PhD student, The Hong Kong University of Science and Technology
2 papers at NeurIPS 2025
HoloV prunes 88.9% visual tokens in MLLMs while retaining 95.8% performance via spatial visual holistic context retention.
The paper systematically analyzes the phenomenon of varying sparsity ratios across views in multi-view learning, and proposed a targeted data-driven network architecture based on Sparse Autoencoder with Adaptive Constraints.
