Work

Improved Robustness for Brain Activity Decoding Based on Information Theoretic Learning

Year: 2023
Type: dissertation
Abstract: Brain activity decoding aims to predict the intentions or mental states of the brain by the utilization of brain recording signals.One significant obstacle for brain activity decoding is the brain recording noise that has complicated distribution, which may lead to large performance deterioration for existing brain decoding algorithms.To address this problem, the present thesis aims to propose robust brain activity decoding algorithms under the framework of information theoretic learning to alleviate the negative effects of the adverse brain recording noise.In particular, minimum error entropy criterion and maximum correntropy criterion were utilized to build robust objective functions for brain decoding algorithms.In addition, another significant problem for brain activity decoding, the high-dimensional issue, is also taken into account in this thesis.The proposed algorithms were evaluated systematically with synthetic datasets and real-world brain data.The experimental results demonstrated that information theoretic learning based robust brain decoding algorithms effectively reduce the performance deterioration caused by the noise and realize higher brain decoding accuracy on the real-world noisy brain data.
Authors: LI, Yuanhao, Li, Yuanhao
Language: English
Open Access status: green
OpenAlex ID: W7047602591
Cites: 136
Cited by:
Related to: 10