Decoding of Visual Information Based on EEG Data

Abstract

We perhaps fantasized about reading the mind of the human brain directly and use it in other ways. In this paper, we tried to analyze human brain activity to identify the feature space employed by humans for visual classification. The brain signals driven by visual information were classified, and the visual information classifier were trained, and the visual information were decoded by the brain signals. The 64 channels EEG data induced by 40 types of images were collected, and recurrent neural network(RNN) was used for extracting features of electroencephalogram(EEG) signals and predicting the types of EEG signals. The accuracy of decoding could reach 80.90%. The technology of decoding of visual information based on brain EEG data could be applied to the brain-computer interface(BCI).