Disentanglement of Color and Shape Representations for Continual Learning
This is an incremental improvement for continual learning systems, addressing forgetting in task-incremental settings.
The paper tackles catastrophic forgetting in continual learning by disentangling color and shape features, showing that this approach improves performance when combined with existing methods like Elastic Weight Consolidation on the Oxford-102 Flowers dataset.
We hypothesize that disentangled feature representations suffer less from catastrophic forgetting. As a case study we perform explicit disentanglement of color and shape, by adjusting the network architecture. We tested classification accuracy and forgetting in a task-incremental setting with Oxford-102 Flowers dataset. We combine our method with Elastic Weight Consolidation, Learning without Forgetting, Synaptic Intelligence and Memory Aware Synapses, and show that feature disentanglement positively impacts continual learning performance.