Energy-Based Models for Continual Learning

Energy-Based Models for Continual Learning

Energy-Based Models for Continual Learning --> Energy-Based Models for Continual Learning ICML 2021 --> Shuang Li 1 , Yilun Du 1 , Gido M. van de Ven 2 , Igor Mordatch 3 1 MIT CSAIL 2 Baylor College of Medicine 3 Google Brain Paper Code We motivate Energy-Based Models (EBMs) as a promising model class for continual learning problems. Instead of tackling continual learning via the use of external memory, growing models, or regularization, EBMs change the underlying training objective to causes less interference with previously learned information. Our proposed version of EBMs for continual lear

related reading

• Energy-Based Modelsenergy-based-model.github.io
• The Continual Learning Problemjessylin.com
• [2302.00487] A Comprehensive Survey of Continual Learning: Theory, Method and Applicationarxiv.org
• [2512.23675] End-to-End Test-Time Training for Long Contextarxiv.org
• [2111.00396] Efficiently Modeling Long Sequences with Structured State Spacesarxiv.org
• [1606.09282] Learning without Forgettingarxiv.org
• Self-Distillation Enables Continual Learningarxiv.org
• d89a66c7c80a29b1bdbab0f2a1a94af8-Paper.pdfproceedings.neurips.cc
• [2605.15220] Always Learning, Always Mixing: Efficient and Simple Data Mixing All The Timearxiv.org
• Continual Learning Bench 1.0 — Newscontinual-learning-bench.com
• Contrastive Learning as Goal-Conditioned Reinforcement Learningarxiv.org
• 2006.10742arxiv.org