Title: Continuous Depth Deep Learning with Uncertainty Aware Neural Ordinary Differential Equations
Day, Date, Time: Friday, 6th January, 5.15 pm to 6.15 pm.
Venue: Seminar hall (IE 211), Second Room, IEOR Building.
Abstract : Deep learning has led to tremendous progress in many fields of artificial intelligence like computer vision and natural language processing. However, designing a deep learning model is tedious process requiring expert knowledge and is a time-consuming process. Deep learning models involve a large number of parameters, which might be a challenge for learning from limited data. Further, they are not robust and may pose a problem for high-risk applications like autonomous driving and healthcare. In this talk, we discuss the continuous depth deep learning models based on differential equations, specifically the neural ordinary differential equations (NODE), which could overcome the drawbacks of the standard deep learning models to some extent. We discuss the design and development of uncertainty-aware neural ordinary differential equations considering the concepts from Bayesian learning such as Gaussian processes and variational inference. We demonstrate the effectiveness of the proposed techniques in solving image classification problems in computer vision.
Speaker Bio: Dr. Srijith P. K. is an Associate Professor at the Department of Computer Science and Engineering, IIT Hyderabad. He is interested in developing machine learning, deep learning and Bayesian learning models to solve problems arising in various domains such as vision, natural language processing and social media. He has published papers in top venues such as NeurIPS, AAAI, UAI, ACL, EMNLP, ECML, WACV, UMAP etc. He was the local organizing chair for ACML 2022 and has served as an area chair and reviewer for several ML and NLP conferences such as NeurIPS, ICML, AAAI, EMNLP, ACML etc. Prior to joining IITH, he worked as a post-doctoral researcher at the University of Melbourne and at the University of Sheffield. Dr. Srijith did his Ph.D. at the department of Computer Science and Automation, Indian Institute of Science, Bangalore. He holds an M.Tech degree in Computer Science from IIT Bombay and a B.Tech degree in Computer Science from NIT Calicut. More details on his research can be found at: https://sites.google.com/site/pksrijith/home.