机器学习与统计学(MLSTAT)会议是由中国现场统计研究会机器学习分会主办的学术会议。会议旨在促进机器学习与统计领域的国内外学者进行学术交流,引领机器学习与统计共同交叉发展的学术文化,推动作为数据科学与人工智能的奠基性学科的进步,以此助力相关数字经济产业的发展。
第四届机器学习与统计学会议(MLSTAT2026)将于2026年7月15日-17日在西南财经大学(四川成都市)举行。会议将邀请20位左右青年学者就机器学习、人工智能、统计学和应用数学等相关领域的前沿进展做大会主题报告,同时欢迎在读博士生进行墙报展示。

主题:基于Koopman理论的连续谱动力系统表示与预测方法
报告人简介:束俊,西安交通大学数学与统计学院副教授,博导。于2016年和2023年分别获西安交通大学理学学士与理学博士学位。目前从事于机器学习基础理论与算法研究;提出基于"模拟学习方法论"(简称SLeM)的学习框架,建立起相应的统计学习理论,为机器学习自动化提供了一个可行的理论途径,并在此框架下针对训练数据自选择、样本标记自校正、网络结构自调节、表现度量自构建、优化算法自设计等机器学习自动化任务研发了系列有效的机器学习自动化核心算法与技术。已在JMLR/TPAMI/TMLR/NSR/NeurIPS/ICML等国际顶级期刊和会议发表学术论文25余篇;ESI高被引论文2篇。谷歌学术引用2000余次,单篇论文引用超1100次。主持科技部重点研发计划青年科学家项目、基金委面上项目以及多项企业横向课题等。
报告摘要:对高维时空混沌动力系统进行表示与预测,仍然是动力系统理论与机器学习领域中的一个基础性挑战。高维、非线性且具有连续谱结构的动力系统广泛存在于气候演化、湍流流动、复杂网络传播以及神经动力学等现实场景中。尽管目前数据驱动方法能够实现较为准确的短期预测,但在以宽频或连续谱为主导的系统中,它们往往缺乏稳定性、可解释性和可扩展性。Koopman 理论为非线性动力学的表征和预测提供了一个线性化视角,但现有方法通常依赖于有限维逼近,这在高维场景下往往导致性能退化。本报告提出一种新的神经Koopman方法,通过将可逆运动与不可逆耗散分离,实现对动力系统的结构化表示。该方法在提高长期预测精度与稳定性的同时,也有助于揭示混沌行为中哪些方面是可以被理解和学习的。
主题:Evaluating biomarkers for treatment selection from reproducibility studies
报告人简介:宋晓, 佐治亚大学流行病和生物统计系教授,本科和硕士毕业于北京大学数学系,博士毕业于北卡罗莱纳州立大学统计系,曾任华盛顿大学(University of Washington) 公共卫生学院生物统计系研究助理教授,主要研究兴趣包括生存分析、测量误差和缺失数据、非参数方法、高维数据和图像数据、以及基于生物标志物的医学检验预测与精准医疗。
报告摘要:We consider evaluating biomarkers for treatment selection under assay modification. Survival outcome, treatment, and Affymetrix gene expression data were obtained from cancer patients. Consider migrating a gene expression biomarker to the Illumina platform. We propose an approach that allows a quick evaluation of the migrated biomarker with only a reproducibility study needed to compare the two platforms, achieved by treating the original biomarker as an error-contaminated observation of the migrated biomarker. We adopt a nonparametric regression model to characterize the relationship between the event rate and the biomarker and obtain an optimal marker-based treatment regime. Using B-spline approximation for nonparametric regression functions, we estimate the treatment regime via the SIMEX approach. SIMEX is a simple and general method that can handle measurement errors in complex cases. Through establishing the uniform convergence rate of the SIMEX estimator for the nonparametric regression functions, we derive the consistency and asymptotic normality of the empirical estimator of the biomarker performance measure. The approach is assessed by simulation studies and demonstrated through application to lung cancer data.
主题:Some Recent Progress on Matrix-Gradient Optimizers
报告人简介:苏炜杰现任宾夕法尼亚大学沃顿商学院、数学系和计算机系副教授,兼任宾大机器学习研究中心联席主任。2011年获得北京大学数学科学学院基础数学学士学位,2016年获得斯坦福大学博士学位。研究兴趣涵盖生成式人工智能的数学和统计基础、隐私保护机器学习、高维统计以及优化理论。
报告摘要:Introduced in December 2024, Muon is an optimization method for training language models that updates the weight along the direction of an orthogonalized gradient. The superiority of Muon has been quickly recognized, as demonstrated on industry-scale models; for example, it has been successfully used to train a trillion-parameter frontier language model. In this talk, we offer two perspectives to shed light on this matrix-gradient method. First, we introduce a unifying framework that precisely distinguishes between preconditioning for curvature anisotropy (like Adam) and gradient anisotropy (like Muon). This perspective not only offers new insights into Adam's instabilities and Muon's accelerated convergence but also leads to a new extension, such as PolarGrad. Next, we introduce a second perspective based on an isotropic curvature model. We derive this model by assuming isotropy of curvature (including Hessian and higher-order terms) across all perturbation directions. We show that under a general growth condition, the optimal update is one that makes the gradient's spectrum more homogeneous; that is, making its singular values closer in ratio. We then show that the orthogonalized gradient becomes optimal for this model when the curvature exhibits a phase transition in growth. Taken together, these results suggest that the gradient orthogonalization employed in Muon is directionally correct but may not be strictly optimal, and we will discuss how to leverage this model for designing new optimization methods. This talk is based on arXiv:2505.21799 and 2511.00674.
会议注册信息:
为了确保会议顺利开展,本次会议将少量收取注册费,会务组将承担会议期间的用餐,其他费用敬请自理!
注册费用:学生代表 200元,其他代表500元。
报名截止时间:2026年6月30日。
会议信息将通过本公众号及会议网站及时更新,欢迎大家积极关注。会议注册是通过会议网站注册。会议注册网址:
https://ml-stat.github.io/MLSTAT2026/register/
联系方式:
邮箱:mlstat2026@126.com
电话:028-87092330