Organizers
Eric Moulines, MBZUAI, Ecole Polytechnique ✉ 🌐
Alexey Naumov, HSE University, Steklov Mathematical Institute of RAS ✉ 🌐
Sergey Samsonov, HSE University ✉ 🌐
Yuhao Wang (王禹皓), Tsinghua University ✉ 🌐
Abstract
The Statistical AI workshop focuses on the statistical ideas that make modern AI reliable - how to quantify uncertainty, make decisions under uncertainty, and design theoretically grounded algorithms that still scale to large data settings. As models grow larger and are deployed more widely, these questions become both more urgent and more technically challenging. The workshop will bring together researchers across statistics, machine learning, and applied AI to discuss recent results and open problems. The workshop will cover, but is not limited to, the following key topics:
● Statistical decision making
● Generative modelling and sampling
● Conformal prediction
● Statistical and causal inference
● Stochastic optimisation, stochastic approximation, and optimization methods for large-scale and non-convex AI models
统计人工智能研讨会聚焦于可以使现代人工智能系统具备可靠性的核心统计思想——包括如何量化不确定性、在不确定条件下进行决策，以及设计既能保持理论严谨性、又能扩展到大规模数据场景的算法。随着模型规模不断扩大、部署日益广泛，这些问题变得愈发紧迫，技术挑战也愈加复杂。本次研讨会将汇聚来自统计学、机器学习及应用人工智能领域的研究人员，共同探讨最新成果与开放问题。研讨议题涵盖但不限于以下关键方向：
● 统计决策论
● 生成式模型采样
● 共形预测
● 统计与因果推断
● 随机优化、随机逼近，以及面向大规模与非凸人工智能模型的优化方法
