Task Uncertainty Loss Reduce Negative Transfer in Asymmetric Multi-task Feature Learning
This work is significant for researchers and practitioners using multi-task learning, as it provides a method to mitigate negative transfer, a common issue hindering the reliability of MTL models.
This paper addresses negative transfer in multi-task learning, where individual tasks perform worse than in single-task learning. The authors propose using aleatoric homoscedastic uncertainty to weigh task losses, demonstrating a reduction in negative transfer across image recognition and pharmacogenomics datasets.
Multi-task learning (MTL) is frequently used in settings where a target task has to be learnt based on limited training data, but knowledge can be leveraged from related auxiliary tasks. While MTL can improve task performance overall relative to single-task learning (STL), these improvements can hide negative transfer (NT), where STL may deliver better performance for many individual tasks. Asymmetric multitask feature learning (AMTFL) is an approach that tries to address this by allowing tasks with higher loss values to have smaller influence on feature representations for learning other tasks. Task loss values do not necessarily indicate reliability of models for a specific task. We present examples of NT in two orthogonal datasets (image recognition and pharmacogenomics) and tackle this challenge by using aleatoric homoscedastic uncertainty to capture the relative confidence between tasks, and set weights for task loss. Our results show that this approach reduces NT providing a new approach to enable robust MTL.