Continual learning baselines and strategies from popular papers, using Avalanche. We include EWC, SI, GEM, AGEM, LwF, iCarl, GDumb, and other strategies.
DEPRECATED & OBSOLETE! Previously StackStorm Enterprise (EWC) Workflow Editor. Now integrated directly into StackStorm OSS Core platform (st2web).
Keras-based framework for implementing continual learning methods.
Implementation of ews weight constraint mentioned in recent Deep Mind paper: http://www.pnas.org/content/early/2017/03/13/1611835114.full.pdf
comparative evaluation of incremental machine learning methods
Tensorflow 1.x implementation of EWC, evaluated on permuted MNIST
This respository hosts the Trust List for the EWC Large Scale Pilots and is co-founded by EU Commission
A spaCy library for Named Entity Recognition with Elastic Weight Consolidation.
Experimental study of causal and Bayesian approaches to continual learning
A short script to search for optimal values of lambda in the sequential learning technique Elastic Weight Consolidation
Continual Reinforcement Learning for Adaptive Ocean Thermal Energy Conversion Control using PPO and Elastic Weight Consolidation
A continual learning thesis project evaluating pre-trained ViT-Tiny with adapters and EWC for reducing catastrophic forgetting in sequential image classification.
Federated continual learning system for planetary climate forecasting. Combines FedAvg, Elastic Weight Consolidation, Physics-Informed Neural Networks, and Multi-Agent PPO across 3 geographic nodes. 29 tests · CI/CD · PyTorch
StackStorm pack containing demo workflows and automations
aria-trl v2 (archived snapshot): 3-way comparison including a real EWC baseline, validated on 8 seeds. Ties EWC on accuracy, cuts BWT forgetting further. Frozen at this state; see the main ARIA-TRL repo for ongoing development.
A continual learning architecture that restructures itself while it learns. Beats EWC on Split MNIST; honest results on Split CIFAR 10 too.
A comparative evaluation of continual learning strategies to mitigate catastrophic forgetting in neural networks.
To associate your repository with the ewc topic, visit your repo's landing page and select "manage topics."