Hi, I am Chen Liu, a PhD student in Computer Science at Yale University.
My research focuses on the geometry of learned latent representations, investigating how neural networks organize information internally, when it breaks down, and how to fix it.
For LLMs, I identified and systematically characterized embedding condensation, a geometric phenomenon where token embeddings collapse into a narrow cone, and showed how to mitigate it to improve generalization in pre- and mid-training (ICML 2026).
For MLLMs, I co-developed VIGIL, an RL post-training framework that punishes visual laziness by penalizing over-reliance on language priors when visual evidence is present (ECCV 2026).
I apply the same perspective to on-manifold generative modeling (RNAGenScape), multimodal representation learning (ImmunoStruct @ Nature Machine Intelligence), unsupervised image segmentation (CUTS @ MICCAI 2024, DiffKillR @ ICASSP 2025 Oral) and latent dynamics modeling (ImageFlowNet @ ICASSP 2025 Oral).
โ๏ธ How I make high-quality figures for publications in top AI conferences and journals
โ๏ธ The first free tool to generate your Google Scholar Citation World Map
โ๏ธ The first tool to quantify your Github repo star percentile
ย ๐ [ICML 2026] Dispersion loss counteracts embedding condensation and improves generalization in small language models
ย ๐ [Nature Machine Intelligence] ImmunoStruct enables multimodal deep learning for immunogenicity prediction
ย ๐ [ICASSP 2025 Oral] ImageFlowNet: Forecasting Multiscale Image-Level Trajectories of Disease Progression with Irregularly-Sampled Longitudinal Medical Images
ย ๐ [ICASSP 2025 Oral] DiffKillR: Killing and Recreating Diffeomorphisms for Cell Annotation in Dense Microscopy Images
ย ๐ [MICCAI 2024] CUTS: A deep learning and topological framework for multigranular unsupervised medical image segmentation
ย ๐ [ICMLW 2023] A novel method to compute entropy and mutual information in neural net representations
โ๏ธ An on-the-fly evaluator for GANs, with a simple working example of DCGAN on SVHN
โ๏ธ A Technical Deep Dive into Drag Your GAN (DragGAN)