MmAP : Multi-modal Alignment Prompt for Cross-domain Multi-task Learning

TL;DR

This paper introduces MmAP, a multi-modal alignment prompt for CLIP that enhances multi-task learning by aligning text and visual modalities, enabling efficient parameter use and improved performance across tasks.

Contribution

The paper proposes a novel multi-modal alignment prompt (MmAP) for CLIP, enabling effective multi-task learning with minimal trainable parameters and preserving task-specific features.

Findings

Achieves significant performance gains over full fine-tuning.

Uses only approximately 0.09% of trainable parameters.

Demonstrates effectiveness on large multi-task datasets.

Abstract

Multi-Task Learning (MTL) is designed to train multiple correlated tasks simultaneously, thereby enhancing the performance of individual tasks. Typically, a multi-task network structure consists of a shared backbone and task-specific decoders. However, the complexity of the decoders increases with the number of tasks. To tackle this challenge, we integrate the decoder-free vision-language model CLIP, which exhibits robust zero-shot generalization capability. Recently, parameter-efficient transfer learning methods have been extensively explored with CLIP for adapting to downstream tasks, where prompt tuning showcases strong potential. Nevertheless, these methods solely fine-tune a single modality (text or visual), disrupting the modality structure of CLIP. In this paper, we first propose Multi-modal Alignment Prompt (MmAP) for CLIP, which aligns text and visual modalities during fine-tuning process. Building upon MmAP, we develop an innovative multi-task prompt learning framework. On the one hand, to maximize the complementarity of tasks with high similarity, we utilize a gradient-driven task grouping method that partitions tasks into several disjoint groups and assign a group-shared MmAP to each group. On the other hand, to preserve the unique characteristics of each task, we assign an task-specific MmAP to each task. Comprehensive experiments on two large multi-task learning datasets demonstrate that our method achieves significant performance improvements compared to full fine-tuning while only utilizing approximately 0.09% of trainable parameters.