Kernel-based Online Regression with Canal Loss
EUROPEAN JOURNAL OF OPERATIONAL RESEARCH(2022)
摘要
•Propose a kernel-based online regression algorithm (NROR) to deal with noisy labels.•Prove that the regularized canal loss is strongly pseudo-convex.•Sublinear regret bound of NROR is established under certain assumptions.•Validate the numerical performance of NROR on the datasets of heavy noisy labels.
更多查看译文
关键词
Data science,Regression,Nonconvex optimization,Regret bound,Noisy label
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要