A list of papers and studies about online learning and less forgettable neural network models.
○ Few-Shot Class-Incremental Learning (CVPR 2020) [Paper](https://arxiv.org/pdf/2004.10956.pdf]
○ Mnemonics Training: Multi-Class Incremental Learning without Forgetting (CVPR 2020) Paper
○ Continual learning with hypernetworks (ICLR 2020) Paper
○ Continual Learning with Adaptive Weights (ICLR 2020) Paper
○ Just Scan - Scalable and Order-robust Continual Learning with Additive Parameter Decomposition (ICLR 2020) Paper
○ Compacting, Picking and Growing for Unforgetting Continual Learning (NIPS 2019) Paper
○ IL2M: Class incremental Learning With Dual Memory (ICCV 2019) Paper
○ Online Continual Learning with Maximally Interfered Retrieval (NIPS 2019) Paper
○ Learning without Memorizing (CVPR2019) Paper
○ Large Scale Incremental Learning (CVPR2019) Paper
○ Learning to Learn without Forgetting By Maximizing Transfer and Minimizing Interference (ICLR2019) Paper
○ Overcoming Catastrophic Forgetting via Model Adaptation (ICLR2019)
○ Overcoming Catastrophic Forgetting with Hard Attention to the Task (ICML 2018) Paper
○ Generative replay with feedback connections as a general strategy for continual learning Paper
- Shows that EWC fails when task identity needed to be inferred.
- Replay-through-Feedback approach substantially shortened training time with no or negligible loss in performance.