STG-MTL: Scalable Task Grouping for Multi-Task Learning Using Data Map

TL;DR

This paper introduces STG-MTL, a scalable, data-driven method for task grouping in multi-task learning that effectively handles large numbers of tasks by using Data Maps to capture training dynamics.

Contribution

The paper presents a novel data-driven approach for scalable task grouping in MTL, utilizing Data Maps to improve performance and handle up to 100 tasks, which was previously unachievable.

Findings

Superior scalability demonstrated compared to existing methods.

Effective task grouping on datasets with up to 100 tasks.

Modular implementation facilitates integration and testing.

Abstract

Multi-Task Learning (MTL) is a powerful technique that has gained popularity due to its performance improvement over traditional Single-Task Learning (STL). However, MTL is often challenging because there is an exponential number of possible task groupings, which can make it difficult to choose the best one because some groupings might produce performance degradation due to negative interference between tasks. That is why existing solutions are severely suffering from scalability issues, limiting any practical application. In our paper, we propose a new data-driven method that addresses these challenges and provides a scalable and modular solution for classification task grouping based on a re-proposed data-driven features, Data Maps, which capture the training dynamics for each classification task during the MTL training. Through a theoretical comparison with other techniques, we manage to show that our approach has the superior scalability. Our experiments show a better performance and verify the method's effectiveness, even on an unprecedented number of tasks (up to 100 tasks on CIFAR100). Being the first to work on such number of tasks, our comparisons on the resulting grouping shows similar grouping to the mentioned in the dataset, CIFAR100. Finally, we provide a modular implementation for easier integration and testing, with examples from multiple datasets and tasks.