Deep Generative Models:- From Theory to Applications

Generative Models are statistical methods that are used to generative samples from an unknown distributions. Deep Learning, owing to his ubiquitous nature, has dominated the area of generative models, with models such as Generative Adversarial Networks (GANs), Variational Auto-Encoders (VAEs) and Diffusion based models. In this talk, I shall give a rigorous overview of the general working principles of some of these models and discuss a few possible applications in domains such as healthcare.

Prathosh is currently a faculty member in the department of Electrical Communication Engineering at the Indian Institute of Science (IISc), Bangalore. He received his PhD from the same institute in 2015 after which he worked in corporate research labs including Xerox Research India, Philips research, and a start-up in CA, USA. His work in the industry, focussing on healthcare analytics, led to the generation of several IP, comprising 15 (US) patents of which 11 are granted and 6 are commercialized. He joined IIT Delhi in 2017 as an Assistant Professor in the computer technology group of Electrical Engineering where was engaged in research and teaching of the machine and deep learning courses. His current research includes guided deep-representational learning, cross-domain generalization, signal processing, and their applications in healthcare. He has co-founded a startup Cogniable.Tech which builds learning algorithms for behavioural healthcare (first-place winner of the recent AI startup challenge by Govt. of India) and also actively engaged with several corporate industries, start-ups, and medical centres (E.g., AIIMS) in solving interesting technical problems.

Abstract

Prathosh A P, IISc Bengaluru