November 7, 2025 LimiX: Unleashing Structured-Data Modeling Capability for Generalist Intelligence 首个统一的大规模结构化数据模型(LDM)，通过上下文条件掩码建模实现分类、回归、缺失值填补、数据生成等多任务的单模型泛化 Tabular Foundation Model Large Structured-Data Model In-Context Learning Masked Modeling Structural Causal Models Scaling Law Classification Regression Missing Value Imputation Data Generation
February 14, 2024 Causal Deep Learning: Encouraging Impact on Real-World Problems Through Causality 剑桥大学提出的CDL框架，通过结构、参数和时间三维度对因果深度学习进行系统分类，弥合因果推理与深度学习的鸿沟 Causal Deep Learning Causal Inference Deep Learning Structural Causal Models Treatment Effects Framework Paradigm Shift