Gradient Correlation Subspace Learning against Catastrophic Forgetting
This addresses the problem of performance degradation in continual learning for AI systems, but appears incremental as it builds on existing subspace-based methods.
The paper tackles catastrophic forgetting in incremental class learning by introducing Gradient Correlation Subspace Learning (GCSL), which projects weights into a subspace least affected by previous tasks, but no concrete performance numbers are provided in the abstract.
Efficient continual learning techniques have been a topic of significant research over the last few years. A fundamental problem with such learning is severe degradation of performance on previously learned tasks, known also as catastrophic forgetting. This paper introduces a novel method to reduce catastrophic forgetting in the context of incremental class learning called Gradient Correlation Subspace Learning (GCSL). The method detects a subspace of the weights that is least affected by previous tasks and projects the weights to train for the new task into said subspace. The method can be applied to one or more layers of a given network architectures and the size of the subspace used can be altered from layer to layer and task to task. Code will be available at \href{https://github.com/vgthengane/GCSL}{https://github.com/vgthengane/GCSL}