Storing Encoded Episodes as Concepts for Continual Learning
This addresses memory-efficient continual learning for AI systems, offering a novel method with significant performance gains.
The paper tackles catastrophic forgetting and memory limitations in continual learning by encoding images with autoencoders and Neural Style Transfer, then replaying reconstructed or pseudo-images during new task training. It increases classification accuracy by 13-17% over state-of-the-art methods and reduces storage space by 78% on benchmark datasets.
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.