Multimodal correlations-based data clustering

<p style='text-indent:20px;'>This work proposes a novel technique for clustering multimodal data according to their information content. Statistical correlations present in that contain similar are exploited perform the task. Specifically, multiset canonical correlation analysis is equipped with norm-one regularization mechanisms identify clusters within different types of share same A pertinent minimization formulation put forth, while block coordinate descent employed derive batch algorithm which achieves better performance than existing alternatives. Relying on subgradient descent, an online approach derived substantially lowers computational complexity compared approach, not compromising significantly performance. It established increasing number regularized framework able correctly cluster entries. Further, it proved scheme converges probability one stationary point ensemble cost having potential recover correct clusters. Extensive numerical tests demonstrate outperforms alternatives, substantial savings.</p>