AI Seminar Series 2022: Decebal Constantin Mocanu and Elena Mocanu

The AI Seminar is a weekly meeting at the University of Alberta where researchers interested in artificial intelligence (AI) can share their research. Presenters include both local speakers from the University of Alberta and visitors from other institutions. Topics can be related in any way to artificial intelligence, from foundational theoretical work to innovative applications of AI techniques to new fields and problems.

On July 15, Decebal Constantin Mocanu and Elena Mocanu, both with the University of Twente, presented "Sparse training in supervised, unsupervised, and deep reinforcement learning" at the AI Seminar.

In the first part of the seminar, Decebal starts with a quick overview of his research line. He introduces along this line one of the many challenges which prevent it from having truly scalable artificial neural networks. Next, he presents an emerging state-of-the-art possible solution: sparse-to-sparse training with static and dynamic sparsity.

In the second part, Elena explores how the reinforcement learning paradigm has a high potential for autonomous agents, although it suffers from scalability issues. She introduces dynamic sparse training in deep reinforcement learning and paves the ground for scalable deep reinforcement learning. Finally, she describes some very recent progress in the field that could be used to foster the generalization performance of sparsely trained RL agents over their densely trained counterparts — while at the same time considerably reducing their computational and memory requirements in both training and inference.

Keep up-to-date on the AI Seminar Series by signing up for the mailing list.

Learn how Amii advances world-leading artificial intelligence and machine learning research: visit our Research page.