Showing papers on "Autoencoder published in 2001"

TL;DR: This paper proposes a mixture model by concurrently performing global data partition and local linear PCA, and describes a modular classification scheme to solve the problem of handwritten digits recognition.

TL;DR: Clustering is thus formulated as the problem of finding the ground state of Potts spins Hamiltonians, and the partition, provided by this procedure, identifies clusters with dense connected regions in the data space.