[2Win5] Poster session 2

[2Win5-62] Reconstructing Visual Stimuli from Brain Representations using a Diffusion Model

〇Fumi Ishizaki¹, Ichiro KOBAYASHI¹ (1.Ochanomizu University)

Keywords:Brain decoding, Diffusion model, Image stimuli

The human brain possesses advanced mechanisms for processing complex external information, and understanding these mechanisms is expected to lead to further advancements in image recognition and speech recognition technologies within the field of artificial intelligence. Research focusing on the decoding of visual experiences has significantly contributed to the development of methods for accurately interpreting how the brain processes visual information and determines what a person perceives. In this study, we propose a novel approach that utilizes Stable Diffusion, a generative model, to predict and reconstruct visual experiences from brain activity data. Rather than relying on conventional text-based inputs, this method uses brain activity data as input to reconstruct internal representations. This approach enables the decoding of visual experiences with greater precision, offering the potential for deeper insights into the relationship between the brain and vision.