Adjusting Model Size in Continual Gaussian Processes: How Big is Big Enough?

TL;DR

This paper introduces a method for automatically adjusting the size of Gaussian process models in continual learning, ensuring near-optimal performance without extensive hyperparameter tuning across diverse datasets.

Contribution

We propose a novel approach to dynamically determine model size in continual Gaussian processes, reducing the need for dataset-specific hyperparameter tuning.

Findings

Method maintains high performance across datasets

Requires less hyperparameter tuning than existing approaches

Automatically adapts model size during training

Abstract

Many machine learning models require setting a parameter that controls their size before training, e.g. number of neurons in DNNs, or inducing points in GPs. Increasing capacity typically improves performance until all the information from the dataset is captured. After this point, computational cost keeps increasing, without improved performance. This leads to the question "How big is big enough?" We investigate this problem for Gaussian processes (single-layer neural networks) in continual learning. Here, data becomes available incrementally, and the final dataset size will therefore not be known before training, preventing the use of heuristics for setting a fixed model size. We develop a method to automatically adjust model size while maintaining near-optimal performance. Our experimental procedure follows the constraint that any hyperparameters must be set without seeing dataset properties, and we show that our method performs well across diverse datasets without the need to adjust its hyperparameter, showing it requires less tuning than others.