Publications

Preprints

• Provably Efficient RLHF Pipeline: A Unified View from Contextual Bandits. [PDF, arXiv]
  Long-Fei Li*, Yu-Yang Qian*, Peng Zhao, and Zhi-Hua Zhou.
  
  
• Heavy-Tailed Linear Bandits: Huber Regression with One-Pass Update. [PDF, arXiv]
  Jing Wang, Yu-Jie Zhang, Peng Zhao, and Zhi-Hua Zhou.
  
  
• Stochastic Approximation Approaches to Group Distributionally Robust Optimization and Beyond. [PDF, arXiv]
  Lijun Zhang, Haomin Bai, Peng Zhao, Tianbao Yang, and Zhi-Hua Zhou.
  

Conference Papers

2024:

• Provably Efficient Reinforcement Learning with Multinomial Logit Function Approximation. [PDF, arXiv, bibtex]
  Long-Fei Li, Yu-Jie Zhang, Peng Zhao, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 37 (NeurIPS 2024), Vancouver, Canada, 2024. Page: 58539-58573.
  
  
• Gradient-Variation Online Learning under Generalized Smoothness. [PDF, arXiv, bibtex]
  Yan-Feng Xie, Peng Zhao, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 37 (NeurIPS 2024), Vancouver, Canada, 2024. Page: 37865-37899.
  
  
• A Simple and Optimal Approach for Universal Online Learning with Gradient Variations. [PDF, bibtex]
  Yu-Hu Yan, Peng Zhao, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 37 (NeurIPS 2024), Vancouver, Canada, 2024. Page: 11132-11163.
  
  
• Near-Optimal Dynamic Regret for Adversarial Linear Mixture MDPs. [PDF, arXiv, bibtex]
  Long-Fei Li, Peng Zhao, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 37 (NeurIPS 2024), Vancouver, Canada, 2024. Page: 55858-55883.
  
  
• Universal Online Convex Optimization with 1 Projection per Round. [PDF, arXiv, bibtex]
  Wenhao Yang, Yibo Wang, Peng Zhao, and Lijun Zhang.
  In: Advances in Neural Information Processing Systems 37 (NeurIPS 2024), Vancouver, Canada, 2024. Page: 31438-31472.
  
  
• Efficient Non-stationary Online Learning by Wavelets with Applications to Online Distribution Shift Adaptation. [PDF, bibtex]
  Yu-Yang Qian, Peng Zhao, Yu-Jie Zhang, Masashi Sugiyama, and Zhi-Hua Zhou.
  In: Proceedings of the 41st International Conference on Machine Learning (ICML 2024), Vienna, Austria, 2024. Page: 41383-41415.
  
  
• Learning with Adaptive Resource Allocation. [PDF, bibtex]
  Jing Wang, Miao Yu, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 41st International Conference on Machine Learning (ICML 2024), Vienna, Austria, 2024. Page: 52099-52116.
  
  
• Handling Heterogeneous Curvatures in Bandit LQR Control. [PDF, bibtex]
  Yu-Hu Yan, Jing Wang, and Peng Zhao.
  In: Proceedings of the 41st International Conference on Machine Learning (ICML 2024), Vienna, Austria, 2024. Page: 55839-55858.
  
  
• Improved Algorithm for Adversarial Linear Mixture MDPs with Bandit Feedback and Unknown Transition. [PDF, arXiv, bibtex]
  Long-Fei Li, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 27th International Conference on Artificial Intelligence and Statistics (AISTATS 2024), Valencia, Spain, 2024. Page: 3061-3069.
  
  
• Dynamic Regret of Adversarial MDPs with Unknown Transition and Linear Function Approximation. [PDF, bibtex]
  Long-Fei Li, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 38th AAAI Conference on Artificial Intelligence (AAAI 2024), Vancouver, Canada, 2024. Page: 13572-13580.
  
  

2023:

• Universal Online Learning with Gradient Variations: A Multi-layer Online Ensemble Approach. [PDF, arXiv, bibtex] (Spotlight)
  Yu-Hu Yan, Peng Zhao, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 36 (NeurIPS 2023), New Orleans, Louisiana, 2023. Page: 37682-37715.
  
  
• Dynamic Regret of Adversarial Linear Mixture MDPs. [PDF, bibtex]
  Long-Fei Li, Peng Zhao, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 36 (NeurIPS 2023), New Orleans, Louisiana, 2023. Page: 60685-60711.
  
  
• Adapting to Continuous Covariate Shift via Online Density Ratio Estimation. [PDF, arXiv, bibtex]
  Yu-Jie Zhang, Zhen-Yu Zhang, Peng Zhao, and Masashi Sugiyama.
  In: Advances in Neural Information Processing Systems 36 (NeurIPS 2023), New Orleans, Louisiana, 2023. Page: 29074-29113.
  
  
• Stochastic Approximation Approaches to Group Distributionally Robust Optimization. [PDF, arXiv, bibtex]
  Lijun Zhang, Peng Zhao, Zhen-Hua Zhuang, Tianbao Yang, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 36 (NeurIPS 2023), New Orleans, Louisiana, 2023. Page: 52490-52522.
  
  
• Handling New Class in Online Label Shift. [PDF, bibtex]
  Yu-Yang Qian*, Yong Bai*, Zhen-Yu Zhang, Peng Zhao, and Zhi-Hua Zhou. (* indicates equal contribution)
  In: Proceedings of the 23rd IEEE International Conference on Data Mining (ICDM 2023), Shanghai, China, 2023. Page: 1283-1288.
  
  
• Optimistic Online Mirror Descent for Bridging Stochastic and Adversarial Online Convex Optimization. [PDF, long version, arXiv, bibtex]
  Sijia Chen, Wei-Wei Tu, Peng Zhao, and Lijun Zhang.
  In: Proceedings of the 40th International Conference on Machine Learning (ICML 2023), Hawaii, Honolulu, 2023. Page: 5002-5035.
      ♣ Journal version [PDF] published at JMLR, with many more results (improved bounds for strongly convex function, new bounds for non-smooth scenarios).
  
  
• Fast Rates in Time-Varying Strongly Monotone Games. [PDF, bibtex]
  Yu-Hu Yan, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 40th International Conference on Machine Learning (ICML 2023), Hawaii, Honolulu, 2023. Page: 39138-39164.
  
  
• Revisiting Weighted Strategy for Non-stationary Parametric Bandits. [PDF, arXiv, bibtex]
  Jing Wang, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 26th International Conference on Artificial Intelligence and Statistics (AISTATS 2023), Valencia, Spain, 2023. Page: 7913-7942.
  
  
• Beyond Performative Prediction: Open-environment Learning with Presence of Corruptions. [PDF, bibtex]
  Jia-Wei Shan, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 26th International Conference on Artificial Intelligence and Statistics (AISTATS 2023), Valencia, Spain, 2023. Page: 7981-7998.
  
  

2022:

• Efficient Methods for Non-stationary Online Learning. [PDF, long version, arXiv, bibtex] (Oral)
  Peng Zhao, Yan-Feng Xie, Lijun Zhang, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 35 (NeurIPS 2022), New Orleans, Louisiana, 2022. Page: 11573-11585.
  
  
• Adapting to Online Label Shift with Provable Guarantees. [PDF, arXiv, code, bibtex]
  Yong Bai*, Yu-Jie Zhang*, Peng Zhao, Masashi Sugiyama, and Zhi-Hua Zhou. (* indicates equal contribution)
  In: Advances in Neural Information Processing Systems 35 (NeurIPS 2022), New Orleans, Louisiana, 2022. Page: 29960-29974.
  
  
• Corralling a Larger Band of Bandits: A Case Study on Switching Regret for Linear Bandits. [PDF, arXiv, bibtex]
  Haipeng Luo, Mengxiao Zhang, Peng Zhao, and Zhi-Hua Zhou. (alphabetical order)
  In: Proceedings of the 35th Annual Conference on Learning Theory (COLT 2022), London, UK, 2022. Page: 3635-3684.
  
  
• Adaptive Bandit Convex Optimization with Heterogeneous Curvature. [PDF, arXiv, bibtex]
  Haipeng Luo, Mengxiao Zhang, and Peng Zhao. (alphabetical order)
  In: Proceedings of the 35th Annual Conference on Learning Theory (COLT 2022), London, UK, 2022. Page: 1576-1612.
  
  
• No-Regret Learning in Time-Varying Zero-Sum Games. [PDF, arXiv, bibtex]
  Mengxiao Zhang*, Peng Zhao*, Haipeng Luo, and Zhi-Hua Zhou. (* indicates equal contribution)
  In: Proceedings of the 39th International Conference on Machine Learning (ICML 2022), Baltimore, Maryland, 2022. Page: 26772-26808.
  
  
• Dynamic Regret of Online Markov Decision Processes. [PDF, arXiv, full version, bibtex]
  Peng Zhao, Long-Fei Li, and Zhi-Hua Zhou.
  In: Proceedings of the 39th International Conference on Machine Learning (ICML 2022), Baltimore, Maryland, 2022. Page: 26865-26894.
  
  
• Non-stationary Online Learning with Memory and Non-stochastic Control. [PDF, arXiv, bibtex]
  Peng Zhao, Yu-Xiang Wang, and Zhi-Hua Zhou.
  In: Proceedings of the 25th International Conference on Artificial Intelligence and Statistics (AISTATS 2022), online, 2022. Page: 2101-2133.
      ♣ Journal version [PDF] published at JMLR with improved memory dependence (as well as lower bound).
  
  
• Optimal Rates of (Locally) Differentially Private Heavy-tailed Multi-Armed Bandits. [PDF, arXiv, bibtex]
  Youming Tao*, Yulian Wu*, Peng Zhao, and Di Wang. (* indicates equal contribution)
  In: Proceedings of the 25th International Conference on Artificial Intelligence and Statistics (AISTATS 2022), online, 2022. Page: 1546-1574.
  
  

2021:

• Improved Analysis for Dynamic Regret of Strongly Convex and Smooth Functions. [PDF, arXiv, bibtex]
  Peng Zhao and Lijun Zhang.
  In: Proceedings of the 3rd Conference on Learning for Dynamics and Control (L4DC 2021), online, 2021. Page: 48-59.
  
  
• Exploratory Machine Learning with Unknown Unknowns. [PDF, code, bibtex]
  Peng Zhao, Yu-Jie Zhang, and Zhi-Hua Zhou.
  In: Proceedings of the 35th AAAI Conference on Artificial Intelligence (AAAI 2021), online, 2021. Page: 10999-11006.
      ♣ Journal version [PDF] published at Artificial Intelligence.
  
  
• Towards Enabling Learnware to Handle Unseen Jobs. [PDF, code, bibtex]
  Yu-Jie Zhang, Yu-Hu Yan, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 35th AAAI Conference on Artificial Intelligence (AAAI 2021), online, 2021. Page: 10964-10972.
  
  
• Storage Fit Learning with Feature Evolvable Streams. [PDF, code, bibtex]
  Bo-Jian Hou, Yu-Hu Yan, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 35th AAAI Conference on Artificial Intelligence (AAAI 2021), online, 2021. Page: 7729-7736.
  
  

2020:

• Dynamic Regret of Convex and Smooth Functions. [PDF, arXiv, bibtex]
  Peng Zhao, Yu-Jie Zhang, Lijun Zhang, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 33 (NeurIPS 2020), Vancouver, Canada, 2020. Page: 12510-12520.
      ♣ Journal version [PDF] finally got published at JMLR (with an almost three-year review period..)
      ♣ Many new results are included, e.g., one-gradient complexity for gradient-variation dynamic regret and collaborative online ensemble framework.
  
  
• An Unbiased Risk Estimator for Learning with Augmented Classes. [PDF, arXiv, code, bibtex]
  Yu-Jie Zhang, Peng Zhao, Lanjihong Ma, and Zhi-Hua Zhou.
  In: Advances in Neural Information Processing Systems 33 (NeurIPS 2020), Vancouver, Canada, 2020. Page: 10247-10258.
  
  
• Learning with Feature and Distribution Evolvable Streams. [PDF, code, bibtex]
  Zhen-Yu Zhang, Peng Zhao, Yuan Jiang, and Zhi-Hua Zhou.
  In: Proceedings of the 37th International Conference on Machine Learning (ICML 2020), Vienna, Austria, 2020. Page: 11317-11327.
  
  
• A Simple Online Algorithm for Competing with Dynamic Comparators. [PDF, bibtex]
  Yu-Jie Zhang, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI 2020), Toronto, Canada, 2020. Page: 390-399.
  
  
• Bandit Convex Optimization in Non-stationary Environments. [PDF, journal, arXiv, bibtex]
  Peng Zhao, Guanghui Wang, Lijun Zhang, and Zhi-Hua Zhou.
  In: Proceedings of the 23rd International Conference on Artificial Intelligence and Statistics (AISTATS 2020), Palermo, Italy, 2020. Page: 1508-1518.
      ♣ Journal version [PDF] published at JMLR with new results for adaptive regret bound of BCO.
  
  
• A Simple Approach for Non-stationary Linear Bandits. [PDF, arXiv version2, errata, bibtex]
  Peng Zhao, Lijun Zhang, Yuan Jiang, and Zhi-Hua Zhou.
  In: Proceedings of the 23rd International Conference on Artificial Intelligence and Statistics (AISTATS 2020), Palermo, Italy, 2020. Page: 746-755.
      ♣ A correct and self-contained version [PDF] is updated at arXiv version2.
  
  
• Optimal Margin Distribution Learning in Dynamic Environments. [PDF, bibtex]
  Teng Zhang, Peng Zhao, and Hai Jin.
  In: Proceedings of the 34th AAAI Conference on Artificial Intelligence (AAAI 2020), New York, NY, 2020. Page: 6821-6828.
  
  

Before 2019:

• Nearest Neighbor Ensembles: An Effective Method for Difficult Problems in Streaming Classification with Emerging New Classes. [PDF, code, bibtex]
  Xin-Qiang Cai, Peng Zhao, Kai Ming Ting, Xin Mu, and Yuan Jiang.
  In: Proceedings of the 19th International Conference on Data Mining (ICDM 2019), Beijing, China, 2019. Page: 970-975.
  
  
• Learning from Incomplete and Inaccurate Supervision. [PDF, code, bibtex]
  Zhen-Yu Zhang, Peng Zhao, Yuan Jiang, and Zhi-Hua Zhou.
  In: Proceedings of the 25th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2019), Anchorage, AL, 2019. Page: 1017-1025.
      ♣ Journal version [PDF] published at IEEE TKDE.
  
  
• Improving Deep Forest by Confidence Screening. [PDF, code, bibtex]
  Ming Pang, Kai Ming Ting, Peng Zhao, and Zhi-Hua Zhou.
  In: Proceedings of the 18th IEEE International Conference on Data Mining (ICDM 2018), Singapore, 2018. Page: 1194-1199.
      ♣ Journal version [PDF] published at IEEE TKDE.
  
  
• Label Distribution Learning by Optimal Transport. [PDF, supp, code, bibtex]
  Peng Zhao and Zhi-Hua Zhou.
  In: Proceedings of the 32nd AAAI Conference on Artificial Intelligence (AAAI 2018), New Orleans, Louisiana, 2018. Page: 4506-4513.
  
  
• Dual Set Multi-Label Learning. [PDF, supp, code, bibtex]
  Chong Liu, Peng Zhao, Sheng-Jun Huang, Yuan Jiang, and Zhi-Hua Zhou.
  In: Proceedings of the 32nd AAAI Conference on Artificial Intelligence (AAAI 2018), New Orleans, Louisiana, 2018. Page: 3635-3642.
  
  
• Multi-View Matrix Completion for Clustering with Side Information. [PDF, code, bibtex]
  Peng Zhao, Yuan Jiang, and Zhi-Hua Zhou.
  In: Proceedings of the 21st Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD 2017), LNCS, Jeju, Korea, 2017. Page: 403-415.

Journal Papers

• Learning Objective Adaptation by Correlation-Based Model Reuse. [PDF, official version, bibtex]
  Lanjihong Ma, Yao-Xiang Ding, Peng Zhao, and Zhi-Hua Zhou.
  IEEE Transactions on Neural Networks and Learning Systems (TNNLS), in press, 2024.
  
  
• Learning with Asynchronous Labels. [PDF, code, bibtex]
  Yu-Yang Qian, Zhen-Yu Zhang, Peng Zhao, and Zhi-Hua Zhou.
  ACM Transactions on Knowledge Discovery from Data (TKDD), in press, 2024.
  
  
• Optimistic Online Mirror Descent for Bridging Stochastic and Adversarial Online Convex Optimization. [PDF, arXiv, bibtex]
  Sijia Chen, Yu-Jie Zhang, Wei-Wei Tu, Peng Zhao, and Lijun Zhang.
  Journal of Machine Learning Research (JMLR), 25(178):1−62, 2024.
  
  
• Adaptivity and Non-stationarity: Problem-dependent Dynamic Regret for Online Convex Optimization. [PDF, arXiv, bibtex]
  Peng Zhao, Yu-Jie Zhang, Lijun Zhang, and Zhi-Hua Zhou.
  Journal of Machine Learning Research (JMLR), 25(98):1−52, 2024.
  
  
• Exploratory Machine Learning with Unknown Unknowns. [PDF, arXiv, code, bibtex]
  Peng Zhao, Jia-Wei Shan, Yu-Jie Zhang, and Zhi-Hua Zhou.
  Artificial Intelligence (AIJ), Volume 327:104059, 2024.
  
  
• Online Non-stochastic Control with Partial Feedback. [PDF, bibtex]
  Yu-Hu Yan, Peng Zhao, and Zhi-Hua Zhou.
  Journal of Machine Learning Research (JMLR), 24(273):1−50, 2023.
  
  
• Non-stationary Online Learning with Memory and Non-stochastic Control. [PDF, arXiv, bibtex]
  Peng Zhao, Yu-Hu Yan, Yu-Xiang Wang, and Zhi-Hua Zhou.
  Journal of Machine Learning Research (JMLR), 24(206):1−70, 2023.
  
  
• Learning from Incomplete and Inaccurate Supervision. [PDF, official version, code, bibtex]
  Zhen-Yu Zhang, Peng Zhao, Yuan Jiang, and Zhi-Hua Zhou.
  IEEE Transactions on Knowledge and Data Engineering (TKDE), 34(12), 5854-5868, 2022.
  
  
• Improving Deep Forest by Screening. [PDF, official version, bibtex]
  Ming Pang, Kai Ming Ting, Peng Zhao, and Zhi-Hua Zhou.
  IEEE Transactions on Knowledge and Data Engineering (TKDE), 34(9), 4298-4312, 2022.
  
  
• Bandit Convex Optimization in Non-stationary Environments. [PDF, arXiv, bibtex]
  Peng Zhao, Guanghui Wang, Lijun Zhang, and Zhi-Hua Zhou.
  Journal of Machine Learning Research (JMLR), 22(125):1−45, 2021.
  
  
• 基于决策树模型重用的分布变化流数据学习. [PDF]
  赵鹏, 周志华.
  中国科学：信息科学, 2021, 51(1): 1-12.（封面文章）
  
  
• Distribution-Free One-Pass Learning. [PDF, official version, code, bibtex]
  Peng Zhao, Xinqiang Wang, Siyu Xie, Lei Guo, and Zhi-Hua Zhou.
  IEEE Transaction on Data Engineering (TKDE), 33(3): 951-963, 2021.
  
  
• Handling Concept Drift via Model Reuse. [PDF, official version, code, bibtex]
  Peng Zhao, Le-Wen Cai, and Zhi-Hua Zhou.
  Machine Learning (Special Issue of the ACML 2019 Journal Track), 109(3): 533-568, 2020.
  

Technical Notes

• Non-stationary Linear Bandits Revisited. [PDF, arXiv]
  Peng Zhao and Lijun Zhang. Technical Note, 2021.