Measuring Fine-Grained Relatedness in Multitask Learning via Data Attribution

TL;DR

This paper introduces the MultiTask Influence Function (MTIF), a novel data attribution method that measures fine-grained task relatedness in Multitask Learning, enabling better data selection and reducing negative transfer.

Contribution

The work extends influence functions to MTL, providing a fine-grained, instance-level relatedness measure and a data selection strategy to improve MTL performance.

Findings

MTIF accurately approximates model performance on data subsets.

Data selection via MTIF consistently enhances MTL model performance.

The method offers an efficient way to measure task relatedness at the instance level.

Abstract

Measuring task relatedness and mitigating negative transfer remain a critical open challenge in Multitask Learning (MTL). This work extends data attribution -- which quantifies the influence of individual training data points on model predictions -- to MTL setting for measuring task relatedness. We propose the MultiTask Influence Function (MTIF), a method that adapts influence functions to MTL models with hard or soft parameter sharing. Compared to conventional task relatedness measurements, MTIF provides a fine-grained, instance-level relatedness measure beyond the entire-task level. This fine-grained relatedness measure enables a data selection strategy to effectively mitigate negative transfer in MTL. Through extensive experiments, we demonstrate that the proposed MTIF efficiently and accurately approximates the performance of models trained on data subsets. Moreover, the data selection strategy enabled by MTIF consistently improves model performance in MTL. Our work establishes a novel connection between data attribution and MTL, offering an efficient and fine-grained solution for measuring task relatedness and enhancing MTL models.