Pinned Publications
No papers matching your search. Try different keywords.
📈 Contribution Graph
Less
More
📅 Contribution Activity
Representative Publications
Open-Source Repositories
Reviewer Service
Conference Reviewing
CVPR 2026
Conference
ECCV 2026
Conference
AAAI 2026
Conference
NeurIPS 2026
Conference
EMNLP 2026
Conference
NeurIPS 2025
Conference
NeurIPS 2024
Conference
ICML 2025
Conference
ICLR 2024
Conference
ACM MM 2025
Conference
ACM MM 2024
Conference
ACM MM 2023
Conference
Journal Reviewing
IEEE TNNLS
Journal
IEEE TMM
Journal
IEEE TYCB
Journal
Work Experience
🏭
Research Intern
Worked with Jiasheng Tang, Zhiwei Tang, Yizeng Han, and Bohan Zhuang on MLLM post-training and agent.
- Conducted research on hallucination and post-training robustness of MLLMs.
- Worked on Agent-as-a-Router for model routing in coding agent workflows; submitted related work to NeurIPS 2026.
- Analyzed MLLMs failure modes including multimodal hallucination and improved the generalization robustness of RLVR post-training; finished two related papers submitted to NeurIPS 2026 and EMNLP 2026.
- Contribute to evaluation framework for benchmarking medical foundation models, including Lingshu-32B.
🏢
Research Intern
Worked with Kaipeng Zhang on multimodal intelligence and open-ended interleaved image-text generation.
- Led research on MLLMs for interleaved image-text generation and evaluation.
- Proposed OpenING, the first comprehensive benchmark for open-ended interleaved image-text generation, accepted as an Oral at CVPR 2025.
- Studied multimodal knowledge-point retrieval-augmented generation for MLLM reasoning; published related work at AAAI 2026.
- Explored neural-driven generation by integrating EEG signals with diffusion-based image editing; published related work at NeurIPS 2025.
🎬
Technical Director
Institute of Intelligent Computing Technology, CAS
- Led a team of 8 employees to develop a visual system for dietary management and nutritional estimation.
- Proposed a zero-shot food detection algorithm with 99%+ accuracy in real-world scenarios.
- Deployed dietary systems in hospitals, companies, and schools with 100M+ usage, leading to a 3x checkout rate and 80% manpower reduction.
- Authorized a Chinese patent and published related papers in ACM MM 2023 and IEEE TIP.
🚀
Co-founder, CTO
Hangzhou Xiao Bin Technology Co. LTD
- Directed the Smart Home System Research and Development Laboratory with 10+ employees.
- Developed a smart bin recognition system with 99%+ classification accuracy, 30%+ higher than competitors.
- Sold products in major Chinese cities including Shanghai and Hangzhou, with annual turnover of CNY 100K+.
- Won 10+ competition prizes, including a Silver Award in Zhejiang International "Internet+" Undergraduate Entrepreneurship Competition and First Prize in the National Undergraduate IoT Design Contest.
About Pengfei Zhou
I am a PhD Candidate in Computer Science at the National University of Singapore (NUS), advised by Prof. Yang You. I obtained my M.E. from the Institute of Computing Technology, Chinese Academy of Sciences (ICT-CAS), supervised by Prof. Weiqing Min and Prof. Shuqiang Jiang, and my B.E. from Zhejiang University of Technology (ZJUT), advised by Prof. Cong Bai.
Research Interests
- Multimodal Learning & Post-Training
- LLMs & Agents & World Models
- Brain-Computer Interface for Generative Tasks
- Food Computing & Healthcare Applications
Education
- PhD — National University of Singapore (NUS), ongoing
- M.E. — Institute of Computing Technology, Chinese Academy of Sciences
- B.E. — Zhejiang University of Technology
Contact
Email: zpf4wp@outlook.com • GitHub • Google Scholar