Changhao He

Changhao He / 何长浩

I am Changhao He (何长浩), a first-year Ph.D. student in the College of Computer Science at Sichuan University, where I am fortunate to be advised by Prof. Peng Hu (胡鹏) and Prof. Xi Peng (彭玺).

I received my Bachelor's degree (2019-2023) from the College of Mechanical Engineering at Sichuan University. After that, I pursued my Master's in the College of Computer Science at Sichuan University and transitioned into the Ph.D. stage through the direct doctoral program (2+4) in 2025. I have been with the XLearning Lab since my Master's studies.

Email / Google Scholar / Github / CV / WeChat

Taken at Kelingking Beach, Nusa Penida

Research

My research focuses on the robustness of multi-modal models during inference. Our group maintains the Awesome Noisy Correspondence repository , a resource hub for research in noisy correspondence. If you have any suggestions or ideas, we warmly invite you to get in touch with us.

News

[2026-07] 🎉🎉 One paper was accepted by ACM MM 2026 as Oral Presentation! Thanks to all coauthors.

[2026-07] 🥳 First-time reviewer for AAAI 2027, BMVC 2026, IEEE TPAMI.

[2026-05] 🎉🎉 Two papers were accepted by ICML 2026! One was accepted as Oral Presentation (accept rate≈0.70%)! Thanks to all coauthors.

[2026-04] Gave a talk @ SouthWest Petroleum University. Slides available.

[2026-03] 🥳 First-time reviewer for NeurIPS 2026.

[2026-02] 🎉🎉 One paper was accepted by CVPR 2026! Thanks to all coauthors.

[2026-01] 🥳 First-time reviewer for ICML 2026, ECCV 2026.

[2025-10] 🥳 First-time reviewer for CVPR 2026. Looking forward to serving the community!

[2025-04] Gave a talk @ Chongqing University of Posts and Telecommunications. Slides available.

[2025-02] 🎉🎉 One paper was accepted by CVPR 2025! Thanks to all coauthors.

[2025-01] ✅ Transitioned into Ph.D. stage through the direct doctoral program (2+4)!

[2024-07] 🎉🎉 One paper was accepted by ACM MM 2024! Thanks to all coauthors.

[2023-06] 🎓🎓 Graduated from Sichuan University!

Publications (*: equal contribution, †: corresponding author)

Robust Multiview Learning Under Noisy Correspondence

Shuxian Li, Changhao He, Xi Peng, Peng Hu^†

ACM MM, 2026 (Oral Presentation)
Paper(Coming soon) /

Code(Coming soon)

Formalize noisy correspondence in multiview learning and propose CARF, a consistency-aware robust fusion framework that jointly learns cross-view consensus and estimates view reliability for robust prediction.

RLSF-V: Mitigating Hallucinations in MLLMs via Fuzzy Semantic Self-Feedback

Changhao He, Shuhao Yan, Shuxian Li, Xi Peng, Peng Hu

ICML, 2026
Paper /

Code / Model / Dataset

Introduce a self-feedback preference optimization framework that assesses hallucinations through local fuzzy semantics derived from internal logits, eliminating the need for external large-model evaluators or human annotations.

DOUBT: Decoupled Object-level Understanding and Bridging via vMF-based Trustworthiness for Hallucination Detection in MLLMs

Kaiqi Chen, Yang Qin, Changhao He, Xi Peng, Peng Hu

ICML, 2026 (Oral Presentation, Top 0.70%)
Paper /

Code

Decouple object recognition from complex reasoning and introduce a model-agnostic vMF-based uncertainty metric for stable hallucination detection in MLLMs.

Bootstrapping Multi-view Learning for Test-time Noisy Correspondence

Changhao He, Di Xue, Shuxian Li, Yanji Hao, Xi Peng, Peng Hu

CVPR, 2026
Paper /

Code / Dataset

Formalize Test-time Noisy Correspondence (TNC) in multimodal learning and introduce an in-place bootstrapping framework together with an RGB-Depth-Text benchmark for robust inference.

Learning with Noisy Triplet Correspondence for Composed Image Retrieval

Shuxian Li*, Changhao He*, Xiting Liu, Joey Tianyi Zhou, Xi Peng, Peng Hu

CVPR, 2025
Paper /

Code /

Model

Introduce learning with noisy triplet correspondence in composed image retrieval, offering a new design perspective for existing supervised methods.

Robust Variational Contrastive Learning for Partially View-unaligned Clustering

Changhao He, Hongyuan Zhu, Peng Hu, Xi Peng

ACM MM, 2024
Paper /

Code

Introduce a variational contrastive learning paradigm for robust clustering under the Partially View-aligned Problem (PVP).

Honors and Awards

• Outstanding Student of Sichuan University, 2025.10

• First-Class Academic Scholarship, Sichuan University, 2025.07

• National Scholarship, 2024.11

• Outstanding Student of Sichuan University, 2024.10

Services

• Conference Reviewer: CVPR, ICML, NeurIPS, AAAI, ECCV, BMVC.

• Journal Reviewer: IEEE TPAMI, IEEE TCSVT.

Last updated:
Design and source code from Jon Barron