Ziheng Zhang

I'm a first year Ph.D. student at the University of Texas at Austin, working with Prof. Ying Ding. Previously, I obtained my B.S. from The Ohio State University where I work with Prof. Weilun (Harry) Chao and Dr. Jianyang Gu.

I develop simple but effective algorithms to solve machine learning problems, with a focus on multimodal, long-tail, and fine-grained settings.

Selected Publications

BioCAP: Exploiting Synthetic Captions Beyond Labels in Biological Foundation Models

TLDR: Integrate instance-level synthetic captions in multimodal alignment and demonstrate rich understanding of biological semantics.

ICLR 2026

arXiv Project page Code

Finer-CAM: Spotting the Difference Reveals Finer Details for Visual Explanation

TLDR: Compare the target class with similar classes for more fine-grained and discriminative explanation.

CVPR 2025

arXiv Demo Code

AISPA: User-Centric System Prompt Auditing for Large Language Model Applications

arXiv, 2026

arXiv

Probing Identity-Specific Motion Signatures: A Controlled Diagnostic Study

arXiv, 2026

arXiv

Beyond Flat Labels: Level-Restricted Contrastive Learning for Hierarchical Fine-Grained Vision Classification

arXiv, 2026

arXiv

MMSpec: Benchmarking Speculative Decoding for Vision-Language Models

arXiv, 2026

arXiv

Speculative Decoding for Multimodal Models: A Survey

Preprint, 2026

Preprint

Prompt-CAM: Making Vision Transformers Interpretable for Fine-Grained Analysis

CVPR 2025

arXiv Demo Code

BeetleFlow: An Integrative Deep Learning Pipeline for Beetle Image Processing

NeurIPS 2025 Workshop for Imageomics

arXiv

Building Machine Learning Challenges for Anomaly Detection in Science

arXiv

ADS