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:20210101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20211117T193000
DTEND;TZID=Asia/Kolkata:20211117T203000
DTSTAMP:20260403T235518
CREATED:20211108T224619Z
LAST-MODIFIED:20211108T225102Z
UID:238961-1637177400-1637181000@ee.iisc.ac.in
SUMMARY:M.Tech.(Research) Thesis Defense of Mr. Vinayak Killedar
DESCRIPTION:Title of the thesis: Solving Inverse Problems Using a Deep Generative Prior\nSupervisor: Prof. Chandra Sekhar Seelamantula (EE)\nExaminer: Prof. Sumohana Channappayya (EE\, IIT Hyderabad) \nAbstract: The objective in an inverse problem is to recover a signal from its measurements\, given the knowledge of the measurement operator. In this thesis\, we address the problems of compressive sensing (CS) and compressive phase retrieval (CPR) using a generative prior model with sparse latent sampling. These problems are ill-posed and have infinite solutions. Structural assumptions such as smoothness\, sparsity and non-negativity are imposed on the solution to obtain a unique solution. \nThe standard CS and CPR formulations impose a sparsity prior on the signal. Recently\, generative modeling approaches have removed the sparsity constraint and shown superior performance over traditional CS and CPR techniques in recovering signals from fewer measurements. Generative model uses a pre-trained network\, the generator of a Generative Adversarial Network (GAN) or the decoder of a Variational Autoencoder (VAE) to model the distribution of the signal and impose a Set-Restricted Eigenvalue Condition (S-REC) on the measurement operator. The S-REC property places a condition on the l-2 norm of the difference in signal and measurement domain for signals coming from the set S. Solving CS and CPR using generative models have some limitations. The reconstructed signal is constrained to lie in the range-space of the generator. The reconstruction process is slow because the latent space is optimized through gradient-descent (GD) and requires several restarts. It has been argued that the distribution of natural images is not confined to a single manifold\, but a union of submanifolds. To take advantage of this property\, we propose a sparsity-driven latent space sampling (SDLSS) framework\, where sparsity is imposed in the latent space. The effect is to divide the latent space into subspaces such that the generator models map each subspace into a submanifold. We propose a proximal meta-learning (PML) algorithm to optimize the parameters of the generative model along with the latent code. The PML algorithm reduces the number of gradient steps required during testing and imposes sparsity in the latent space. We derive the sample complexity bounds within the SDLSS framework for the linear CS model\, which is a generalization of the result available in the literature. The results demonstrate that\, for a higher degree of compression\, the SDLSS method is more efficient than the state-of-the-art deep compressive sensing (DCS) method. We consider both linear and learned nonlinear sensing mechanisms\, where the nonlinear operator is a learned fully connected neural network or a convolutional neural network and show that the learned nonlinear version is superior to the linear one. \nAs an application of the nonlinear sensing operator\, we consider compressive phase retrieval\, wherein the problem is to reconstruct a signal from the magnitude of its compressed linear measurements. We adapt the S-REC imposed on the measurement operator and propose a novel cost function. The SDLSS framework along with PML algorithm is applied to optimize the sparse latent space such that the adapted $\mathcal{S}$-REC loss and data-fitting error are minimized. The reconstruction process is fast and requires few gradient steps during testing compared with the state-of-art deep phase retrieval technique. \nExperiments are conducted on standard datasets such as MNIST\, Fashion-MNIST\, CIFAR-10\, and CelebA to validate the efficiency of SDLSS framework for CS and CPR. The results show that\, for a given dataset\, there exists an effective input latent dimension for the generative model. Performance quantification is carried out by employing three objective metrics: peak signal-to-noise ratio (PSNR)\, structural similarity index measure (SSIM)\, and reconstruction error (RE) per pixel\, which are averaged across the test dataset. \nAbout the speaker: Vinayak Killedar obtained a B.E. (ECE) degree from M. S. Ramaiah Institute of Technology (MSRIT)\, Bangalore in 2008. During 2008-2010\, he worked for Robert Bosch Engineering and Business Solution (RBEI)\, Coimbatore. He joined the M.Tech.(Signal Processing) program in National Institute of Technology (NIT) Calicut and graduated in 2013. He worked for Continental AG during 2014-2018 in the areas of autonomous driving and radar signal processing. Subsequently\, he joined the Spectrum Lab\, Department of Electrical Engineering\, Indian Institute of Science for M.Tech.(Research) and specialized in Compressed Sensing and Machine Learning. He is presently a Senior Technical Specialist at Ansys\, Kempten\, Germany.
URL:https://ee.iisc.ac.in/event/m-tech-research-thesis-defense-of-mr-vinayak-killedar/
LOCATION:Online\, India
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20211117T164500
DTEND;TZID=Asia/Kolkata:20211117T180000
DTSTAMP:20260403T235518
CREATED:20211108T230750Z
LAST-MODIFIED:20211116T032837Z
UID:238966-1637167500-1637172000@ee.iisc.ac.in
SUMMARY:PhD Thesis Defense of Mr. Shome Subhra Das
DESCRIPTION:Date and Time: November 17\, 2021 (Wednesday)  11.15 AM\nClick here to join the meetingExternal Examiner: Prof. Gaurav Harit\, IIT Jodhpur\n\nTitle: Techniques for estimating the direction of pointing gestures using depth images in the presence of orientation and distance variations from the depth sensor\nAbstract: Currently\, we interact with computers\, robots\, drones\, and virtual reality interfaces using pointing devices such as mouse\, touchpad\, joystick\, virtual reality wand\, drone controller\, etc. These devices have one or more of the following limitations: being cumbersome\, non-immersive\, immobile\, and having a steep learning curve. The target of this work is to explore ways to replace existing pointing devices with pointing gesture-based interfaces.  \n This thesis addresses two problems\, namely estimating the direction being indicated by a pointing gesture (PDE) and detection of pointing gestures. The proposed techniques use a single depth sensor and use only the hand region. To our knowledge\, this is the maiden attempt at creating depth and orientation tolerant\, accurate methods for estimating the pointing direction using only depth images of the hand region. The proposed methods achieve accuracies comparable to or better than those of existing methods while avoiding their limitations.  \n Significant contributions of the thesis:  \n (i) Proposing a real-time technique for estimating the pointing direction using a nine-axis inertial motion unit (IMU) and an RGB-D sensor. It is the first method to compute the pointing direction (PD) by finding the axis vector of the index finger. It is also the first method to fuse information from the IMU and depth sensor to obtain the PD. Further\, this is the first method to obtain the ground-truth pointing direction of pointing gestures using depth data of the index finger region.  \n (ii) Creation of a large (100k+ samples) dataset with accurate ground truth for PDE from depth images. Each sample consists of the segmented depth image of a hand\, the fingertip location (2D + 3D)\, the pointing vector (as a unit vector and in terms of the yaw and pitch values)\, and the mean depth of the hand. This is the first public dataset for depth image based PDE that has accurate ground truth and a large number of samples.  \n(iii) Proposing a new 3D convolutional neural network-based method to estimate pointing direction. This is the first deep learning-based method for PDE that uses only the depth images of the hand region for PDE\, without the use of RGB data. It is tolerant to variation in orientation and depth of the hand with respect to the camera and is suitable for real-time applications.  \n (iv) Proposing another technique for estimating the pointing direction using global registration of the test data point cloud with a pointing hand model captured using Kinect fusion-based method. It is tolerant to the variation in the orientation and depth of the hand w.r.t. the RGB-D sensor. It does not have the limitation of the previously proposed methods since it does not require the attachment of any device such as IMU nor does it require any dataset for training. It achieves less net angular error than most techniques in the literature using only the hand region.  \n (v) Creation of a large dataset of positive and negative samples for detection of pointing gestures from depth images of the hand region. A technique is also proposed using deep learning to distinguish pointing gestures from other hand gestures. This achieves higher accuracy than the only other existing technique by Cordo et al.  for detection of pointing gestures from depth images of the hand. 
URL:https://ee.iisc.ac.in/event/phd-thesis-defense-of-mr-shome-subhra-das/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20211014T160000
DTEND;TZID=Asia/Kolkata:20211015T050000
DTSTAMP:20260403T235518
CREATED:20211110T013900Z
LAST-MODIFIED:20211110T013900Z
UID:239069-1634227200-1634274000@ee.iisc.ac.in
SUMMARY:Thesis Defence of Indla Rajitha Sai Priyamvada
DESCRIPTION:Thesis Title: Analysis and Enhancement of Stability of Power Systems with Utility-scale Photovoltaic Power Plants \nGuide: Dr. Sarasij Das \nAbstract: Owing to the negative impact of carbon emissions on the environment\, power systems are experiencing a paradigm shift in power generation. The fossil fuel-based generators that utilize synchronous machines are increasingly being replaced by the renewables such as Photovoltaic (PV) generators. Utility-scale PV power plants are coming up in the various parts of the world. Power electronic interface\, control strategies and lack of inherent rotational element are the main factors that distinguish PV generation from Synchronous Generators (SGs). In addition\, the time constants of the PV control loops and Phase Locked Loop (PLL) are of the same order unlike the SGs. The power electronic interface offers a better control over the electrical energy generated by the PV generators. However\, the power electronic interface brings new challenges to power system stability. This research work focuses on addressing transient and small-signal stability issues of grid connected utility scale PV power plants.\nIn conventional power systems\, swing equation of SGs and (extended) equal area criterion are used to assess the transient stability of power system. However\, the same analysis techniques may not be applicable for PV generators. In this research work\, transient stability assessment criteria are developed for grid connected PV generator with two different control strategies viz.\, Vdc-Q control and PQ control (with/without support functionalities). The proposed criteria are developed considering the outer and inner control loop\, PLL and filter dynamics of PV generator. PSCAD simulations are carried out on a two-bus system and a modified IEEE-39 bus system to validate the proposed criterion. The stability criteria are found to effectively assess the stability of grid connected utility scale PV generators.\nThe power transfer capability of transmission network is limited by thermal limits\, voltage limits and stability limits. Power transfer capability of transmission lines emanating from PV generators considering thermal and voltage limits is explored well in the literature. However\, there is a lack of literature on stability constrained power transfer capability limit. In this research work\, adaptive control-based tuning laws are proposed for grid connected PV generators to improve the stability constrained power transfer capability. The adaptive tuning laws are derived based on the Lyapunov energy function analysis. The Lyapunov functions are formulated using the summation of squares of the PI block errors and difference between the PI parameter values from their optimal values. Time domain simulations are carried out on a two-bus system and a modified IEEE-39 bus system to validate the proposed tuning laws. From time domain simulations\, it is observed that the proposed tuning laws are found to effectively improve the stability limit on power transfer to the voltage limit.\nThe increased penetration of PV generations into power systems has also brought qualitative changes in small signal stability of power systems. Two new categories of oscillation modes are introduced into power systems which have participation from PV state variables. As the mode shape of the two new categories of oscillation modes is different from that of SG modes\, the power system stabilizer design should be revisited. In this research work\, control-based power system stabilizer is developed considering the controllability and observability of the new categories of oscillation modes. The effectiveness of the developed stabilizer in providing sufficient damping to the new categories of oscillation modes is validated through PSCAD simulations on a modified IEEE-39 bus system.\nAs power systems are large interconnected systems\, the increased penetration of PV generation has resulted in notable interaction among PV generators and SGs. Investigation of the interaction among generators is important to understand the dynamic behaviour of overall power system when subjected to disturbances. This research work is carried out to understand the interaction among PV and SGs. The interaction is analysed through investigation of interaction among oscillation modes of PV generation and SG. A mathematical formulation to quantify the interaction among the oscillation modes of PV generations and SGs is proposed. A modified IEEE-39 bus system is considered to carry out the interaction study and validate the results obtained from mathematical formulations.
URL:https://ee.iisc.ac.in/event/thesis-defence-of-indla-rajitha-sai-priyamvada/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20211011T210000
DTEND;TZID=Asia/Kolkata:20211011T220000
DTSTAMP:20260403T235518
CREATED:20211110T012956Z
LAST-MODIFIED:20211110T014300Z
UID:239067-1633986000-1633989600@ee.iisc.ac.in
SUMMARY:Seminar by Dr. Prem Ranjan
DESCRIPTION:Title: Sustainability through High Voltage Engineering and Research \nAbstract : Through this talk\, we will take a glance at applications of high voltage engineering in three different sustainable technologies. (a) Direct application of high voltage and pulsed power will be shown for economical generation of nanoparticles (NPs) through wire explosion process (WEP). Control of NPs size\, phase and formation mechanism will be discussed through modelling studies and different material characterisation techniques. Application of WEP-synthesized semiconductor NPs will be discussed for wastewater treatment. (b) Then\, we will go through the need of high voltage electric system in more electric aircraft (MEA) to reduce the carbon footprint. Different tools available to evaluate the arc faults and damage caused to the neighbouring systems will be detailed through mathematical and experimental tools. (c) Drive towards sustainable environment is leading to search of SF6 alternatives in power equipment\, which are responsible for more than 80% of total SF6 emission. Research towards SF6 alternatives in gas insulated systems to reduce the global warming potential will be discussed in brief. Finally\, the prospective of high voltage engineering and research in some other areas will be discussed. \nSpeaker Biodata: Prem Ranjan is working as a postdoc researcher at High Voltage Lab\, The University of Manchester\, UK\, since Nov. 2019. He obtained the B. Tech. degree in Electrical and Electronics Engineering from NIT Calicut\, India in 2015 and the integrated MS\, PhD degrees in Electrical Engineering from IIT Madras\, Chennai\, India in 2019. He worked as an exchange researcher at Nagaoka University of Technology\, Japan for 4 months during 2016 and 2017. His research interests are focused on sustainable applications of high voltage engineering including exploding wire\, gas insulation (SF6 alternatives)\, arc-tracking in more electric aircraft and condition monitoring of power apparatus. \n 
URL:https://ee.iisc.ac.in/event/seminar-by-dr-prem-ranjan/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20210929T163000
DTEND;TZID=Asia/Kolkata:20210929T173000
DTSTAMP:20260403T235518
CREATED:20211110T014011Z
LAST-MODIFIED:20211110T014011Z
UID:239071-1632933000-1632936600@ee.iisc.ac.in
SUMMARY:PhD Thesis Defence of Aravind Illa
DESCRIPTION:Thesis Title: Acoustic-Articulatory Mapping: Analysis and Improvements with Neural Network Learning Paradigms \nAbstract:  Human speech is one of many acoustic signals we perceive\, which carries linguistic and paralinguistic (e.g: speaker identity\, emotional state) information. Speech acoustics are produced as a result of different temporally overlapping gestures of speech articulators (such as lips\, tongue tip\, tongue body\, tongue dorsum\, velum\, and larynx) each of which regulates constriction in different parts of the vocal tract. Estimating speech acoustic representations from articulatory movements is known as articulatory-to-acoustic forward (AAF) mapping i.e.\, articulatory speech synthesis. While estimating articulatory movements back from the speech acoustics is known as acoustic-to-articulatory inverse (AAI) mapping. These acoustic-articulatory mapping functions are known to be complex and nonlinear. \nComplexity of this mapping depends on a number of factors. These include the kind of representations used in the acoustic and articulatory spaces. Typically these representations capture both linguistic and paralinguistic aspects in speech. How each of these aspects contributes to the complexity of the mapping is unknown. These representations and\, in turn\, the acoustic-articulatory mapping are affected by the speaking rate as well. The nature and quality of the mapping varies across speakers. Thus\, complexity of mapping also depends on the amount of the data from a speaker as well as number of speakers used in learning the mapping function. Further\, how the language variations impact the mapping requires detailed investigation. This thesis analyzes few of such factors in detail and develops neural network based models to learn mapping functions robust to many of these factors. \nElectromagnetic articulography (EMA) sensor data has been used directly in the past as articulatory representations (ARs) for learning the acoustic-articulatory mapping function. In this thesis\, we address the problem of optimal EMA sensor placement such that the air-tissue boundaries as seen in the mid-sagittal plane of the real-time magnetic resonance imaging (rtMRI) is reconstructed with minimum error. Following optimal sensor placement work\, acoustic-articulatory data was collected using EMA from 41 subjects with speech stimuli in English and Indian native languages (Hindi\, Kannada\, Tamil and Telugu) which resulted in a total of ~23 hours of data\, used in this thesis. Representations from raw waveform are also learnt for AAI task using convolutional and bidirectional long short term memory neural networks (CNN-BLSTM)\, where the learned filters of CNN are found to be similar to those used for computing Mel-frequency cepstral coefficients (MFCCs)\, typically used for AAI task. In order to examine the extent to which a representation having only the linguistic information can recover ARs\, we replace MFCC vectors with one-hot encoded vectors representing phonemes\, which were further modified to remove the time duration of each phoneme and keep only phoneme sequence. Experiments with phoneme sequence using attention network achieve an AAI performance that is identical to that using phoneme with timing information\, while there is a drop in performance compared to that using MFCC. \nExperiments to examine variation in speaking rate reveal that\, the errors in estimating the vertical motion of tongue articulators from acoustics with fast speaking rate\, is significantly higher than those with slow speaking rate. In order to reduce the demand for data from a speaker\, low resource AAI is proposed using a transfer learning approach. Further\, we show that AAI can be modeled to learn acoustic-articulatory mappings of multiple speakers through a single AAI model rather than building separate speaker-specific models. This is achieved by conditioning an AAI model with speaker embeddings\, which benefits AAI in seen and unseen speaker evaluations. Finally\, we show the benefit of estimated ARs in voice conversion application. Experiments revealed that ARs estimated from speaker independent AAI preserves linguistic information and suppress speaker-dependent factors. These ARs (from unseen speaker and language) are used to drive target speaker specific AAF to synthesis speech\, which preserves linguistic information and target speaker’s voice characteristics.
URL:https://ee.iisc.ac.in/event/phd-thesis-defence-of-aravind-illa/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20210806T213000
DTEND;TZID=Asia/Kolkata:20210806T223000
DTSTAMP:20260403T235518
CREATED:20211110T014202Z
LAST-MODIFIED:20211110T014202Z
UID:239074-1628285400-1628289000@ee.iisc.ac.in
SUMMARY:PhD Thesis Colloquium of Praveen Kumar Pokala @ 4pm
DESCRIPTION:Event : Thesis Colloquium\nTitle : Robust Nonconvex Penalties for Solving Sparse Linear Inverse Problems and Applications to Computational Imaging\nSpeaker : Praveen Kumar Pokala\nDegree Registered : PhD\nAdvisor : Prof. Chandra Sekhar Seelamantula\nDate : 06/08/2021\nVenue : Online\nAbstract : Sparse linear inverse problems require the solution to the l-0-regularized least-squares cost\, which is not computationally tractable. Approximate and computationally tractable solutions are obtained by employing convex/nonconvex relaxations of the l-0-pseudonorm. One such approximation is obtained by considering the l-1-norm\, which is a convex relaxation of the l-0-pseudonorm. However\, l-1 regularization is known to result in biased estimates due to over-relaxation of the l-0-pseudonorm but it comes with the advantage of convexity of the regularized least-squares cost. Several nonconvex approximations of the l-0 pseudonorm have been proposed to overcome the bias introduced by the l-1-norm and to ensure better sparsity. However\, certain aspects of nonconvex sparse regularization have not been explored. Some of these are as follows:\nNonconvex sparse priors have been explored in the synthesis-sparse framework\, but not in the analysis-sparse framework due to the unavailability of proximal operators in closed-form in the analysis setting. Existing nonconvex approaches attach the same regularization weights across all the components of a sparse vector and treat them as fixed hyperparameters. Considering different weights for the entries and adapting them iteratively is likely to result in a superior performance.\nPrior learning networks based on deep-unfolded architectures for solving nonconvex penalties have not been explored. This thesis addresses the above aspects in three parts and considers applications to various computational imaging problems.\nPart-1: Nonconvex Analysis-sparse Recovery\nIn this part\, we solve the analysis-sparse recovery problem based on three regularization approaches:\nConvexity-preserving nonconvex regularization: We propose the analysis variants of the generalized Moreau envelope and generalized minimax concave penalty (GMCP) over a complex domain. Since the cost is a real-valued function defined over a complex domain\, it is nonholomorphic\, i.e.\, it does not satisfy Cauchy-Riemann (CR) conditions. To circumvent this problem\, we rely upon on Wirtinger calculus to derive the proximal operator for the analysis l-1 prior and develop an efficient optimization strategy employing projected proximal algorithms. The projection transform maps the analysis-sparse recovery problem into an equivalent constrained synthesis-sparse formulation.\nNonconvex sparse regularization: We consider the problem of nonconvex analysis sparse recovery in which the signal is assumed to be sparse in a redundant analysis operator. Standard nonconvex sparsity promoting priors do not have a proximal operator in closed-form under a redundant analysis operator and therefore\, proximal approaches cannot be applied directly. This led us to develop two alternatives — Moreau envelope regularization and projected transformation.\nGeneralized weighted l-1 regularization: We develop a generalized weighted l-1 regularization strategy\, which allows for efficient weight-update strategies for iteratively reweighted l-1-minimization under tight frames. Further\, we impose sufficient conditions on the weight function that leads to a reweighting strategy\, which follows the interpretation originally given by Candès et al.\, but is more efficient than theirs. Since the objective function is nonholomorphic\, we resort to Wirtinger calculus for deriving the update equations. We develop an algorithm called generalized iteratively reweighted soft-thresholding algorithm (GIRSTA) and its fast variant\, namely\, generalized fast iteratively reweighted soft-thresholding algorithm (GFIRSTA). We provide convergence guarantees for GIRSTA and empirical convergence results for GFIRSTA.\nWe demonstrate the efficacy of the proposed regularization strategies in comparison with the benchmark techniques considering compressive-sensing magnetic resonance image (CS-MRI) reconstruction under a redundant analysis operator\, more specifically\, shift-invariant discrete wavelet transform (SIDWT).\nPart-2: Weighted Minimax Concave p-pseudonorm Minimization\nIn this part\, we develop techniques for accurate low-rank plus sparse matrix decomposition (LSD) and low-rank matrix recovery. We proposed weighted minimax-concave penalty (WMCP) as the nonconvex regularizer and show that it admits a certain equivalent representation that is more amenable to weight adaptation. Similarly\, an equivalent representation to the weighted matrix gamma norm (WMGN) enables weight adaptation for the low-rank part. The optimization algorithms are based on the alternating direction method of multipliers. The optimization frameworks relying on the two penalties\, WMCP and WMGN\, coupled with a novel iterative weight-update strategy\, result in accurate low-rank plus sparse matrix decomposition and low-rank matrix recovery techniques. Further\, we derive an algorithm\, namely\, iteratively reweighted MGN (iReMaGaN) algorithm\, which has a superior low-rank matrix recovery performance. The proposed algorithms are shown to satisfy descent properties and convergence guarantees. On the applications front\, we consider the problems of foreground-background separation and image denoising. Simulations and validations on standard datasets show that the proposed techniques outperform the benchmark techniques. Next\, we extended the idea to obtain a generalized l-p-penalty\, namely\, minimax concave p-pseudonorm (MCpN) based on a novel p-Huber function as the sparsity promoting function\, and its weighted counterpart\, weighted MCpN (WMCpN) as a regularizer for solving the sparse linear inverse problem. WMCpN is a generalization of which several penalties\, namely\, l-1-norm\, minimax concave penalty (MCP)\, l-p penalty\, weighted l-1-norm\, and weighted l-p penalty become special cases. However\, MCpN and WMCpN regularizers do not have closed-form proximal operators\, which makes the optimization problem challenging. To overcome this hurdle\, we develop an equivalent representation that is more amenable to optimization and allows for an analytical weight-update strategy. MCpN is a special case of WMCpN where all the weights are fixed and equal. The optimization algorithms are based on the alternating direction method of multipliers. Considering the application of interferometric phase estimation\, we demonstrate that MCpN and WMCpN result in accurate interferometric phase estimation. Simulations and experimental validations on standard datasets show that the proposed techniques outperform the benchmark techniques.\nPart-3: Nonconvex Sparse Regularization and Deep-Unfolding\nIn the final part\, we transition from fixed analytical priors to data-driven priors. To begin with\, we develop a deep-unfolded architecture\, namely\, FirmNet\, for sparse recovery. FirmNet has two parameters — one that controls the noise variance\, and the other that allows for explicit sparsity control. We show that FirmNet is better than Learned-ISTA (LISTA) by at least three-fold in terms of the probability of error in support (PES)\, and about 2 to 4 dB higher reconstruction SNR. Further\, we solve the problem of reflectivity inversion\, which deals with estimating the subsurface structure from seismic data through FirmNet. As an application\, we consider the problem of seismic reflectivity inversion. We demonstrate the efficacy of FirmNet over the benchmark techniques for the reflectivity inversion problem by testing on synthetic 1-D seismic traces and 2-D wedge models. We also report validations on simulated 2-D Marmousi2 model and real data from the Penobscot 3D survey off the coast of Nova Scotia\, Canada. Next\, we propose convolutional FirmNet (ConFirmNet)\, which is an extension of the FirmNet approach to solve the problem of convolutional sparse coding. As an application\, we build a ConFirmNet based sparse autoencoder (ConFirmNet-SAE) and demonstrate suitability for image denoising and inpainting. Further\, we also show that training ConFirmNet-SAE with the Huber loss imparts robustness to outliers. ConFirmNet-SAE also proves to be robust to mismatch between training and test noise conditions than convolutional learned iterative soft-thresholding algorithm (CLISTA). Finally\, we propose a sparse recovery formulation that employs a nonuniform\, nonconvex synthesis sparse model comprising a combination of convex and nonconvex regularizers\, which results in accurate approximations of the l-0 pseudo-norm. The resulting iterative optimization employs proximal averaging. When unfolded\, the iterations give rise to a nonuniform sparse proximal average network (NuSPAN) that can be optimized in a data-driven fashion. We demonstrate the efficacy of NuSPAN also for solving the problem of seismic reflectivity inversion.\nSpeaker Biodata : Praveen Kumar Pokala received his B.Tech. degree in Electronics and Telecommunication Engineering from Jawaharlal Nehru Technological University\, Hyderabad\, India\, in 2006 and M. Tech degree in Signal Processing from Indian Institute of Technology (IIT)\, Guwahati\, India\, in 2009. Subsequently\, he worked as an Assistant Professor in LPU university\, Jalandhar\, India and GITAM university\, Hyderabad\, India. He is currently pursuing Ph.D. in the Department of Electrical Engineering\, Indian Institute of Science\, Bangalore. His current research interests are machine learning\, deep learning\, and nonconvex optimization algorithms\, with applications to inverse problems in computational imaging.
URL:https://ee.iisc.ac.in/event/phd-thesis-colloquium-of-praveen-kumar-pokala-4pm/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20210801T053000
DTEND;TZID=Asia/Kolkata:20210802T053000
DTSTAMP:20260403T235518
CREATED:20211124T230306Z
LAST-MODIFIED:20211207T014334Z
UID:239340-1627795800-1627882200@ee.iisc.ac.in
SUMMARY:Events earlier to Aug 2021
DESCRIPTION:Click to see the Content
URL:https://ee.iisc.ac.in/event/events-earlier-to-aug-2021/
END:VEVENT
END:VCALENDAR