NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:3426
Title:Copula Multi-label Learning


		
This paper presents a copula-based approach to multi-label multi-class classification. Under some conditions it is shown that the estimated conditional expectation is unbiased and bounds on the Mean Square Error are derived. The paper is well written and I found the analyzes and derivations correct. The use of copula for the study of the multi-label setting is new and the study can help to orient more theoretically driven works in this direction.