Meta-Continual Learning Via Dynamic Programming

Krishnan Raghavan, Prasanna Balaprakash

Keywords: Meta Continual Learning, Supervised Learning, Dynamic Programming, Catastrophic Forgetting, Generalization

Abstract: Meta continual learning algorithms seek to train a model when faced with similar tasks observed in a sequential manner. Despite promising methodological advancements, there is a lack of theoretical frameworks that enable analysis of learning challenges such as generalization and catastrophic forgetting. To that end, we develop a new theoretical approach for meta continual learning~(MCL) where we mathematically model the learning dynamics using dynamic programming, and we establish conditions of optimality for the MCL problem. Moreover, using the theoretical framework, we derive a new dynamic-programming-based MCL method that adopts stochastic-gradient-driven alternating optimization to balance generalization and catastrophic forgetting. We show that, on MCL benchmark data sets, our theoretically grounded method achieves accuracy better than or comparable to that of existing state-of-the-art methods.

One-sentence Summary: A dynamic-programming-based meta-continual approach that adopts a stochastic-gradient-driven alternating optimization to balance generalization and catastrophic forgetting.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Reviewed Version (pdf): https://openreview.net/references/pdf?id=x4h3r5fCcU