We developed a learning-based automatic localizer to accurately identify hippocampal sublayer in mice from high-density extracellular recordings by aligning neural manifolds across sessions and animals.
DeepSeparator is a deep learning framework designed to separate neural signals from artifacts in EEG recordings and reconstruct clean signals.
We developed a transformer-based agent that achieves meta-reinforcement learning through in-context learning, demonstrating one-shot learning in a 4x4 recurrent spatial navigation task.
Satori is a novel AR assistant framework that integrates the Belief-Desire-Intent (BDI) model of user and a multi-modal large language model (LLM), to provide proactive guidance.