Stat-ML 讨论班笔记（二）：估计与泛化理论
经验过程与度量熵、泛化界、PAC-Bayes、极小极大下界（Le Cam / Fano / Assouad）以及约束风险不等式（强数据处理不等式、本地隐私、通信复杂度）
Communication Compelxity and its Application in Estimation
Classical communication complexity problems, deterministic communication: lower bound and structure, randomization, conditional information complexity and direct sum, direct sum …