Source-free Subject Adaptation for EEG-based Visual Recognition

TL;DR

This paper introduces a source-free approach for EEG-based visual recognition that adapts to new subjects using pre-trained models and classifier-based data generation, addressing privacy concerns and improving accuracy.

Contribution

It proposes a novel source-free subject adaptation method using classifier responses to generate EEG samples, enabling effective adaptation without access to source data.

Findings

Achieves 74.6% top-1 accuracy in 5-shot setting on EEG-ImageNet40.

Demonstrates consistent improvements across different subject-independent learning methods.

Shows the method's generalizability and effectiveness without source data access.

Abstract

This paper focuses on subject adaptation for EEG-based visual recognition. It aims at building a visual stimuli recognition system customized for the target subject whose EEG samples are limited, by transferring knowledge from abundant data of source subjects. Existing approaches consider the scenario that samples of source subjects are accessible during training. However, it is often infeasible and problematic to access personal biological data like EEG signals due to privacy issues. In this paper, we introduce a novel and practical problem setup, namely source-free subject adaptation, where the source subject data are unavailable and only the pre-trained model parameters are provided for subject adaptation. To tackle this challenging problem, we propose classifier-based data generation to simulate EEG samples from source subjects using classifier responses. Using the generated samples and target subject data, we perform subject-independent feature learning to exploit the common knowledge shared across different subjects. Notably, our framework is generalizable and can adopt any subject-independent learning method. In the experiments on the EEG-ImageNet40 benchmark, our model brings consistent improvements regardless of the choice of subject-independent learning. Also, our method shows promising performance, recording top-1 test accuracy of 74.6% under the 5-shot setting even without relying on source data. Our code can be found at https://github.com/DeepBCI/Deep-BCI/tree/master/1_Intelligent_BCI/Source_Free_Subject_Adaptation_for_EEG.