ICML

TrialCalibre: A Fully Automated Causal Engine for RCT Benchmarking and Observational Trial Calibration

TrialCalibre：一种完全自动化的因果引擎用于RCT基准测试和观察性试验校准

Amir Habibdoust, Xing Song

AI总结 本文提出TrialCalibre，一种多智能体系统，用于自动化和扩展BenchExCal流程，通过专门的智能体协调整个过程，实现适应性、可审计和透明的因果效应估计。

Comments 5 pages , 2 figures

Journal ref Proceedings of the 42nd International Conference on Machine Learning, Vancouver, Canada. PMLR 267, 2025. Copyright 2025 by the author(s)

Online learning with Erdős-Rényi side-observation graphs

基于Erdős-Rényi边观察图的在线学习

Tomáš Kocák, Gergely Neu, Michal Valko

机构 * SequeL team INRIA Lille - Nord Europe（INRIA里尔-北欧SequeL团队） ； Universitat Pompeu Fabra Barcelona, Spain（巴塞罗那大学庞培法华大学） 纠错

AI总结 研究了对抗性多臂老虎机问题，其中学习者可观察非选择臂的损失。提出两种算法处理不同r范围，第一种在r≥(log T)/(2N)时达到O(√(T/r log N))的期望遗憾，第二种在较小r时达到O(√(T/r log(N+T)))。

Comments Published at International Conference on Machine Learning (ICML) 2015. 11 pages

Journal ref Proceedings of the 32nd International Conference on Machine Learning (ICML), 2015

Tabular Data with Class Imbalance: Predicting Electric Vehicle Crash Severity with Pretrained Transformers (TabPFN) and Mamba-Based Models

表格数据中的类别不平衡：利用预训练转换器（TabPFN）和Mamba-based模型预测电动汽车碰撞严重程度

Shriyank Somvanshi, Pavan Hebli, Gaurab Chhetri, Subasish Das

机构 * College of Science and Engineering, Texas State University, San Marcos, Texas, USA（科学与工程学院，德克萨斯州立大学，圣马科斯，德克萨斯州，美国） 纠错

AI总结 本文提出了一种深度表格学习框架，利用真实世界碰撞数据预测电动汽车碰撞严重程度，通过SMOTEENN处理类别不平衡问题，评估了TabPFN、MambaNet和MambaAttention三种模型，发现MambaAttention在严重伤害分类上表现更优。

Comments This is the author's preprint version of a paper accepted for presentation at the 24th International Conference on Machine Learning and Applications (ICMLA 2025), December 3-5, 2025, Florida, USA. The final published version will appear in the official IEEE proceedings. Conference site: https://www.icmla-conference.org/icmla25/

A Transformer-Based Cross-Platform Analysis of Public Discourse on the 15-Minute City Paradigm

基于Transformer的跨平台公共 discourse 对15分钟城市范式的分析

Gaurab Chhetri, Darrell Anderson, Boniphace Kutela, Subasish Das

机构 * 1 College of Science ； Engineering, Texas State University, San Marcos, Texas, USA Email ； 3 Texas A\&M Transportation Institute, Texas A\&M University, Houston, Texas, USA Email 纠错

AI总结 本研究首次跨平台分析15分钟城市概念的公共 discourse，采用压缩Transformer模型进行多平台情感分类，发现DistilRoBERTa表现最佳，MiniLM一致性最高，揭示平台特性差异及压缩模型的高效优势。

Comments This is the author's preprint version of a paper accepted for presentation at the 24th International Conference on Machine Learning and Applications (ICMLA 2025), December 3-5, 2025, Florida, USA. The final published version will appear in the official IEEE proceedings. Conference site: https://www.icmla-conference.org/icmla25/