Bio

Yang Yu is now a graduate student at University of Science and Technology of China (USTC), BDAA Lab. His advisor is Prof. Qi Liu. His current research interests include recommender systems, user modeling, natural language processing, and model compression.

Yang Yu, Qi Liu, Kai Zhang, Yuren Zhang, Chao Song, Min Hou, Yuqing Yuan, Zhihao Ye, Zaixi Zhang, and Sanshi Lei Yu. AdaptSSR: Pre-training User Model with Augmentation-Adaptive Self-Supervised Ranking. In NeurIPS, 2023. [Paper] [Code] [Poster] [Slides]
Yang Yu, Qi Liu, Likang Wu, Runlong Yu, Sanshi Lei Yu, and Zaixi Zhang. Untargeted Attack against Federated Recommendation Systems via Poisonous Item Embeddings and the Defense. In AAAI, 2023. [Paper] [Code] [Poster] [Slides] [Video]
Yang Yu, Fangzhao Wu, Chuhan Wu, Jingwei Yi, and Qi Liu. Tiny-NewsRec: Effective and Efficient PLM-based News Recommendation. In EMNLP, 2022. [Paper] [Code] [Poster]
Sanshi Lei Yu, Qi Liu, Fei Wang, Yang Yu, and Enhong Chen. Federated News Recommendation with Fine-grained Interpolation and Dynamic Clustering. In CIKM, 2023. [Paper] [Code]
Chao Zhang, Fangzhao Wu, Jingwei Yi, Derong Xu, Yang Yu, Jindong Wang, Yidong Wang, Tong Xu, Xing Xie, and Enhong Chen. Non-IID always Bad? Semi-Supervised Heterogeneous Federated Learning with Local Knowledge Enhancement. In CIKM, 2023. [Paper] [Code]
Tao Qi, Fangzhao Wu, Chuhan Wu, Peiru Yang, Yang Yu, Xing Xie, and Yongfeng Huang. HieRec: Hierarchical User Interest Modeling for Personalized News Recommendation. In ACL-IJCNLP, 2021. [Paper]
Chuhan Wu, Fangzhao Wu, Yang Yu, Tao Qi, Yongfeng Huang, and Qi Liu. NewsBERT: Distilling Pre-trained Language Model for Intelligent News Application. In Findings of EMNLP, 2021. [Paper]

Anhui Province Key Laboratory of Big Data Analysis and Application (BDAA). Advisor: Prof. Qi Liu.
Working on recommender systems, federated learning, and natural language processing.
Designed an effective clustering-based untargeted attack method against federated recommendation systems and a general uniformity-based defense mechanism (AAAI'23).

Working on empowering various user-oriented services (e.g., user profiling, personalized recommendation) by pre-training the user model with users’ app-related behaviors.
Designed a new self-supervised ranking task for user model pre-training (NeurIPS'23).

Social Computing Group. Mentor: Dr. Fangzhao Wu and Dr. Xing Xie.
Mainly worked on news understanding and recommendation.
Developed multilingual news recommendation models for MSN online services.
Designed a self-supervised domain-specific post-training method and a two-stage knowledge distillation framework for effective and efficient PLM-based news recommendation (EMNLP'22).

Worked on the compression of generative models for edge devices.
Implemented several methods for image super-resolution (e.g. EDSR, RCAN, ESRGAN) and model compression (e.g., pruning, quantization, knowledge distillation).

Aiming to gather, standardize, and analyze massive multi-modal educational resources on the Internet and provide various intelligent education services for students and teachers.
In charge of the educational resources search system and the test paper analysis service.
Empowered the recall and ranking pipeline of the search system with pre-trained language models.

Implemented various attack methods and defense mechanisms for federated recommendation systems in a unified framework.

Implemented several deep learning-based news recommendation methods (e.g., NAML, NRMS) in PyTorch, with the support of multi-GPU training and evaluation.

2020, 2023China National Scholarship (Top 0.2% nationwide), Ministry of Education of the People’s Republic of China

2019, 2022Huawei Scholarship (Top 2%), University of Science and Technology of China

2021, 2022, 2023First Prize Academic Scholarship, University of Science and Technology of China

2021"Stars of Tomorrow" Internship Award of Excellence, Microsoft Research Asia

Publications