# Storing Encoded Episodes as Concepts for Continual Learning

arXiv: https://arxiv.org/abs/2007.06637 · HTML: https://mlsanity.com/paper/2007.06637/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 26 Jun 2020.

## Built on by (most cited first, 8 of 8)

- [EEC: Learning to Encode and Regenerate Images for Continual Learning](https://mlsanity.com/paper/2101.04904/) (2021)
- [F-SIOL-310: A Robotic Dataset and Benchmark for Few-Shot Incremental Object Learning](https://mlsanity.com/paper/2103.12242/) (2021)
- [Continual Learning Approaches for Anomaly Detection](https://mlsanity.com/paper/2212.11192/) (2022)
- [Continual Learning via Manifold Expansion Replay](https://mlsanity.com/paper/2310.08038/) (2023)
- [Learning Novel Objects Continually Through Curiosity](https://mlsanity.com/paper/2103.07758/) (2021)
- [CBCL-PR: A Cognitively Inspired Model for Class-Incremental Learning in Robotics](https://mlsanity.com/paper/2308.00199/) (2023)
- [RECALL: Rehearsal-free Continual Learning for Object Classification](https://mlsanity.com/paper/2209.14774/) (2022)
- [Continual Learning of Visual Concepts for Robots through Limited Supervision](https://mlsanity.com/paper/2101.10509/) (2021)

## Abstract (arXiv)

The two main challenges faced by continual learning approaches are catastrophic forgetting and memory limitations on the storage of data. To cope with these challenges, we propose a novel, cognitively-inspired approach which trains autoencoders with Neural Style Transfer to encode and store images. Reconstructed images from encoded episodes are replayed when training the classifier model on a new task to avoid catastrophic forgetting. The loss function for the reconstructed images is weighted to reduce its effect during classifier training to cope with image degradation. When the system runs out of memory the encoded episodes are converted into centroids and covariance matrices, which are used to generate pseudo-images during classifier training, keeping classifier performance stable with less memory. Our approach increases classification accuracy by 13-17% over state-of-the-art methods on benchmark datasets, while requiring 78% less storage space.