LSTM-assisted evolutionary self-expressive subspace clustering

Massive volumes of high-dimensional data that evolve over time are continuously collected by contemporary information processing systems, which bring up the problem organizing these into clusters, i.e. achieving purpose dimensional reduction, and meanwhile learning their temporal evolution patterns. In this paper, a framework for evolutionary subspace clustering, referred to as LSTM–ESCM, is introduced, aims at clustering set evolving points lie in union low-dimensional subspaces. order obtain parsimonious representation each step, we propose exploit so-called self-expressive trait point. At same time, LSTM networks implemented extract inherited patterns behind overall frame. An efficient algorithm has been proposed. Numerous experiments carried out on real-world datasets demonstrate effectiveness our proposed approach. The results show suggested dramatically outperforms other known similar approaches terms both run accuracy.