RBLA: Rank-Based-LoRA-Aggregation for Fine-tuning Heterogeneous Models in FLaaS

TL;DR

This paper introduces RBLA, a novel aggregation method for federated learning that effectively combines models with different LoRA ranks, improving performance in heterogeneous FLaaS environments.

Contribution

The paper proposes RBLA, a new rank-based aggregation technique that preserves diverse model features, addressing limitations of current padding methods in heterogeneous LoRA federated learning.

Findings

RBLA outperforms existing aggregation methods in heterogeneous settings.

RBLA maintains model accuracy across diverse hardware capabilities.

Experimental results validate RBLA's effectiveness in FLaaS environments.

Abstract

Federated Learning (FL) is a promising privacy-aware distributed learning framework that can be deployed on various devices, such as mobile phones, desktops, and devices equipped with CPUs or GPUs. In the context of server-based Federated Learning as a Service (FLaaS), FL enables a central server to coordinate the training process across multiple devices without direct access to local data, thereby enhancing privacy and data security. Low-Rank Adaptation (LoRA) is a method that efficiently fine-tunes models by focusing on a low-dimensional subspace of the model's parameters. This approach significantly reduces computational and memory costs compared to fine-tuning all parameters from scratch. When integrated with FL, particularly in a FLaaS environment, LoRA allows for flexible and efficient deployment across diverse hardware with varying computational capabilities by adjusting the local model's rank. However, in LoRA-enabled FL, different clients may train models with varying ranks, which poses challenges for model aggregation on the server. Current methods for aggregating models of different ranks involve padding weights to a uniform shape, which can degrade the global model's performance. To address this issue, we propose Rank-Based LoRA Aggregation (RBLA), a novel model aggregation method designed for heterogeneous LoRA structures. RBLA preserves key features across models with different ranks. This paper analyzes the issues with current padding methods used to reshape models for aggregation in a FLaaS environment. Then, we introduce RBLA, a rank-based aggregation method that maintains both low-rank and high-rank features. Finally, we demonstrate the effectiveness of RBLA through comparative experiments with state-of-the-art methods.