Online Continual Learning with Maximally Interfered Retrieval

Abstract

Continual learning, the setting where a learning agent is faced with a never ending stream of data, continues to be a great challenge for modern machine learning systems. In particular the online or "single-pass through the data" setting has gained attention recently as a natural setting that is difficult to tackle. Methods based on replay, either generative or from a stored memory, have been shown to be effective approaches for continual learning, matching or exceeding the state of the art in a number of standard benchmarks. These approaches typically rely on randomly selecting samples from the replay memory or from a generative model, which is suboptimal. In this work, we consider a controlled sampling of memories for replay. We retrieve the samples which are most interfered, i.e. whose prediction will be most negatively impacted by the foreseen parameters update. We show a formulation for this sampling criterion in both the generative replay and the experience replay setting, producing consistent gains in performance and greatly reduced forgetting. We release an implementation of our method at https://github.com/optimass/Maximally_Interfered_Retrieval.

References in corpus (12)

• Overcoming catastrophic forgetting in neural networks
• Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigence
• An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networks
• Efficient Lifelong Learning with A-GEM
• Experience Replay for Continual Learning
• On Tiny Episodic Memories in Continual Learning
• Variational Continual Learning
• Reinforced Continual Learning
• Towards Robust Evaluations of Continual Learning
• Lifelong Generative Modeling
• Gradient based sample selection for online continual learning
• Continual Classification Learning Using Generative Models

Cited by in corpus (7)

• Continual Learning and Catastrophic Forgetting
• Inherit with Distillation and Evolve with Contrast: Exploring Class Incremental Semantic Segmentation Without Exemplar Memory
• Learning latent representations across multiple data domains using Lifelong VAEGAN
• Online Distillation with Continual Learning for Cyclic Domain Shifts
• Block Contextual MDPs for Continual Learning
• Frosting Weights for Better Continual Training
• Toward industrial use of continual learning : new metrics proposal for class incremental learning