参数恒定的专家数目在基于对数损耗的在线学习中极大极小最优问题
Minimax Optimality in Online Learning under Logarithmic Loss with Parametric Constant Experts
作者：Fares Hedayati
作者单位：University of California
加工时间：2014-03-18
信息来源：EECS
索取原文[80 页]
关键词：在线学习；最大似然估计；在线预测；极大极小最优
摘 要：We focus on the minimax regret, which is the regret of the strategy with the minimum of the worst-case regret over outcome sequences.
027-87841330