PredictingLabelDistributionfromMulti-label Ranking

Therefore, Eq.(5) holds for k = 2,3, . A.4 ProofofCorollary2 Corollary2 If an instance is annotated by a multi-label rankingσ, m is the number of relevant labels,0 δ m 1,m(m+1) 2 ˆδ m 1,thentheEAEofσisboundedby: A.10 DetailsofExperiments The information of the datasets we used is shown in Table 1. The first four rows in Table 1 are the existing label distribution datasets; the last three rows in Table 1 are the datasets we created. Since some examples in the original label distribution datasets do not satisfy the prerequisites of our paper (i.e., there are some examples(x,d) such that there exist relevant labels with identical label description degrees), we remove these examples from the dataset to obtain such a dataset: {(x,d) D| (di 6= 0,dj 6= 0),di 6= dj}, where D = {(xn,dn)}Nn=1.