Gradient Projection Memory for Continual Learning

Abstract

The ability to learn continually without forgetting the past tasks is a desired attribute for artificial learning systems. Existing approaches to enable such learning in artificial neural networks usually rely on network growth, importance based weight update or replay of old data from the memory. In contrast, we propose a novel approach where a neural network learns new tasks by taking gradient steps in the orthogonal direction to the gradient subspaces deemed important for the past tasks. We find the bases of these subspaces by analyzing network representations (activations) after learning each task with Singular Value Decomposition (SVD) in a single shot manner and store them in the memory as Gradient Projection Memory (GPM). With qualitative and quantitative analyses, we show that such orthogonal gradient descent induces minimum to no interference with the past tasks, thereby mitigates forgetting. We evaluate our algorithm on diverse image classification datasets with short and long sequences of tasks and report better or on-par performance compared to the state-of-the-art approaches.

Accepted for Oral Presentation at ICLR 2021 https://openreview.net/forum?id=3AOj0RCNC2

References in corpus (8)

• Overcoming catastrophic forgetting in neural networks
• Continual Learning Through Synaptic Intelligence
• Efficient Lifelong Learning with A-GEM
• Re-evaluating Continual Learning Scenarios: A Categorization and Case for Strong Baselines
• Continual Learning with Deep Generative Replay
• Learn to Grow: A Continual Structure Learning Framework for Overcoming Catastrophic Forgetting
• A Low Effort Approach to Structured CNN Design Using PCA
• Generalisation Guarantees for Continual Learning with Orthogonal Gradient Descent

Cited by in corpus (5)

• Online Continual Learning on Class Incremental Blurry Task Configuration with Anytime Inference
• Task-agnostic Continual Learning with Hybrid Probabilistic Models
• OER: Offline Experience Replay for Continual Offline Reinforcement Learning
• Optimizing Reusable Knowledge for Continual Learning via Metalearning
• Prototypical quadruplet for few-shot class incremental learning