Continual Learning via Sequential Function-Space Variational Inference

TL;DR

This paper introduces a novel continual learning method using sequential function-space variational inference, enabling neural networks to adapt better to new tasks with improved accuracy and less reliance on previous task data.

Contribution

It formulates continual learning as sequential function-space variational inference, allowing more flexible adaptation and improved performance over existing parameter-regularization methods.

Findings

Achieves better predictive accuracy across various task sequences.

Requires less dependence on representative points from previous tasks.

Enables neural networks to adapt more flexibly to new data.

Abstract

Sequential Bayesian inference over predictive functions is a natural framework for continual learning from streams of data. However, applying it to neural networks has proved challenging in practice. Addressing the drawbacks of existing techniques, we propose an optimization objective derived by formulating continual learning as sequential function-space variational inference. In contrast to existing methods that regularize neural network parameters directly, this objective allows parameters to vary widely during training, enabling better adaptation to new tasks. Compared to objectives that directly regularize neural network predictions, the proposed objective allows for more flexible variational distributions and more effective regularization. We demonstrate that, across a range of task sequences, neural networks trained via sequential function-space variational inference achieve better predictive accuracy than networks trained with related methods while depending less on maintaining a set of representative points from previous tasks.