文件列表:

NMF.m (3649, 2015-06-26)
NMF2D.m (4081, 2015-06-26)
NMFD (0, 2015-06-26)
NMFD\NMFD.m (3241, 2015-06-26)
NMFD\drumLoops (0, 2015-06-26)
NMFD\drumLoops\DL6.wav (705644, 2015-06-26)
NMFD\script_test_NMFD.m (2231, 2015-06-26)
PLCA.m (4474, 2015-06-26)
python (0, 2015-06-26)
python\NMF.py (2224, 2015-06-26)
python\NMFD.py (3255, 2015-06-26)

# NMF-matlab Matlab code of Non-Negative matrix factorization (NMF) and variants, using multiplicative update rules for a beta-divergence cost (including Itakura Saito divergence, Kullback Leibler divergence and Froebenius distance). Implemented variants: - Probabilistic Latent Component Analysis (PLCA) as proposed by Smaragdis. This is actually a probabilistic model for NMF that corresponds to a Kullback-Leibler cost. - Non Negative Factor Deconvolution (NMFD) as proposed by Smaragdis. This is a time-convolutive variant of NMF that factorize time/frequency patterns. Useful for instance to model drum sounds. - Non Negative Matrix Factor 2-D Deconvolution (NMF2D) as proposed in Non Negative Matrix Factor 2-D Deconvolution for Blind Single Channel Source Separation by Mikkel N. Schmidt and Morten Mrup. This is a time-frequency-convolutive variant of NMF that factorize time/frequency patterns that can be shifted both in time and frequency. Useful to decompose representation for which frequency shift make sense, such as Constant-Q spectrogram.

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：