Efficiently Identifying Task Groupings for Multi-Task Learning

TL;DR

This paper introduces a method to efficiently identify optimal task groupings in multi-task learning, improving performance and reducing computational costs by analyzing task gradient effects in a single training run.

Contribution

The proposed approach determines task groupings in one run by quantifying gradient effects, outperforming existing methods in efficiency and accuracy.

Findings

Decreased test loss by 10.0% on Taskonomy dataset.

Operates 11.6 times faster than previous state-of-the-art methods.

Effective in large-scale multi-task learning scenarios.

Abstract

Multi-task learning can leverage information learned by one task to benefit the training of other tasks. Despite this capacity, naively training all tasks together in one model often degrades performance, and exhaustively searching through combinations of task groupings can be prohibitively expensive. As a result, efficiently identifying the tasks that would benefit from training together remains a challenging design question without a clear solution. In this paper, we suggest an approach to select which tasks should train together in multi-task learning models. Our method determines task groupings in a single run by training all tasks together and quantifying the effect to which one task's gradient would affect another task's loss. On the large-scale Taskonomy computer vision dataset, we find this method can decrease test loss by 10.0% compared to simply training all tasks together while operating 11.6 times faster than a state-of-the-art task grouping method.