Publications
You can also find my articles on Google Scholar.
Published papers
ICML 2024 (Oral)
Developed a tabular foundation model for estimating dependence between random variables, achieving 1,000× faster estimation and higher accuracy on low-dimensional benchmarks. Downstream applications include feature selection, image segmentation and keyframe extraction.
Recommended citation: Z. Hu et al. (2024). "InfoNet: Neural Estimation of Mutual Information without Test-Time Optimization." ICML 2024 (Oral).
ICML 2026
Scaled InfoNet to high-dimensional dependence estimation through richer synthetic data and architecture design. Built data-generation and distributed pretraining pipelines for a 600M-parameter model on 32 H200 GPUs.
Recommended citation: Z. Hu et al. (2026). "InfoAtlas: A Foundation Model for Zero-Shot Statistical Dependence Estimate." ICML 2026.
Manuscripts under review
ICLR (under review)
Developed online diagnostics based on statistical dependence among hidden representations to detect anomalies during Transformer training, including overfitting, data anomalies and loss spikes.
Recommended citation: Z. Hu et al. (2026). "From Anisotropy to Anomaly: An Online Geometric Diagnostics Framework for Transformer Training." ICLR (under review).
ICLR (under review)
Pretrained a tabular foundation model for density estimation, outperforming all evaluated classical baselines on synthetic and real-world benchmarks. Applications include tail-risk estimation and outlier detection.
Recommended citation: Z. Hu et al. (2026). "DensityFM: Foundation-Style Pretraining for Density Estimation." ICLR (under review).
