Deep generative view on continual learning

Tomasz Trzciński

Course Summary | About the lecturer | Location and schedule Videos | Assignment

Registration form:

Register here

Course summary:

Neural networks suffer from catastrophic forgetting, defined as an abrupt performance loss on previously learned tasks when acquiring new knowledge. For instance, if a network previously trained for detecting virus infections is now retrained with data describing a recently discovered strain, the diagnostic precision for all previous ones drops significantly. To mitigate that, we can retrain the network on a joint dataset from scratch, yet it is often infeasible due to the size of the data, or impractical when retraining requires more time than it takes to discover a new strain. The catastrophic forgetting severely limits the capabilities of contemporary neural networks and continual learning aims to address this pitfall. During the course, we will introduce continual learning as a domain of machine learning, define its main challenges and existing methods. We will then look at the approaches inspired by recent neuroscientific works, specifically at the generative models employed in continual learning scenario. Based on our research where we hypothesize that the unsupervised way of incorporating knowledge by generative models corresponds to the way biological systems continually learn, we will introduce generative replay as a method to overcome catastrophic forgetting. We will then further explore the landscape of continual learning from a deep generative modeling point of view.

About the lecturer:

Tomasz Trzciński (DSc, WUT'20; PhD, EPFL'14; MSc, UPC/PoliTo'10) is an Associate Professor at Warsaw University of Technology since 2015, where he leads a Computer Vision Lab, and an Assistant Professor at Jagiellonian University of Cracow (GMUM). He was a Visiting Scholar at Stanford University in 2017 and at Nanyang Technological University in 2019. Previously, he worked at Google in 2013, Qualcomm in 2012 and Telefónica in 2010. He is an Associate Editor of IEEE Access and MDPI Electronics and frequently serves as a reviewer in major computer science conferences (CVPR, ICCV, ECCV, NeurIPS, ICML) and journals (TPAMI, IJCV, CVIU). He is a Senior Member of IEEE and an expert of National Science Centre and Foundation for Polish Science. He is a Chief Scientist at Tooploox and a co-founder of Comixify, a technology startup focused on using machine learning algorithms for video editing.

Location and schedule:

Wednesday, April 6th in 5440
14:15 - 15:15 lecture
15:30 - 17:00 class
Thursday, April 7th in 5440
14:15 - 15:15 lecture
15:30 - 17:00 class
Friday, April 8th in 5440
14:15 - 15:15 lecture
15:30 - 17:00 class

Here are all slides.
Videos: The video recordings of the lectures are available:
Assignment: link to the assignment. Please submit the solutions to wojciech.masarczyk.dokt (at) by 08.05.22 .