\, M0x) is useful for localizing a minor/incipient mechanical fault in the winding. A physically realizable empirical relationship to estimate M0x is initially proposed. The corresponding constants of the empirical relationship are then calculated from the measured DPI. The proposed method requires three DPI measurements: one with neutral-end open and the other with neutral-end shorted. The third DPI is measured with a known external lumped capacitance connected between the neutral and ground. This method requires only the first few dominant natural frequencies observable in the first two of the DPIs. \nFeasibility of proposed methods for estimating Cs and M0x was initially verified by simulation on an N-section ladder network and then by experiments on small-sized continuous-disk and interleaved-disk windings\, and finally on a large-sized 33 kV\, 3.5 MVA continuous-disk winding. Salient features of the proposed methods are – they are simple\, elegant and involve minimum post-processing after measuring the DPI. Given its inherent simplicity and their relevance\, the author is hopeful that industry will come forward to implement these procedures on an existing FRA measuring instruments – thus opening a new dimension to FRA measurements. \n* * * * * * * * ALL ARE CORDIALLY INVITED * * * * * * * *
URL:https://ee.iisc.ac.in/event/thesis-defence-of-ashiq-muhammed-p-e/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221021T213000
DTEND;TZID=Asia/Kolkata:20221021T223000
DTSTAMP:20260404T023037
CREATED:20221020T042250Z
LAST-MODIFIED:20221020T042943Z
UID:240068-1666387800-1666391400@ee.iisc.ac.in
SUMMARY:EE Faculty Colloquium by Prof. A G Ramakrishnan
DESCRIPTION:Title: Analyzing patterns in EEG: from biometrics to altered states of consciousness \nSpeaker: Prof. Ramakrishnan A. G.\, MILE Lab\, Dept of Electrical Engineering\, Indian Institute of Science \nVenue: EE MMCR \nTime: 4pm\, 21 October 2022 \nAbstract: The electrical activity of brain\, the key control organ of our body\, carries a lot of information about our current activity\, health\, mental status and identity. We\, in MILE Lab\, have been focussing for the past few years on analysing the various patterns present in the electroencephalogram (EEG). When someone is in deep sleep\, coma or under anesthesia\, the level of consciousness is lower than that of waking state. When someone has locked-in syndrome\, the level of consciousness is the same as that of waking state. On the other hand\, during meditation\, the consciousness level is higher than that of waking. Hypnosis is another completely different altered state of consciousness. The interrelationship between the different EEG channels is also distinctly different during inhalation\, breath-hold and exhalation. We are studying the patterns in EEG under all the above conditions. While working on the above topics\, we unexpectedly made a discovery that some measures based on the functional connectivity between the channels are distinct for each individual and can very well be used to identify people. Using other measures\, we are also able to predict the word imagined by a person out of a set of words.Speaker bio: Ramakrishnan A. G. is a professor of Electrical Engineering and an associate faculty member at the Centre for Neuroscience. He obtained his Masters in Electrical Engineering and Ph D in Biomedical Engineering from the Indian Institute of Technology\, Madras. He has graduated 19 Ph.D.s\, 16 M.Tech.s by research\, and guided over 100 M. Tech. projects at IISc. He is a Fellow of the Indian National Academy of Engineering\, As the leader of a research consortium\, he was instrumental in creating handwriting recognition technologies for eight Indian languages. He received Manthan award (South East Asia and Asia Pacific) twice for creating audio books for blind students through his OCR and TTS in Tamil and Kannada. His current areas of research include speech recognition in Indic languages\, decoding of imagined words from EEG\, brain functional connectivity analysis in modified states of consciousness and the study of the neural control and physiological mechanisms behind the health and therapeutic effects of deep breathing. For his earlier work on evoked potentials from leprosy patients\, he had received Sir Watt Kay Young Researcher’s Prize from the Royal College of Physicians and Surgeons\, Glasgow. He was a Senior Research Scientist at Hewlett Packard Research Labs\, Bangalore India from May 2002 to August 2003. He is an invited member of the Senate of IIIT-Allahabad\, Prayagraj and the Federation of Indian Chambers of Commerce and Industry – Indian Language Internet Alliance. He was a member of the Knowledge Commission\, Government of Karnataka during 2017-2020. He is also one of the founder directors of RaGaVeRa Indic Technologies private limited recognized by Karnataka Government as one of the Elevate 2019 Startup winners. The Kannada TTS developed by RaGaVeRa has been evaluated to be better in quality than Google’s Wavenet TTS and Nuance’s Kannada TTS. He is also the Advisor-Neuroscience of Feedfront Technologies Pvt Ltd\, Bengaluru.
URL:https://ee.iisc.ac.in/event/ee-faculty-colloquium-by-prof-a-g-ramakrishnan/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221021T194500
DTEND;TZID=Asia/Kolkata:20221021T210000
DTSTAMP:20260404T023037
CREATED:20221024T234607Z
LAST-MODIFIED:20221024T234711Z
UID:240071-1666381500-1666386000@ee.iisc.ac.in
SUMMARY:EE and CPS Seminar by Prof. Ketan Savla
DESCRIPTION:Title: Microscopic Traffic Flow Control\nDate: 21 October\nTime: 2:15pm\nVenue: EE MMCR \nAbstract: Design and performance evaluation of traffic control techniques such as ramp metering are typically based on macroscopic traffic flow models. These models\, obtained by spatio-temporal averaging of microscopic vehicle-to-vehicle/infrastructure interactions\, do not have sufficient resolution to model safety\, or to study the impact of emerging paradigms  of autonomy and connectivity. We present coordinated ramp metering algorithms that regulate entry into the freeway network at the vehicle level\, based on information about state of vehicles in the network\, but do not require information about travel demand. Under these algorithms\, each on-ramp operates under cycles during which it does not release more vehicles than its queue size at the beginning of the cycle. \nAdditionally\, the algorithms\, dynamically\, either introduce pause at the end of the cycle\, or modulate the release rate during the cycle\, or modulate safety distance for release during the cycle. Under standard safe vehicle-following and merging protocols\, these algorithms are shown to keep the network undersaturated for maximal travel demand and result in lower travel time than known ramp metering algorithms.Biography of the speaker: Ketan Savla is an associate professor and the John and Dorothy Shea Early Career Chair in Civil Engineering at the University of Southern California. His current research interest is in distributed optimal and robust control\, dynamical networks\, state-dependent queuing systems\, and mechanism design\, with applications in civil infrastructure systems. His recognitions include NSF CAREER\, George S. Axelby Outstanding Paper Award\, and the Donald P. Eckman Award. He serve(d) as an associate editor of the IEEE Transactions on Control of Network Systems\, IEEE Control Systems Letters\, and IEEE Transactions on Intelligent Transportation Systems. He is also a co-founder and the chief science officer of Xtelligent\, Inc.
URL:https://ee.iisc.ac.in/event/ee-and-cps-seminar-by-prof-ketan-savla/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221019T213000
DTEND;TZID=Asia/Kolkata:20221019T223000
DTSTAMP:20260404T023037
CREATED:20221019T032748Z
LAST-MODIFIED:20221019T032809Z
UID:240063-1666215000-1666218600@ee.iisc.ac.in
SUMMARY:Thesis Defence of Anwesha Roy
DESCRIPTION:Degree Registered:  M. Tech. (Research) \nGuide: Prof. Prasanta Kumar Ghosh \nDate  & Time: 19th October\, 2022\, Wednesday\, 4:00 PM \nVenue:     Online link : https://tinyurl.com/3rystxed \nTitle: Improved air-tissue boundary segmentation in real-time magnetic resonance imaging videos using speech articulator specific error criterion \nAbstract: Real-time Magnetic Resonance Imaging (rtMRI) is a tool used exhaustively in speech science and linguistics to understand the dynamics of the speech production process across languages and health conditions. rtMRI has two advantages over other methods which capture articulatory movement\, like X-ray\, Ultrasound and Electromagnetic articulography – it is non-invasive\, and it captures a complete view of the vocal tract including pharyngeal structures. The rtMRI video provides spatio-temporal information of speech articulatory movements\, which helps in modeling speech production. For this purpose\, a common step is to obtain the air-tissue boundary (ATB) segmentation in all frames of the rtMRI video. The accurate estimation of ATBs of the upper airway of the vocal tract is essential for many speech processing applications like speaker verification\, text-to-speech synthesis\, visual augmentation for synthesized articulatory videos\, and analysis of vocal tract movement. Thus\, it is necessary to have an accurate air-tissue boundary segmentation in every frame of the rtMRI videos. \nThe best performance in ATB segmentation of rtMRI videos in speech production\, in unseen subject conditions\, is known to be achieved by a 3-dimensional convolutional neural network (3D-CNN) model. In seen subject conditions\, both 3D-CNN and 2-dimensional deep convolutional encoder-decoder network (SegNet) show similar performance. However\, the evaluation of these models\, as well as other ATB segmentation techniques reported in literature\, has been done using Dynamic Time Warping (DTW) distance between the entire original and predicted boundaries or contours. Such an evaluation measure may not capture local errors in the predicted contour. Careful analysis of predicted contours reveals errors in regions like the velum part and tongue base section\, which are not captured in a global evaluation metric like DTW distance. In this thesis\, we automatically detect such errors and propose a novel correction scheme for them. We also propose two new evaluation metrics for ATB segmentation\, separately for each contour\, to explicitly capture errors in these contours. \nMoreover\, the state-of-the-art models use overall binary cross entropy as the loss function during model training. However\, such a global loss function does not give enough emphasis on regions which are more prone to errors. In this thesis\, together with global loss\, we explore the use of regional loss functions which focus on areas of the contours which have been analyzed as error prone in our analysis. Two different losses are considered in the regions around velum and tongue base – binary cross entropy (BCE) loss and dice loss. It is observed that dice-loss based models perform better than their BCE loss based counterparts.
URL:https://ee.iisc.ac.in/event/thesis-defence-of-anwesha-roy/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221013T203000
DTEND;TZID=Asia/Kolkata:20221013T220000
DTSTAMP:20260404T023037
CREATED:20221019T033655Z
LAST-MODIFIED:20221019T033655Z
UID:240065-1665693000-1665698400@ee.iisc.ac.in
SUMMARY:Thesis Defense of Kiran Kumar Challa
DESCRIPTION:Degree registered: PhD \nThesis Title: Algorithms and Testbed for Synchronous Generator Parameter Estimation \nGuide: Prof. Gurunath Gurrala \nExaminer: Prof. K. Shanti Swarup\, IIT Madras \nDate & TIme: 13th October 2022\, Thursday\, 3pm – 4.30pm \nMode:   Hybrid mode\, both Teams and Physical \nVenue: MMCR\, 1st floor C-wing\, EE Department\, IISc \n\n\nMicrosoft Teams meeting \n\n\n\nJoin on your computer\, mobile app or room device \n\nClick here to join the meeting \n\n\n\nMeeting ID: 438 079 328 744Passcode: q4pugu \n\n\n\n\n\nAbstract: The development of dynamic power system components models became increasingly important in the modern grids dominated by high penetration of renewables because of the increased dependency of planning and operational decisions on dynamic simulation studies. The parameters of synchronous machines and associated control models play significant role in the overall model of the grid\, which need to be updated regularly by the utilities. So\, the parameters of the power plants are calibrated/estimated either using off-line testing or online measurements from phasor measurement units (PMU) or digital fault recorders (DFR). Development of individual generator models is feasible only if the PMU/DFR data is available for each generator in a power plant. Otherwise\, they can provide only aggregate model of a generating plant as PMU/DFRs are usually placed in substations. Digital protective relay (DPR) records are available for individual generators in any generating plant. This thesis explores the possibilities of utilizing DPR records of individual generators for parameter estimation. About 200 relay records have been collected from a hydro plant and a thermal plant in Karnataka. It is found that most of the records contain at the most 3 seconds data. Existing methods of parameter estimation using PMU/DFR data failed to work with the short duration records. There is no prior work reported in the literature which uses short relay records for parameter estimation of the synchronous generators. Constrained iterated unscented Kalman filter (CIUKF) and enhanced scattered search (eSS) algorithms are proposed for the parameter estimation using DPR records in this thesis. The parameters of a turbo alternator and its excitation system (210 MW) are estimated from the relay records collected using the proposed algorithms and the results are found be accurate. This is a first of its kind effort in the literature to the best of our knowledge. It is also found that the relay records should contain pre-fault data\, during fault data and some post-fault data for accurate estimation. However\, from the collected records only a small percentage of the records are found to be useful. To generate realistic data in the laboratory an experimental test bed development\, replicating the field implementation aspects of the digital relays\, is proposed in this thesis. A realistic scaled-down generalized substation model for translational research in smart grids is developed\, which can be configured to operate in 7 widely used substation bus bar schemes with prevalent current transformer (CT) configurations. All the potential transformers (PT) and CT measurements\, circuit breaker (CB)\, isolator and earth switch status signals are made available to configure any protection strategy like bus-bar protection\, unit protection schemes\, etc. precisely the same way they get implemented in the field. A systematic procedure for the development of an experimental scaled-down frequency-dependent transmission line model of a 230 kV transmission line is proposed. A lumped parameter frequency dependent transmission line model using modal transformation is derived for a 230 kV transmission line and scaled-down to 220 V. Clarke and inverse Clarke transformations are implemented using specially designed 1-phase transformers. The inductances of the scaled-down model are realized using amorphous cores. A new algorithm is proposed to fit a reduced-order R-L equivalent circuit to the frequency response of the modal impedances of the transmission lines. A close enough fitting is achieved with lesser number of passive elements using the proposed method compared to the widely used vector fitting algorithm. This kind of physical realization of a frequency dependent power transmission line model in the laboratory is first of its kind effort in the literature to the best of our knowledge. \n\nNote: Know how generated from the implementation of the generalized substation panels and transmission line models has been licensed to MCore Technologies Pvt Ltd\, Bangalore for commercialization. \nAcknowledgements: This work is supported by Fund for Improvement of Science and Technology (FIST) program\, DST\, India\, No.SR/FST/ETII-063/2015 (C) and (G) under the project “Smart Energy Systems Infrastructure – Hybrid Test Bed”. Acknowledge partial funding support from Robert Bosch Centre for Cyber Physical Systems (RBCCPS)\, IISc. Also acknowledge the Tata Trust Travel Grant.
URL:https://ee.iisc.ac.in/event/thesis-defense-of-kiran-kumar-challa/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221011T173000
DTEND;TZID=Asia/Kolkata:20221011T183000
DTSTAMP:20260404T023037
CREATED:20221006T052743Z
LAST-MODIFIED:20221006T052956Z
UID:240024-1665509400-1665513000@ee.iisc.ac.in
SUMMARY:Thesis Defence of Mr. Jerrin Thomas Panachakel
DESCRIPTION:Degree Registered:      Ph.D.\n\nDate and Time:            Oct. 11\, 2022 (Tuesday)  12 Noon\n\nVenue:                           MMCR\, Hall No. C 241\, II Floor\, Dept. of Electrical Engineering.\nMS Teams Link:            https://teams.microsoft.com/l/channel/19%3a7nfRrQdIH7fPf5u003TVCBAp18cRM7dyFwv1eqJRGjY1%40thread.tacv2/General?groupId=53beb192-2a2f-4f20-8a0c-dbb67a65f532&tenantId=6f15cd97-f6a7-41e3-b2c5-ad4193976476\n \nTitle: Machine Learning for Decoding Imagined words and Altered State of Consciousness from EEG\n\nAbstract: The thesis explores several architectures for accurately decoding the cognitive activity from EEG recorded during speech imagery and Rajayoga meditation. The major contributions of the thesis are listed below:\n\n\n\n\n\n\n\n\n\n\n\n\n\nNeural Correlates of Phonological Category in Speech Imagery EEG\n\n\nWe have shown that neural correlates of phonological categories exist in the EEG recorded during speech imagery. These correlates lead to significant differences in the mean phase coherence (MPC) values of the EEG across several cortical regions.\nWe have also shown that MPC values can be used for accurately classifying the EEG recorded during speech imagery based on the phonological category of the prompts. The proposed architecture for this task has an accuracy of 84.9%.\n\n\n\nDecoding Imagined Words from EEG\n\n\nOne of the challenges in designing systems for decoding imagined words from EEG is the limited availability of data. We have presented three architectures for decoding imagined words from EEG. All three architectures alleviate this problem of limited availability of data.\nThe transfer learning-based architecture employs MPC and magnitude-squared coherence values along with a ResNet50-based classifier. This architecture achieves an accuracy of 92.8% on a publicly available EEG dataset in classifying speech imagery.\n\nClassification of Altered State of Consciousness from Resting State\n\n\nWe have presented three architectures for classifying the altered state of consciousness during Rajayoga meditation from the resting state.\nBoth CSP-LDA-LSTM (common spatial pattern-linear discriminant analysis-long short-term memory) and SVD-DNN (singular value decomposition-deep neural network) architectures are able to capture subject-invariant features.\nThe best intra-subject accuracy obtained is 98.2% and the best inter-subject accuracy is 96.4%.
URL:https://ee.iisc.ac.in/event/thesis-defence-of-mr-jerrin-thomas-panachakel/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221007T203000
DTEND;TZID=Asia/Kolkata:20221007T213000
DTSTAMP:20260404T023037
CREATED:20221006T043709Z
LAST-MODIFIED:20221006T045715Z
UID:240021-1665174600-1665178200@ee.iisc.ac.in
SUMMARY:Thesis Colloquium of Mr. Shreyas Ramoji @3pm
DESCRIPTION:Date: October 7th\, Friday\, 3-4pm \nDegree Registered: PhD \nVenue: EE\, MMCR [1st Floor\, C241] and in Microsoft Teams at https://tinyurl.com/2rfbs7ke \nThesis Title: Supervised Learning Approaches for Language and Speaker Recognition \nAbstract: In the age of artificial intelligence\, one of the important goals of the research community is to get machines to automatically figure out who is speaking and in what language – a task that humans are naturally capable of. Developing algorithms that automatically infer the speaker\, language\, or accent from a given segment of speech are challenging tasks for machines and has been a topic of research for at least three decades. While most of the prior successes have been through the development of unsupervised embedding extractors\, the main aim of this doctoral research is to propose novel supervised approaches for robust speaker and language recognition. \nIn the first part of this talk\, we propose a supervised version of a popular embedding extraction approach called the i-vector. The i-vector is a popular technique for front-end embedding extraction in speaker and language recognition. In this approach\, a database of speech recordings (in the form of a sequence of short-term feature vectors) is modeled with a Gaussian Mixture Model\, called the Universal Background Model (GMM-UBM). The deviation in the mean components is captured in a lower dimensional latent space called the i-vector space using a factor analysis framework. In our work\, we proposed a fully supervised version of the i-vector model\, where each label class is associated with a Gaussian prior with a class-specific mean parameter. The joint prior (marginalized over the sample space of classes) on the latent variable becomes a GMM. The choice of prior is motivated by the Gaussian back-end\, where the conventional i-vectors for each language are modeled with a single Gaussian distribution. With detailed data analysis and visualization\, we showed that the supervised i-vector (s-vector) features yield representations succinctly capture the language (accent) label information and do a significantly better job distinguishing the various accents of the same language. We performed language recognition experiments on the NIST Language Recognition Evaluation (LRE) 2017 challenge dataset\, which has test segments ranging from 3 to 30 seconds. With the s-vector framework\, we observe relative improvements between 8% to 20% in terms of the Bayesian detection cost function\, 4% to 24% in terms of EER\, and 9% to 18% in terms of classification accuracy over the conventional i-vector framework. We also perform language recognition experiments showing similar improvements on the RATS dataset and Mozilla Common Voice dataset\, and speaker classification experiments using LibriSpeech. \nIn the second part of the talk\, we explore the problem of speaker verification\, where a binary decision has to be made on a test speech segment as to whether it is spoken by a target speaker or not\, based on a limited duration of enrollment speech. The state-of-the-art approach to speaker verification was to extract fixed-dimensional embeddings from speech of arbitrary duration and train a back-end generative model called the Probabilistic Linear Discriminant Analysis (PLDA) which was used to make decisions using a Bayesian decision framework. We proposed a neural network approach for back-end modeling\, where the likelihood ratio score of the generative PLDA model is posed as a discriminative similarity function\, and the learnable parameters of the score function are optimized using a verification cost. The proposed model\, termed as neural PLDA (NPLDA)\, is initialized using the generative PLDA model parameters. The loss function for the NPLDA model is an approximation of the minimum detection cost function (DCF) used as one of the evaluation metrics in various speaker verification challenges. Further\, we explore a fully neural approach where the neural model outputs the verification score directly\, given the acoustic feature inputs. This Siamese neural network (SiamNN) model combines embedding extraction and back-end modeling into a single processing pipeline. The development of the single neural Siamese model allows the joint optimization of all the modules using a verification cost. We provide a detailed analysis of the influence of hyper-parameters\, choice of loss functions\, and data sampling strategies for training these models. Several speaker recognition experiments were performed using Speakers in the Wild (SITW)\, VOiCES\, and NIST SRE datasets where the proposed NPLDA and SiamNN models are shown to improve over the state-of-art significantly. \nWe conclude the talk by highlighting some of the noteworthy approaches that were published during the course of this research work and identifying some important future directions that can be explored. \nBio: Shreyas Ramoji is a Ph.D. scholar at the Learning and Extraction of Acoustic Patterns (LEAP) Laboratory\, Department of Electrical Engineering\, Indian Institute of Science\, Bengaluru. He obtained his Bachelor of Engineering degree from the Department of Electronics and Communication Engineering\, PES Institute of Technology\, Bangalore South Campus in 2016. He is a student member of the IEEE Signal Processing Society and ISCA. His research interests include Speaker Verification\, Language and Accent Identification\, Neuroscience\, Machine learning\, and Artificial Intelligence. \n—————– \n​All are welcome. \n  \n 
URL:https://ee.iisc.ac.in/event/thesis-colloquium-of-mr-shreyas-ramoji-3pm/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20221006T203000
DTEND;TZID=Asia/Kolkata:20221006T213000
DTSTAMP:20260404T023037
CREATED:20220920T231630Z
LAST-MODIFIED:20220920T231758Z
UID:239899-1665088200-1665091800@ee.iisc.ac.in
SUMMARY:Thesis Defence of  Mr.  Paawan Kirankumar Dubal
DESCRIPTION:Thesis Title: Cyber Attack Resilient Breaker Failure Protection Scheme Using Wide Area measurements \nName of the Advisor: Prof. Sarasij Das \nDegree Registered: M.Tech-Research \nDate and Time: 6th October\, 2022\, 3:00 PM IST Online \nMeeting Link: https://teams.microsoft.com/l/meetup-join/19%3ameeting_ZWZkYzViOWQtNWIzYi00MWQ1LWE1ZTMtNzcwOTUxM2JkNGM0%40thread.v2/0?context=%7b%22Tid%22%3a%226f15cd97-f6a7-41e3-b2c5-ad4193976476%22%2c%22Oid%22%3a%2240bc61c0-cc62-49a2-80af-a06745a651ac%22%7d \nAbstract: Breaker Failure Protection (BFP) is a backup protection that comes into play when a circuit breaker fails to isolate the fault. If the circuit breaker fails to clear the fault\, the BFP scheme commands other required breakers to isolate the fault. BFP schemes are usually incorporated in microprocessor-based digital relays. Commonly employed BFP schemes use overcurrent element (50BF) and Breaker Failure Initiation (BFI) signals as inputs. The BFI signal is issued to the BFP relays from other digital relays. Line current sensed via current transformer is fed to the BFP relay for the overcurrent element (50BF). When both 50BF and BFI are high\, it waits for a specified time for the primary protection to operate. The 50BF element is usually set much lower than the rated load currents. So\, it will be high during the normal loading conditions. A cyber-attack can be launched by issuing false BFI or blocking a legitimate BFI signal to the BFP relay. Operation of BFP scheme usually leads to disconnection of a larger amount of loads. As a result\, mal-operation of BFP schemes can cause major disturbances in power systems. There is a need to make the BFP schemes resilient to cyber-attacks for reliable operation of power systems. Currently\, there is a lack of literature on the cyber-attack resilient BFP schemes. \nHence\, this thesis proposes a Wide-Area Measurement-Based Cyber-Resilient Breaker Failure Protection Scheme. The scope of the work is to develop an algorithm that will ascertain if the BFI received by the BFP relay is genuine. Blocking a legitimate BFI will cause the backup protection to operate and clear the fault. The proposition assumes that the BFP relay is not compromised in any manner. However\, a fake BFI can be issued by other digital relays\, which may cause unwanted BFP operations. In the proposed algorithm\, when the BFI is received. The BFP relay will communicate the receipt of BFI to the Phasor Data Concentrator (PDC). The proposed algorithm will run at the PDC\, which has access to the time-stamped measurements of the adjacent substations and the substation that triggered the algorithm. The decision of the proposed algorithm is communicated back to the BFP relay\, which will allow the tripping if the BFI is genuine. Hence\, we also propose modifications in the BF scheme in the BFP relay to incorporate the algorithm’s decision in issuing the final trip. The proposal running at PDC is a two-layer algorithm. The first layer randomly samples the bus voltages at the adjacent substations considering different groups of digital relays. The relay which has issued the BFI may be compromised. It makes relays of the same make and family more susceptible to a cyber-attack exploiting the same vulnerabilities. Hence we propose grouping of relays by their make and relay families. The first layer is meant to determine if there is a fault in the vicinity of the BFP relay that issued the trigger. The second layer provides discrimination between fault and cyber-attack by measuring the impedance observed at the two ends of the perceived-faulted line. Since the proposed solution is computationally lightweight\, it adheres to the timing requirement of the BFP. The proposition requires healthy communication between the PMUs and the PDC. Nevertheless\, the proposed method is fail-safe. It will resort to the conventional BFP scheme in case of loss of communication with the PDC. The proposed solution mitigates n number of cyber-attacks in a no-fault scenario. Additionally\, the proposed solution can detect one cyber-attack if the attacker times the cyber-attack during a fault condition. PSCAD simulations were performed to validate the proposition on IEEE 118 bus system. Furthermore\, the hardware was developed emulating the PMU-PDC communication as per IEEE C37.118-2 standard\, and the execution time of the proposal was verified to ensure adherence to the timing requirement of the BFP. \n  \nALL ARE CORDIALLY INVITED \n 
URL:https://ee.iisc.ac.in/event/thesis-defence-of-mr-paawan-kirankumar-dubal/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20220926T163000
DTEND;TZID=Asia/Kolkata:20220926T173000
DTSTAMP:20260404T023037
CREATED:20220926T010719Z
LAST-MODIFIED:20220926T010758Z
UID:240014-1664209800-1664213400@ee.iisc.ac.in
SUMMARY:Thesis Defence of Mr. Ahmad Arfeen @ 11am
DESCRIPTION:Title:  Data Efficient Domain Generalization \nFaculty Advisor: Prof. Soma Biswas \nExaminer: Prof. Venkatesh Babu R. (CDS\, IISc) \nDate: Monday\, September 26\, 2022 \nTime: 11:00 AM \nVenue:  MMCR (EE dept) \nAbstract: For the task of image classification\, in general\, the test data is assumed to come from the same distribution as the training data. But this may not always hold in real-life scenarios. For example\, in night-time surveillance\, we may need to classify images captured using NIR cameras\, but the available model has been trained on RGB images. Domain generalization (DG) addresses the problem of generalizing classification performance across any unknown domain\, by leveraging training samples from multiple source domains. In this thesis\, we address two challenging scenarios for the DG task\, with focus on data efficiency. Currently\, the training process of majority of the state-of-the-art DG-methods is dependent on a large amount of labeled data. This restricts the application of the models in many real-world scenarios\, where collecting and annotating a large dataset is an expensive and difficult task. \nAs the first contribution\, we address the problem of Semi-supervised Domain Generalization (SSDG)\, where the training set contains only a few labeled data\, in addition to a large number of unlabeled data from multiple domains. This is relatively unexplored in literature and poses a considerable challenge to the state-of-the-art DG models\, since their performance degrades under such condition. To address this scenario\, we propose a novel Selective Mixing and Voting Network (SMV-Net)\, which effectively extracts useful knowledge from the set of unlabeled training data\, available to the model. Specifically\, we propose a mixing strategy on selected unlabeled samples on which the model is confident about their predicted class labels to achieve a domain-invariant representation of the data\, which generalizes effectively across any unseen domain. Extensive experiments on two popular DG-datasets demonstrate the usefulness of the proposed framework. \nThe second contribution of this thesis is a novel approach for the task of Zero-Shot Domain Generalization (ZSDG). This is very challenging since the query data can belong to an unseen class as well as unseen domain. For this task\, we address the challenge of class imbalance by learning class specific classifier margins\, which not only maintains the semantic relationship of the classes in the embedding space\, but is also discriminative\, and thus improves the classification performance on the test data. Extensive experiments on multiple datasets justify the effectiveness of the proposed approach. \n*************************************************************************************************** \nALL ARE WELCOME
URL:https://ee.iisc.ac.in/event/thesis-defence-of-mr-%e2%80%afahmad-arfeen-11am/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20220921T223000
DTEND;TZID=Asia/Kolkata:20220921T233000
DTSTAMP:20260404T023037
CREATED:20220920T230706Z
LAST-MODIFIED:20220920T231050Z
UID:239893-1663799400-1663803000@ee.iisc.ac.in
SUMMARY:Online Lecture by Dr. Mohammad Hedayati @5pm
DESCRIPTION:Title: High Voltage DC Circuit Breakers (HVDC CB) \nSpeaker: Dr. Mohammad Hedayati\, Dyson Technology\, UK. \nTime and Venue: 5 pm\, MS Teams/Online mode (Link provided below) \nAbstract: The main challenge of the DC meshed grid is the lack of suitable fault protection devices. The AC circuit breaker cannot be used in the DC system as in DC there is no zero crossing to extinguish the arc. Hence the DC circuit breakers need to create a virtual current zero crossing. There are different types of DC CB\, and three of them (solid state\, mechanical\, hybrid circuit breaker) are explained in this talk. The advantages and disadvantages of each technology are then pointed out. \nSpeaker Bio: Dr. Mohammad Hedayati did his Master and PhD in department of Electrical Engineering\, IISc between 2008 to 2016. Then he moved to University of Aberdeen as a postdoc\, where he was working on DC CB for DC meshed grids in 2016. In 2018 he joined University of Bristol working on GaN Devices Reliability and health monitoring and switching characteristics. In 2021 he joined Dyson technology\, during this time he was designing high speed (150kRPM) motor drive inverters for vacuum cleaner and personal care applications. In October 2022 he will be joining Jaguar Land Rover (owned by Tata group) to design EV motor drive inverters. \nAll are welcome. \nMicrosoft Teams meeting \nJoin on your computer or mobile app \nClick here to join the meeting
URL:https://ee.iisc.ac.in/event/online-lecture-by-dr-mohammad-hedayati-5pm/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20220909T200000
DTEND;TZID=Asia/Kolkata:20220909T210000
DTSTAMP:20260404T023037
CREATED:20220905T052514Z
LAST-MODIFIED:20220905T052514Z
UID:239884-1662753600-1662757200@ee.iisc.ac.in
SUMMARY:Thesis Colloquium of Anurenjan P. R.
DESCRIPTION:Venue\, Date and Time: MMCR\, EE\, IISc. 9-9-2022\, 2.30-3.30pm.\n\n\nTeams link – https://tinyurl.com/3bpmjxkx . \nTitle: Dereverberation of speech using frequency domain linear prediction \nFaculty Advisor: Dr. Sriram Ganapathy \nAbstract : The speech-based technologies are radically changing the way we interact with systems and how we access information. In many of these applications\, the users prefer to interact with the system through a far-field microphone without the nuance of a handheld or body-worn device. Examples of such applications are automated meeting analysis\, speech-based dictation systems\, hands-free interfaces for controlling consumer-products\, IoT\, virtual assistants in mobile phones and smart speakers. The major challenge in capturing speech from the far-field is the degradation of the signal quality due to reverberation. Reverberation refers to the delayed and weighted summation of the direct component of the speech signal with the reflected versions. This talk is focused on developing methods for speech dereverberation\, i.e.\, restoring the functional quality of reverberated speech\, using the signal analysis technique of frequency domain linear prediction (FDLP). \nThe FDLP is the frequency domain dual of the conventional Time Domain Linear Prediction (TDLP). Just as the TDLP estimates the spectrum of a signal\, the FDLP estimates the temporal envelopes of the signal using an autoregressive model. We apply the FDLP approach to the sub-bands of speech signal that are distributed in the mel scale. \nThis talk will describe two broad directions for addressing issues in the far-field speech using the FDLP approach. In the first part of the talk\, we explore a front-end design for automatic speech recognition (ASR) applications that suppresses the reverberation artifacts in the FDLP envelope. In the second part of the thesis\, we develop a speech enhancement model using the envelope and carrier decomposition given by the FDLP technique. \nIn the design of the ASR front end\, I will discuss a novel approach for 3-D acoustic modeling framework\, where the spatio-spectral features from all the sub-band channels are extracted. The features that are input to the 3-D CNN are extracted by modeling the signal peaks in the spatio-spectral domain using a multi-variate autoregressive modeling approach. In the subsequent part of this section\, I will describe a neural model for speech dereverberation using the long-term sub-band envelopes of speech. The neural dereverberation model estimates the envelope gain\, which when applied to reverberant signals\, allows the suppression of the late reflection components. The de-reverberated envelopes are used for feature extraction in speech recognition. The key novelty in this model is the joint learning of the reverberation and the ASR system. In these ASR experiments using the proposed framework\, we illustrate significant performance gains over previously proposed front ends. \nThe second part of the thesis deals with the FDLP based speech dereverberation for enhancement applications\, where the goal is to restore the audible quality of the speech signal. For this task\, we decompose the sub-band speech signal into the constituent envelope and carrier part. A dereverberation neural model is designed that attempts to enhance the envelope and carrier signals jointly. Further\, joint learning of the speech enhancement model with the end-to-end ASR model is proposed with a single neural framework. The proposed model therefore can generate improved audio quality and provide robust representations for far-field ASR. Finally\, I will illustrate the subjective quality improvement of the audio signal as well as the improvement in ASR performance obtained by the proposed envelope-carrier model. \nAcknowledgement \nThis work was partly supported by project grants from Samsung Research India\, Bangalore and the College of Engineering\, Trivandrum\, Kerala.  \nBio: Mr. Anurenjan is a PhD student at the LEAP lab\, Electrical Engineering\, IISc. He is also currently working as Assistant Professor in College of Engineering\, Trivandrum. Mr. Anurenjan completed his Bachelors in Technology from Government Engineering College\, Barton Hill\, Trivandrum\, Kerala in 2006 and his Masters in Technology from College of Engineering\, Trivandrum\, Kerala in 2008. He joined the LEAP lab as a PhD candidate under AICTE-QIP program in the year 2017. He hails from Trivandrum district of Kerala. He is interested in signal processing\, machine learning and speech processing. Mr. Anurenjan is a student member of IEEE SPS and the ISCA. During his free hours\, Mr. Anurenjan likes to play badminton and swimming.  \n——- \nAll are invited. Coffee/Tea will be served before the talk. 
URL:https://ee.iisc.ac.in/event/thesis-colloquium-of-anurenjan-p-r/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20220829T210000
DTEND;TZID=Asia/Kolkata:20220829T223000
DTSTAMP:20260404T023037
CREATED:20220826T083354Z
LAST-MODIFIED:20220826T083354Z
UID:239876-1661806800-1661812200@ee.iisc.ac.in
SUMMARY:Thesis Colloquium of Manish Tathode @3.30pm
DESCRIPTION:Title:  Fast and Compact Voltage Equalizer for Satellite Applications.Advisor: Prof. Vinod John.Date and Time: Monday\, 29th August 2022\, 3.30 pm.Location: MMCR\, EE Department.Meeting Link: click here for Meeting Link \nAbstract: Lithium-ion batteries have now become an critical constituent of Electrical Power System of solar-powered satellites due to their high energy density\, wider operating temperature range and better radiation tolerance. For compact realization and better space utilization\, the series-parallel connected Li-ion batteries are operated with currents close to the design limit of the cells\, speeding up the increase in the inherent initial imbalance in the individual cell voltages in a series connected stack\, demanding fast equalization. Active multicell-to-multicell equalization achieve fast equalization by efficient charge transfer among multiple cells in the series connected stack. PS-MAHB equalizer is a multicell-to-multicell equalizer\, with its open-loop control maintains high equalization current throughout the equalization. Its soft-switched operation and modularization abilities make it an attractive choice for space applications. However\, it lacks the necessary protective features and required redundancy essential for its use in space applications. Hence\, a Modified PS-MAHB (MPS-MAHB) equalizer is developed by incorporating necessary protection features and redundancy in the PS-MAHB equalizer. The Failure Mode Impact Analysis of the MPS-MAHB equalizer reveals that during the most likely switch short circuit failure mode\, the faulty part of the equalizer is disconnected by the protective device and the redundancy does not let the cell get out of the equalization.The existing static phase shift-based control of the equalizer causes direct dependency of the equalization currents on the cell voltages and limits the equalization current levels to lower than the design equalization current value when the cell voltages are lower. Thus\, the control works with reduced rate of equalization\, and causes the under-utilization of the equalizer hardware for significant duration of time in the charge-discharge cycle. A dynamic phase shift-based control is proposed to maximize the equalization current through the cells irrespective of the cell voltages which further increases the rate of equalization and improves the equalizer hardware utilization. In the simulation\, a significant improvement in the equalization rate compared to the static phase shift control is verified with the proposed dynamic phase shift-based control.The compact hardware realization of the equalizer hardware and the voltage sensors addresses the space-volume constraints in satellite applications. The equalizer hardware is realized as 4-cell equalizer modules\, and the compactness of the equalizer hardware is achieved by pushing the switching frequency to 1MHz reducing the values and sizes of the passive components. The challenges faced during the PCB design of the 4-cell equalizer module are addressed by the design. A non-isolated high-precision op-amp based voltage sensing scheme is developed to target the equalization band close to 10mV. The concept of easy-to-design motherboard-based interface is introduced\, which does not require any changes in the design of 4-cell equalizer module and the voltage sensor board\, irrespective of the cell connector geometry.The experimental results verify the operation of the equalizer showing the convergence of cell voltages from the initial imbalance of 300mV to the band of 10mV. The impact of the non-ideal dynamic response of the Li-ion cell voltage on the voltage-sensing-based control algorithm is discussed along with the necessary modifications incorporated in the control.We request your presence to the colloquium.All are welcome.
URL:https://ee.iisc.ac.in/event/thesis-colloquium-of-manish-tathode-3-30pm/
LOCATION:EE\, MMCR
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20220824T213000
DTEND;TZID=Asia/Kolkata:20220824T223000
DTSTAMP:20260404T023037
CREATED:20220822T043420Z
LAST-MODIFIED:20220822T044126Z
UID:239871-1661376600-1661380200@ee.iisc.ac.in
SUMMARY:Thesis Colloquium of Ms. Mani Madhoolika Bulusu
DESCRIPTION:Degree Registered: M.Tech.(Research) \nAdvisor: Prof. Chandra Sekhar Seelamantula \nTitle: Interpolation of Digital Elevation Models Using Generative Adversarial Networks \nVenue: Multimedia Classroom (MMCR)\, Department of Electrical Engineering\, IISc \nDate & Time: August 24\, 2022; 4 PM onward (coffee will be served during the talk) \nAbstract: A digital elevation model (DEM) is a three-dimensional representation of elevation data of a terrain such as a lunar terrain acquired using a Chandrayaan rover or a terrestrial terrain acquired by a reconnaissance aircraft. Terrestrial DEMs are used in hydrological modeling\, geomorphology\, and glaciology. Lunar DEMs can be used to locate natural resources and to identify prospective landing sites for exploratory missions. Hence\, high-quality\, reliable DEMs are of great significance. DEMs are generally captured using LiDAR (Light Detection and Ranging)\, stereophotogrammetry\, and time-of-flight cameras. However\, a reliable DEM cannot be constructed without adequate landmark points/features. This is the case with smooth terrains\, occlusions\, and multiple voids. The measurements are nonuniform in general. Hence\, there is a need for interpolation and void-filling techniques to estimate the elevation with a high accuracy. \nInverse distance weighting (IDW)\, De Launay triangulation\, and Kriging are some of the popular benchmark algorithms for interpolating scattered and nonuniformly spaced data. Manual parameter tuning\, inability to recover high-frequency information\, and high computational complexity are some of the issues that these traditional interpolation techniques suffer from. Deep Learning (DL) has proven to be effective in providing excellent results in the field of image processing and computer vision\, specifically in the tasks of super-resolution\, image in-painting\, extrapolation\, and segmentation. With the massive success of DL in several image processing and computer vision applications\, its applicability has been explored for solving the DEM interpolation problem as well. However\, convolutions are not readily defined if the measurements are nonuniform. Hence\, the recent DL based research on DEM interpolation has only focused only on regularly spaced data. \nWe address the realistic problem of DEM interpolation from irregularly spaced measurements\, with the density of measurements varying spatially. This is a new and unexplored direction in the deep learning setting. We propose a new and robust DL architecture based on Generative Adversarial Networks (GANs) to perform interpolation and result in a uniform DEM with a user-specified resolution. The generator comprises three modules: Learnable Distance Weighting (LDW)\, DEM in-painting\, and Continuous Convolution (CC). We designed the novel LDW module as a learnable counterpart to the popular IDW algorithm that operates on the distances between the measurements and grid locations. This reduces the problem to that of inpainting post the LDW transformation. The proposed method is evaluated on synthetically generated data and on standard publicly available NASA (LOLA LRO) datasets using the mean relative error and PSNR as performance metrics. Extensive experiments justify the effectiveness and accuracy of the proposed approach in comparison with the benchmark techniques. We conclude the presentation by discussing possible future directions for DL based DEM interpolation. \nAll are invited.
URL:https://ee.iisc.ac.in/event/thesis-colloquium-of-ms-mani-madhoolika-bulusu/
LOCATION:EE\, MMCR
END:VEVENT
END:VCALENDAR