Connection Strength-Based Optimization for Multi-Task Learning
Keywords: multi-task learning, optimization, task priority, connection strength
TL;DR: We propose a novel multi-task optimization which takes into account task priority within the shared network.
Abstract: The goal of multi-task learning is to learn diverse tasks within a single unified network. As each task has its own unique objective function, conflicts emerge during training, resulting in negative transfer among them. Earlier research identified these conflicting gradients in shared parameters between tasks and attempted to realign them in the same direction. However, we prove that such optimization strategies lead to sub-optimal Pareto solutions due to their inability to accurately determine the individual contributions of each parameter across various tasks. In this paper, we propose the concept of task priority to evaluate parameter contributions across different tasks. We identify two types of connections to learn and maintain task priority: implicit and explicit connections. Implicit connections relate to the links between parameters influenced by task-specific loss during backpropagation, whereas explicit connections are gauged by the magnitude of parameters. Based on these, we present a new method named connection strength-based optimization for multi-task learning. Our optimization process consists of two phases. The first phase learns the task priority within the network, while the second phase modifies the gradients while upholding this priority. This ultimately leads to finding new Pareto optimal solutions for multiple tasks. Through extensive experiments with different loss scaling techniques, we show that our approach greatly enhances multi-task performance in comparison to earlier gradient manipulation methods.
Primary Area: optimization
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 1628
Loading