Lead ContributorManuscript, 2026
Brain Hierarchical Alignment Improves Generalizable Visual Decoding via Bridge Priors
Matching brain activity to images to retrieve and reconstruct what a person sees.
Kunpeng Xie
I study how AI models can learn from brain signals, with a focus on vision and language.
Neural Computing and Control Laboratory, SUSTech
Advised by Prof. Quanying Liu
Southern University of Science and Technology (SUSTech)
I’m pursuing a B.Eng. in Intelligent Medical Engineering and expect to graduate in 2027.
GPA: 3.73/4.00
Research Interests
I’m interested in how models represent brain activity and relate it to images and language. I also explore how neural feedback can guide image generation as a person interacts with a model.
Selected Research
ContributorACM Multimedia 2025
BrainFLORA: Uncovering Brain Concept Representation via Multimodal Neural Embeddings
Studying how visual concepts are represented across EEG, MEG, and fMRI.
ContributorResearch Project
BrainWhisperer: Aligned Semantic Representations for Language Model-based Decoding
Connecting brain signals with language models to decode meaning.
Friends
A brain-computer interface modeling startup building a flywheel for neural data collection and model training.
A creativity-first independent team focused on practical AI products and reliable web systems.