Geometric Brain-Inspired AI: Leveraging Hyperbolic Neural Codes for Next-Generation BCIs

Neural and artificial systems organize information hierarchically, but current brain-AI alignment ignores this structure. Extending our work showing brain-aligned AI improves decoding, we develop models in non-Euclidean spaces that better capture hierarchical relationships in both systems to advance brain-computer interfaces and brain-inspired AI.

Biography

VafaeiHow does the brain encode what we see, hear, and feel — and can AI learn the same language? Shirin Vafaei is a PhD student at Osaka University supervised by Prof. Takufumi Yanagisawa, developing brain decoding and encoding algorithms that map neural activity to perception. Her research uses machine learning to explore the geometric structure of how the human brain represents sensory information, with the goal of building more accurate brain-computer interfaces that could restore communication and movement for people with neurological conditions.