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:20240101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20240412T093000
DTEND;TZID=Asia/Kolkata:20240412T110000
DTSTAMP:20260527T111944
CREATED:20240409T063344Z
LAST-MODIFIED:20240409T063642Z
UID:241435-1712914200-1712919600@ee.iisc.ac.in
SUMMARY:Colloquium on  Estimation of flashovers in the EHV/UHV lines on the east  coast due to lightning produced by the Bay of Bengal cyclones
DESCRIPTION:Name of the Candidate:       Anirban Chatterjee \nTitle of the Thesis:              Estimation of flashovers in the EHV/UHV lines on the east coast due to lightning produced by the Bay of Bengal cyclones \nDegree Registered:                MTech (Res) in Electrical Engineering \nTime and date:                     9.30 AM\, 12th April 2024 \nVenue:                                   High Voltage Laboratory Seminar Hall of EE Department \nResearch Supervisor:           Professor Udaya Kumar \nAbstract \nThe Bay of Bengal produces a considerable number of cyclones. Many of them invade the east coast of India. They can cause structural damage to towers\, substation flooding\, conductor snapping\, etc. In many cases\, through lightning\, they cause several flashovers on the EHV/UHV grid. However\, there is no serious effort to estimate the possible number of flashovers caused by the lightning produced by such cyclones. The present work aims to fill this serious gap. \nThe estimation of such lightning-induced flashovers requires several aspects\, both electrical and cyclone-related. The lightning stroke could be intercepted by the tower/ground-wire\, or it can strike the phase conductor. The electro-geometric model(EGM)\, suggested in IEEE standards\, is employed for assessing the normalized number of strokes striking the phase conductor and intercepted by the tower/ground-wire. The associated probabilities are also estimated for typical EHV and UHV lines. \nThe relation between the peak stroke current and the rise time is made based on the literature. Then\, by modelling the lines in EMTP with a multi-story model for the tower\, simulations are carried out to deduce the corresponding voltage rise. Using this information and the BIL of the line\, the possibility of flashovers is assessed. \nThe trajectory of the cyclone and the speed\, along with the number of lightning flashes produced by them are assimilated from different sources. Modelling the cyclone as a disc above ground\, the line length shadowed as a function of time is calculated. In addition\, equivalent ground flash density per square km per hour is also calculated. Combining all these information\, the possible number of lighting-induced flashovers in the EHV/UHV grid along the east coast is estimated.  Within three to four days the number of such flashover incidents can range from a few to few tens of flashovers within a time span of two to four days.
URL:https://ee.iisc.ac.in/event/colloquium-on-estimation-of-flashovers-in-the-ehv-uhv-lines-on-the-east-coast-due-to-lightning-produced-by-the-bay-of-bengal-cyclones/
LOCATION:HV seminar Hall
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=Asia/Kolkata:20240412T160000
DTEND;TZID=Asia/Kolkata:20240412T173000
DTSTAMP:20260527T111944
CREATED:20240412T043837Z
LAST-MODIFIED:20240412T043837Z
UID:241439-1712937600-1712943000@ee.iisc.ac.in
SUMMARY:Talk On Making Machine Learning Models Safer and Better: Data and Model Perspectives
DESCRIPTION:  \nAbstract: As machine learning systems are increasingly deployed in real-world settings like healthcare\, finance\, and scientific applications\, ensuring their safety and reliability is crucial. However\, many state-of-the-art ML models still suffer from issues like poor out-of-distribution generalization\, sensitivity to input corruptions\, requiring large amounts of data\, and inadequate calibration – limiting their robustness and trustworthiness for critical real-world applications. \nIn this talk\, I will present a broad overview of different safety considerations for modern ML systems. I will then proceed to discuss our recent efforts in making ML models safer from two complementary perspectives – (i) manipulating data and (ii) enriching the model capabilities by developing novel training mechanisms. First\, I will discuss our work on designing new data augmentation techniques for object detection followed by demonstrating how\, in the absence of data from desired target domains of interest\, one could leverage pre-trained generative models for efficient synthetic data generation. Next\, I will introduce a new paradigm of training deep networks called model anchoring and show how one could achieve similar properties to an ensemble but through a single model. I will specifically discuss how model anchoring can significantly enrich the class of hypothesis functions being sampled and demonstrate its effectiveness through its improved performance on several safety benchmarks. Finally\, I will present our efforts in proactively identifying samples on which a model would fail through a novel model counterfactual synthesis technique by leveraging foundation models (e.g.\, GPT family and CLIP). I will then conclude by highlighting exciting future research directions for producing robust ML models through leveraging multi-modal foundation models. \n\nBio: Kowshik Thopalli is a Machine Learning Scientist and a post-doctoral researcher at Lawrence Livermore National Laboratory. He is currently mentored by Dr. Jay Thiagarajan. His research focuses on developing reliable machine learning models that are robust under distribution shifts. He has published papers on a variety of techniques to address model robustness\, including domain adaptation\, domain generalization\, and test-time adaptation using geometric and meta-learning approaches. His expertise also encompasses integrating diverse knowledge sources\, such as domain expert guidance and generative models\, to improve model data efficiency\, accuracy\, and resilience to distribution shifts. He received his Ph.D. in 2023 from Arizona State University under the mentorship of Dr. Pavan Turaga.\n______\n\nAll are welcome.
URL:https://ee.iisc.ac.in/event/talk-on-making-machine-learning-models-safer-and-better-data-and-model-perspectives/
LOCATION:B306 \, EE Dept
END:VEVENT
END:VCALENDAR