Generating Visual Stimuli from EEG Recordings using Transformer-encoder based EEG encoder and GAN

The study focuses on generating images from EEG signals using a deep learning framework. The objective is to recreate images that subjects view using EEG recordings. This is achieved using a Transformer-encoder based EEG encoder, which serves as input to the generator component of the GAN network. The study also incorporates perceptual loss to enhance the quality of generated images. This research could improve communication with individuals experiencing conditions like locked-in syndrome or paralysis, and could also benefit applications such as memory retrieval or visualizing thoughts.

Publication date: 16 Feb 2024
Project Page: https://arxiv.org/abs/2402.101
Paper: https://arxiv.org/pdf/2402.10115