Ben Xue

I am a researcher working on Computer Vision and Recommendation. I received my B.S. in EECS from Peking University in 2019, and my M.S. from the Academy for Advanced Interdisciplinary Studies (AAIS), PKU in 2022, advised by Prof. Yadong Mu. My research interests include pixel-level vision, image/video generation, generative recommendation, and discrete tokenization.

I previously worked at Alibaba Taobao Tech. and am currently a researcher at the Kuaishou Advertising AIGC Algorithm Team, focusing on generative recommendation and AI-driven content production.

I have published papers at top venues including ECCV, ICCV, KDD, and MobiCom.

Email / Scholar / Github

Research

	Universal Discrete Tokenizers: Principles, Applications, and Future Directions Ben Xue, J. Wang, Y. Li, J. Lan, H. Xu, J. Jia, P. Jiang, Q. Chen, K. Gai, L. Liu, et al. TechRxiv, 2026 paper (survey) A comprehensive study of discrete tokenizers covering their design principles, downstream applications, and future research directions.
	Recommendation as Generation: Unifying Personalized Video Generation and Recommendation at Industrial Scale Y. Cheng^, B. Wang^, H. Zhang^, X. Gao^, Z. Yin, Ben Xue, Y. Li, J. Xue, Y. Ma, et al. ^Equal contribution arXiv*, 2026 paper / project page Unifies personalized video generation and recommendation at industrial scale via a generation-centric formulation.
	Generative Recommendation for Large-Scale Advertising Ben Xue^, D. Liu^, L. Wang^, M. Sun^, P. Wang^, P. Zhang^, S. Shi^, T. Xu^, Y. Sha^, Z. Liu^, et al. ^Equal contribution KDD 2026 ADS Track, 2026 (Oral)* paper A generative recommendation framework tailored for large-scale online advertising systems.
	SweetTok: Semantic-Aware Spatial-Temporal Tokenizer for Compact Video Discretization Z. Tan^, Ben Xue^, J. Jia, J. Wang, W. Ye, S. Shi, M. Sun, W. Wu, Q. Chen, P. Jiang ^Equal contribution ICCV*, 2025 paper / code A semantic-aware spatial-temporal tokenizer that yields compact discrete representations for videos.
	An End-to-End, Low-Cost, and High-Fidelity 3D Video Pipeline for Mobile Devices T. Fang, C. Niu, Y. Sun, C. Lv, X. Jiang, Ben Xue, F. Wu, G. Chen MobiCom, 2024 paper A practical end-to-end 3D video pipeline optimized for mobile devices with low cost and high fidelity.
	DCCF: Deep Comprehensible Color Filter Learning Framework for High-Resolution Image Harmonization Ben Xue, S. Ran, Q. Chen, R. Jia, B. Zhao, X. Tang ECCV, 2022 (Oral) paper / code A deep, interpretable color filter framework for high-resolution image harmonization.
	Video2Subtitle: Matching Weakly-Synchronized Sequences via Dynamic Temporal Alignment Ben Xue, C. Liu, Y. Mu ICMR, 2022 (Oral) paper / code Matches weakly-synchronized video and subtitle sequences via dynamic temporal alignment.

Hobbies

🏀 Basketball ⚾ Baseball 🎸 Guitar 🍳 Cooking & Bakery

Website template based on Jon Barron's website.