Geometric Prior-Guided Federated Prompt Calibration

TL;DR

This paper introduces GGTPC, a novel federated prompt calibration method that uses a global geometric prior to correct local training bias caused by data heterogeneity, significantly improving federated learning performance.

Contribution

It proposes a new framework that directly addresses local training bias in federated prompt learning by leveraging a privacy-preserving global geometric prior and a calibration layer.

Findings

Outperforms state-of-the-art on CIFAR-100 with label skew.

Improves baseline performance under extreme data skew.

Enhances FedAvg on domain-skewed datasets.

Abstract

Federated Prompt Learning (FPL) offers a parameter-efficient solution for collaboratively training large models, but its performance is severely hindered by data heterogeneity, which causes locally trained prompts to become biased. Existing methods, focusing on aggregation or regularization, fail to address this root cause of local training bias. To this end, we propose Geometry-Guided Text Prompt Calibration (GGTPC), a novel framework that directly corrects this bias by providing clients with a global geometric prior. This prior, representing the shape of the global data distribution derived from the covariance matrix, is reconstructed on the server in a privacy-preserving manner. Clients then use a novel Geometry-Prior Calibration Layer (GPCL) to align their local feature distributions with this global prior during training. Extensive experiments show GGTPC's effectiveness. On the label-skewed CIFAR-100 dataset ($\beta$=0.1), it outperforms the state-of-the-art by 2.15\%. Under extreme skew ($\beta$=0.01), it improves upon the baseline by 9.17\%. Furthermore, as a plug-and-play module on the domain-skewed Office-Home dataset, it boosts FedAvg's performance by 4.60\%. These results demonstrate that GGTPC effectively mitigates data heterogeneity by correcting the fundamental local training bias, serving as a versatile module to enhance various FL algorithms.