Measuring Fine-Grained Relatedness in Multitask Learning via Data Attribution

2025-05-27

AI Summary

Existing multi-task learning (MTL) approaches suffer from coarse-grained task relatedness estimation and insufficient suppression of negative transfer. To address these limitations, this paper introduces data attribution into MTL for the first time and proposes the Multi-Task Influence Function (MTIF). Grounded in influence function theory, MTIF is compatible with both hard- and soft-parameter-sharing architectures and integrates reverse Hessian approximation with gradient covariance modeling to enable **instance-level, fine-grained quantification of task relatedness**. Leveraging this metric, we design a principled data selection strategy that actively mitigates negative transfer. Extensive experiments across multiple MTL benchmarks demonstrate that MTIF consistently improves primary task accuracy while robustly suppressing negative transfer, thereby validating the effectiveness and practicality of fine-grained relatedness modeling.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Reasoning under Uncertainty: Relational Probabilistic Models
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

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.

Problem

• Measure fine-grained task relatedness in Multitask Learning
• Mitigate negative transfer via data attribution methods
• Propose MultiTask Influence Function for instance-level analysis

Innovation

• Extends data attribution to MTL setting
• Proposes MultiTask Influence Function (MTIF)
• Enables fine-grained data selection strategy