Skip to content

chrhenning/posterior_replay_cl

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Posterior Meta-Replay for Continual Learning

In this study we propose posterior-replay continual learning, a framework for continually learning task-specific posterior approximations within a single shared meta-model. Across a range of experiments, we compare this approach to prior-focused CL, for which a single trade-off solution across all tasks is recursively obtained. Please see our paper for more details. You can also checkout our poster presentation or this longer talk in which we present the project.

If you are interested in working with hypernetworks in PyTorch, check out the package hypnettorch.

1D Regression Experiments

You can find instructions on how to reproduce our 1D Regression experiments and on how to use the corresponding code in the subfolder probabilistic/regression.

2D Mode Classification Experiments

You can find instructions on how to reproduce our 2D Mode Classification experiments and on how to use the corresponding code in the subfolder probabilistic/prob_gmm.

Split and Permuted MNIST Experiments

You can find instructions on how to reproduce our Split and Permuted MNIST experiments and on how to use the corresponding code in the subfolder probabilistic/prob_mnist.

SplitCIFAR-10/100 Experiments

You can find instructions on how to reproduce our SplitCIFAR-10 and SplitCIFAR-100 experiments and on how to use the corresponding code in the subfolder probabilistic/prob_cifar.

Documentation

Please refer to the README in the subfolder docs for instructions on how to compile and open the documentation.

Citation

Please cite our corresponding paper if you use this code in your research project.

@inproceedings{posterior:replay:2021:henning:cervera,
title={Posterior Meta-Replay for Continual Learning}, 
      author={Christian Henning and Maria R. Cervera and Francesco D'Angelo and Johannes von Oswald and Regina Traber and Benjamin Ehret and Seijin Kobayashi and Benjamin F. Grewe and João Sacramento},
booktitle={Conference on Neural Information Processing Systems},
year={2021},
url={https://arxiv.org/abs/2103.01133}
}

About

Continual learning of task-specific approximations of the parameter posterior distribution via a shared hypernetwork.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published