In Defense of the Unitary Scalarization for Deep Multi-Task Learning

TL;DR

This paper demonstrates that simple unitary scalarization combined with standard regularization techniques can match or outperform complex multi-task optimization algorithms, challenging recent skepticism about its effectiveness.

Contribution

The study shows that basic scalarization with regularization rivals complex multi-task optimizers, prompting a reevaluation of current multi-task learning strategies.

Findings

Unitary scalarization performs as well or better than complex optimizers.

Regularization techniques are key to effective multi-task learning.

Many advanced optimizers can be viewed as forms of regularization.

Abstract

Recent multi-task learning research argues against unitary scalarization, where training simply minimizes the sum of the task losses. Several ad-hoc multi-task optimization algorithms have instead been proposed, inspired by various hypotheses about what makes multi-task settings difficult. The majority of these optimizers require per-task gradients, and introduce significant memory, runtime, and implementation overhead. We show that unitary scalarization, coupled with standard regularization and stabilization techniques from single-task learning, matches or improves upon the performance of complex multi-task optimizers in popular supervised and reinforcement learning settings. We then present an analysis suggesting that many specialized multi-task optimizers can be partly interpreted as forms of regularization, potentially explaining our surprising results. We believe our results call for a critical reevaluation of recent research in the area.