Parameter-Level Soft-Masking for Continual Learning
Jul 24, 2023
Speakers
Tatsuya Konishi
Speaker · 0 followers
Mori Kurokawa
Speaker · 0 followers
Chihiro Ono
Speaker · 0 followers
Gyuhak Kim
Speaker · 0 followers
About
Existing research on task incremental learning in continual learning has primarily focused on preventing catastrophic forgetting (CF). Although several techniques have achieved learning with no CF, they attain it by letting each task monopolize a sub-network in a shared network, which seriously limits knowledge transfer (KT) and causes over-consumption of the network capacity, i.e., as more tasks are learned, the performance deteriorates. The goal of this paper is threefold: (1) overcoming CF, (2) encouraging KT, and (3) tackling the capacity problem. A novel technique (called SPG) is proposed that soft-masks (partially blocks) parameter updating in training based on the importance of each parameter to old tasks. Each task still uses the full network, i.e., no monopoly of any part of the network by any task, which enables maximum KT and reduction in capacity usage. To our knowledge, this is the first work that soft-masks a model at the parameter-level for continual learning. Extensive experiments demonstrate the effectiveness of SPG in achieving all three objectives. More notably, it attains significant transfer of knowledge not only among similar tasks (with shared knowledge) but also among dissimilar tasks (with little shared knowledge) while mitigating CF.Existing research on task incremental learning in continual learning has primarily focused on preventing catastrophic forgetting (CF). Although several techniques have achieved learning with no CF, they attain it by letting each task monopolize a sub-network in a shared network, which seriously limits knowledge transfer (KT) and causes over-consumption of the network capacity, i.e., as more tasks are learned, the performance deteriorates. The goal of this paper is threefold: (1) overcoming CF, (…
Like the format? Trust SlidesLive to capture your next event!
Professional recording and live streaming, delivered globally.
Sharing
Recommended Videos
Presentations on similar topic, category or speaker
Leveraging Offline Data in Online Reinforcement Learning
Watch later
Optimizing NOTEARS Objectives via Topological Swaps
Watch later
Chang Deng, …
Faith-Shap: The Faithful Shapley Interaction Index
Watch later
Understanding Backdoor Attacks through the Adaptability Hypothesis
Watch later
Xun Xian, …
Language Models are Bounded Pragmatic Speakers
Watch later
