PROJECT TITLE :

Multi-Label Learning with Global and Local Label Correlation - 2018

ABSTRACT:

It is well-known that exploiting label correlations is vital to multi-label learning. Existing approaches either assume that the label correlations are international and shared by all instances; or that the label correlations are local and shared only by a information subset. In reality, in the important-world applications, each cases might occur that some label correlations are globally applicable and a few are shared only in a native group of instances. Moreover, it's additionally a usual case that only partial labels are observed, which makes the exploitation of the label correlations a lot of additional troublesome. That is, it's hard to estimate the label correlations when many labels are absent. In this Project, we propose a new multi-label approach GLOCAL addressing each the full-label and the missing-label cases, exploiting global and local label correlations simultaneously, through learning a latent label illustration and optimizing label manifolds. The intensive experimental studies validate the effectiveness of our approach on both full-label and missing-label knowledge.