• Professor 教授、博导；国家青年人才计划入选者、四川大学人才办副主任
• College of Computer Science， Sichuan University, China
• Basic Building 414, Wangjiang Campus, Sichuan University
  No.24 South Section 1, Yihuan Road, Chengdu, China, 610065
• 我每年在计算机学院招生，招生名额为硕士生5-6名，博士生3-4名，专业学位和科学学位均可，相关信息可以在我校研招办网站查询【博士招生目录】。硕士生原则上均可以在研二后去企业实习，如果提前达到优秀毕业条件可以提前实习。课题组的毕业生去向主要是高校、国企/公务员、AI大厂，更多毕业生去向可以点击查看。
• 我的主要研究方向是神经网络模型的设计理论和方法，应用涉及到大模型的优化设计、预训练/后训练、持续学习、鲁棒学习、机器学习自进化等领域，课题组的GPU计算资源充足，对我课题感兴趣的可以通过邮箱联系我 ysun@scu.edu.cn

News

• One paper about Continual Learning of LM reasoning learning, is accepted by NeurIPS2026. Congratulations to Wei!
• One paper about Perforamnce predictor, is accepted by NeurIPS2026. Congratulations to Jiawen!
• I am invited to be Area Chair of ICLR2027!
• One paper about Symbolic regression based on NAS, is accepted by TPAMI. Congratulations to Xiaotian, Peng, Yuwei! This submission was reviewed under TPAMI over 4 years.
• One paper about Efficient Continual Learning, is accepted by IJCV. Congratulations to Wei!
• I am inivited to be Tutorial Chair of IEEE CEC2027!
• I am invited to be Area Chair of NeurIPS2026!
• One paper about SSN LLM is accepted by ICME2026. Congratulations to Long and Xiaotian!
• Our paper about Continious Learning is accepted by CVPR2026, Congratulations to Aojun!
• Two work about schedulingare accepted by TCYB and ICAPS2026, respectively, Congratulations to Yun!
• Our paper about robust learning is accepted by IEEE TNNLS. Congratulations to Yuqi and Yuwei!
• Our paper about Performance Predictor for NAS is accepted by IEEE Transactions on Evolutionary Computation (TEVC). Congratulations to Junhao!
• Two papers about robust learning and SSN LLM are accepted by AAAI2026. Congratulations to Yuqi and Long!
• Congratulations to Jingrong, Han Ji, and Yuqi on winning the National Scholarship!
• Our paper about constrained optimization is accepted by SWEVO. Congratulations to Sri and Jiahao!
• Our work about Performance Predictor for NAS is accepted by IEEE Transactions on Computers. Congratulations to Xiaotian!
• I am awarded 2025 Outstanding Associate Editor of IEEE Transactions on Evolutionary Computation (TEVC)!
• Our work about robust learning is accepted by NeurIPS25. Congratulations to Yuqi!
• Two papers are accepted by PRICAI25. Congratulations to Jiawen (performance predictor) and Chunhui (continous learning)! Chunhui is a senior student.
• Our work about Performance Predictor for NAS is accepted by IEEE Transactions on Systems, Man, and Cybernetics: Systems (TSMC). Congratulations to Xiaotian!
• Our paper about AI4Science (fast design of nuclear reactor) is accepted by Nuclear Engineering and Design. Congratulations to Minxiao!
• Our paper about Exploring GPT-4o’s reasoning capabilities for panoramic radiograph is accepted by Clinical Oral Investigations. Congratulations to Yutao and Prof. Tang!
• Our paper about NAS for long-tailed learning is accepted by CIKM25. Congratulations to Yuhan and Prof. Gong.
• Two papers about performance predictor for NAS is accepted by ICCV25. Congratulations to Han!
• Our paper about Exploring GPT-4 for clinical decision-making is accepted by Front. Public Health. Congratulations to Yutao and Prof. Tang!
• Three papers are accepted by ICML25, Congratulations to Aojun (continous learning from neural architecture perspective), Jingrong (neural architecture design), and Zeqiong (theory of ENAS)!
• I am invited to be Publication Chair of PRICAI2025
• Our work about robust NAS is accepted by TKDE. Congratulations to Yuqi!
• I am invited to be Area Chair of IEEE CEC2025
• Our work about robust NAS is accepted by ICLR25. Congratulations to Yuqi!
• Our work about Evolutionary SNN is accepted by TEVC. Congratulations to Xiaotian!
• Again! We won the first place on Neural Architecture Search Challenge organized by AutoML24. [Media] (https://www.scu.edu.cn/info/1207/26284.htm) Congratulations to Aojun, Xiaotian, and Yuqi!
• I gave one-hour talk as part of specialized tutorial on Evolutionary computation and evolutionary deep learning for image analysis, signal processing and pattern recognition in GECCO2024.
• Our work about optimizer automation motivated by NAS is accepted by ECCV2024. Congratulations to Xiaotian!
• I gave two-hours tutorial on evolutionary neural architecture search in CEC2024.
• Our work about robust training of LLM is accepted by ICML2024. Congratulations to Prof. Pan and Yuhan!
• Three of our work about NAS are accepted by IJCAI2024. Congratulations to Aojun, Xiaotian, and Han!
• Our work about adversarial example generation is appected by FSE2024. Congratulations to Prof. Li!
• Our work about imbalanced data learning is appected by TKDE.
• Our work about robust NAS is accepted by TNNLS. Congratulations to Yuqi!
• Two of our work about theory of ENAS are accepted by GECCO2024 and TETCI. Congratulations to Zeqiong!
• Our work about schedulingis accepted by TEVC. Congratulations to Yun!
• Our work about robust NAS is accepted by CVPR2024. Congratulations to Yuwei and Yuqi!
• Yanan Sun is invited to be Associate Editor of IEEE Transactions on Neural Networks and Learning Systems.
• Our work about theory of ENAS is accepted by TEVC. Congratulations to Zeqiong!
• We won the first and second place on Neural Architecture Search Challenge organized by CVPR23. [Media] [1] [2] Congratulations to Aojun and Jingrong!
• Our work about Performance Predictor for NAS is accepted by TCYB. Congratulations to Xiangning!
• Yanan Sun is invited to be Associate Editor of IEEE Transactions on Evolutionary Computation from 2023.
• Our work about Performance Predictor for NAS is accepted by NeurIPS2022. Congratulations to Yuqiao!
• Our work about Evolving Transformer is awarded as Best Paper by MLMI2022. Congratulations to Jie!
• Our work about Benchmark Platform of ENAS is accepted by TEVC. Congratulations to Xiangning!
• Our work about Constraint NAS is accepted by TNNLS. Congratulations to Siyi!
• Our work about improving efficiency of NAS is accepted by ICCV21. Congratulations to Yuqiao!

Research Interest

My research interest focus on theory and applications of automated machine learning, including:

• explainable/interpretable neural architecture search (NAS)
• convergence analysis for evolutionary computation-based NAS
• multi-/many-objective/ and constrained optimization for NAS
• low-energy consumption NAS with high-inference speed
• feature selection and construction
• auto data augmentation