Welcome to the third iteration of our Knowledge Sharing Sessions! This week, we introduce Saikat Chatterjee, Associate Professor at KTH, who will speak on the ability and inability of deep neural networks in the perspective of intelligence!
Abstract: In this seminar I will raise questions on ability and inability of deep neural networks in the general perspective of intelligence. What do we expect from an intelligent system? Can we think Artificial Intelligence is really intelligent? How do we define it? For example, if an intelligent system learns with hard examples, what the system should be expected to do for simple examples? We will check what computer does for an autonomous driving application context as a computer vision problem. Then, I will raise the question: can a neural network learn its size? Can it learn number of layers and number of nodes by itself as a self size-learning net – does it have that level of intelligence? Can a neural network generate pictures? Can it produce interpolated pictures which do not exist in reality? We will try to argue from signals/systems ways, and see that it is hard to decide and take a side on this ability-inability issue.
Mode of the seminar: This will be an interactive seminar! Hopefully, we will finish within the stated 1.5 hours.
Saikat Chatterjee is an associate professor at the School of Electrical Engineering and Computer Science, KTH-Royal Institute of Technology, Sweden. He received a Ph.D. degree from the Indian Institute of Science, India. He has published more than 120 papers in international journals and conferences. He was a co-author of the paper that won the best student paper award at ICASSP 2010. His current research interests are signal processing, machine learning, bioinformatics, data analytics, and speech and audio processing.
Join the KTH AI Society and gain access to Slack where you can communicate with others interested in the same field as you and get a quick insight on the organization!
From KTH to Spotify: Journey of ML Engineers Alva Liu and Emma Schüldt February 2, 2024 / Arena, KTH Innovation, Stockholm, Sweden
Extraordinary General Meeting December 14, 2023 / Google Meet, Stockholm, Sweden
An introduction to Generative AI: Multimodality and Boosting Specialization and Performance in LLMs3 December 7, 2023 / Arena, KTH Innovation, Stockholm, Sweden