Meta-learnt priors slow down catastrophic forgetting in neural networks

Abstract

Current training regimes for deep learning usually involve exposure to a single task / dataset at a time. Here we start from the observation that in this context the trained model is not given any knowledge of anything outside its (single-task) training distribution, and has thus no way to learn parameters (i.e., feature detectors or policies) that could be helpful to solve other tasks, and to limit future interference with the acquired knowledge, and thus catastrophic forgetting. Here we show that catastrophic forgetting can be mitigated in a meta-learning context, by exposing a neural network to multiple tasks in a sequential manner during training. Finally, we present SeqFOMAML, a meta-learning algorithm that implements these principles, and we evaluate it on sequential learning problems composed by Omniglot and MiniImageNet classification tasks.

References in corpus (13)

• TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
• Overcoming catastrophic forgetting in neural networks
• Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
• Matching Networks for One Shot Learning
• An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networks
• On First-Order Meta-Learning Algorithms
• Re-evaluating Continual Learning Scenarios: A Categorization and Case for Strong Baselines
• Born to Learn: the Inspiration, Progress, and Future of Evolved Plastic Artificial Neural Networks
• Task Agnostic Continual Learning via Meta Learning
• Backpropamine: training self-modifying neural networks with differentiable neuromodulated plasticity
• Meta-Learning by the Baldwin Effect
• Meta Continual Learning
• Meta-Learning Representations for Continual Learning

Cited by in corpus (5)

• A Policy Gradient Algorithm for Learning to Learn in Multiagent Reinforcement Learning
• ARCADe: A Rapid Continual Anomaly Detector
• Meta-Learning with Neural Tangent Kernels
• Few-Shot Unsupervised Continual Learning through Meta-Examples
• Reviewing continual learning from the perspective of human-level intelligence