Title :
Nonlinear ICA through low-complexity autoencoders
Author :
Hochreiter, Sepp ; Schmidhuber, Jürgen
Author_Institution :
Fakultat fur Inf., Tech. Univ. Munchen, Germany
Abstract :
We train autoencoders by flat minimum search (FMS), a regularizer algorithm for finding low-complexity networks describable by few bits of information. As a by-product, this encourages nonlinear independent component analysis (ICA) and sparse codes of the input data
Keywords :
computational complexity; neural nets; principal component analysis; sparse matrices; flat minimum search; independent component analysis; low-complexity autoencoders; low-complexity networks; nonlinear ICA; regularizer algorithm; sparse codes; Decoding; Equations; Flexible manufacturing systems; Independent component analysis; Principal component analysis; Source separation;
Conference_Titel :
Circuits and Systems, 1999. ISCAS '99. Proceedings of the 1999 IEEE International Symposium on
Conference_Location :
Orlando, FL
Print_ISBN :
0-7803-5471-0
DOI :
10.1109/ISCAS.1999.777509
