A Feature Filter for EEG Using Cycle-GAN Structure

摘要

The brain-computer interface (BCI) has become one of the most important biomedical research fields and has created many useful applications. As an important component of BCI, electroencephalography (EEG) is in general sensitive to noise and rich in all kinds of information from our brain. In this paper, we introduce a new strategy to filter out unwanted features from EEG signals using GAN-based autoencoders. Filtering out signals relating to one property of the EEG signal while retaining another is similar to the way we can listen to just one voice during a party. This approach has many potential applications including in privacy and security. We use the UCI EEG dataset on alcoholism for our experiments. Our experiment results show that our novel GAN based structure can filter out alcoholism information for 66% of EEG signals with an average of only 6.2% accuracy lost.