Prioritizing Modalities: Flexible Importance Scheduling in Federated Multimodal Learning

TL;DR

This paper introduces FlexMod, an adaptive resource allocation method for multimodal federated learning that prioritizes important modalities to improve efficiency and model performance.

Contribution

We propose a novel importance scheduling approach using prototype learning, Shapley values, and reinforcement learning to optimize resource allocation in MFL.

Findings

Significant performance improvements on three real-world datasets.

Efficient resource utilization by prioritizing critical modalities.

Enhanced model accuracy through adaptive importance scheduling.

Abstract

Federated Learning (FL) is a distributed machine learning approach that enables devices to collaboratively train models without sharing their local data, ensuring user privacy and scalability. However, applying FL to real-world data presents challenges, particularly as most existing FL research focuses on unimodal data. Multimodal Federated Learning (MFL) has emerged to address these challenges, leveraging modality-specific encoder models to process diverse datasets. Current MFL methods often uniformly allocate computational frequencies across all modalities, which is inefficient for IoT devices with limited resources. In this paper, we propose FlexMod, a novel approach to enhance computational efficiency in MFL by adaptively allocating training resources for each modality encoder based on their importance and training requirements. We employ prototype learning to assess the quality of modality encoders, use Shapley values to quantify the importance of each modality, and adopt the Deep Deterministic Policy Gradient (DDPG) method from deep reinforcement learning to optimize the allocation of training resources. Our method prioritizes critical modalities, optimizing model performance and resource utilization. Experimental results on three real-world datasets demonstrate that our proposed method significantly improves the performance of MFL models.