La-MAML: Look-ahead Meta Learning for Continual Learning

Abstract

The continual learning problem involves training models with limited capacity to perform well on a set of an unknown number of sequentially arriving tasks. While meta-learning shows great potential for reducing interference between old and new tasks, the current training procedures tend to be either slow or offline, and sensitive to many hyper-parameters. In this work, we propose Look-ahead MAML (La-MAML), a fast optimisation-based meta-learning algorithm for online-continual learning, aided by a small episodic memory. Our proposed modulation of per-parameter learning rates in our meta-learning update allows us to draw connections to prior work on hypergradients and meta-descent. This provides a more flexible and efficient way to mitigate catastrophic forgetting compared to conventional prior-based methods. La-MAML achieves performance superior to other replay-based, prior-based and meta-learning based approaches for continual learning on real-world visual classification benchmarks. Source code can be found here: https://github.com/montrealrobotics/La-MAML

arXiv commentAccepted for Oral Presentation at NeurIPS 2020

References in corpus (14)

• Overcoming catastrophic forgetting in neural networks
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Theoretical Models of Learning to Learn
• Meta-SGD: Learning to Learn Quickly for Few-Shot Learning
• Efficient Lifelong Learning with A-GEM
• On First-Order Meta-Learning Algorithms
• On Tiny Episodic Memories in Continual Learning
• Empirical Analysis of the Hessian of Over-Parametrized Neural Networks
• Gradient Surgery for Multi-Task Learning
• Online Meta-Learning
• Selfless Sequential Learning
• Task Agnostic Continual Learning Using Online Variational Bayes
• Alpha MAML: Adaptive Model-Agnostic Meta-Learning
• Deep Online Learning via Meta-Learning: Continual Adaptation for Model-Based RL

Cited by in corpus (11)

• Flattening Sharpness for Dynamic Gradient Projection Memory Benefits Continual Learning
• Complementary Calibration: Boosting General Continual Learning with Collaborative Distillation and Self-Supervision
• Online Continual Learning via the Knowledge Invariant and Spread-out Properties
• A Policy Gradient Algorithm for Learning to Learn in Multiagent Reinforcement Learning
• Formalizing the Generalization-Forgetting Trade-off in Continual Learning
• Sequoia: A Software Framework to Unify Continual Learning Research
• Reviewing continual learning from the perspective of human-level intelligence
• Meta-Learning for Domain Generalization in Semantic Parsing
• Reproducibility Report: La-MAML: Look-ahead Meta Learning for Continual Learning
• TAG: Task-based Accumulated Gradients for Lifelong learning
• Online Continual Learning in Image Classification: An Empirical Survey