End-to-End Multi-Task Learning with Attention

TL;DR

This paper introduces a novel multi-task learning architecture, MTAN, which employs task-specific attention modules within a shared network to improve performance and efficiency across various vision tasks.

Contribution

The paper presents the Multi-Task Attention Network (MTAN), a simple, parameter-efficient, end-to-end trainable architecture that learns task-specific features through attention modules, outperforming existing methods.

Findings

State-of-the-art performance on multiple datasets

Less sensitive to task loss weighting schemes

Effective for both image classification and image-to-image tasks

Abstract

We propose a novel multi-task learning architecture, which allows learning of task-specific feature-level attention. Our design, the Multi-Task Attention Network (MTAN), consists of a single shared network containing a global feature pool, together with a soft-attention module for each task. These modules allow for learning of task-specific features from the global features, whilst simultaneously allowing for features to be shared across different tasks. The architecture can be trained end-to-end and can be built upon any feed-forward neural network, is simple to implement, and is parameter efficient. We evaluate our approach on a variety of datasets, across both image-to-image predictions and image classification tasks. We show that our architecture is state-of-the-art in multi-task learning compared to existing methods, and is also less sensitive to various weighting schemes in the multi-task loss function. Code is available at https://github.com/lorenmt/mtan.