hitcszq / NMF-Tensorflow

Non-negative Matrix Factorization (NMF) Tensorflow Implementation

NMF-Tensorflow

Non-negative Matrix Factorization (NMF) Tensorflow Implementation.

Examples

>>> import numpy as np
>>> from nmf import NMF
>>> V = np.array([[1, 1], [2, 1], [3, 1.2], [4, 1], [5, 0.8], [6, 1]])
>>> model = NMF(max_iter=200,learning_rate=0.01,display_step=10, optimizer='mu')
>>> W, H = model.fit_transform(V, r_components=2, initW=False, givenW=0)
>>> print(V)
>>> print(model.inverse_transform(W, H))

Optimization

• Gradient Descent with Multiplicative Update Rule
• Projected Gradient

TODO:

• Sparse NMF
• Discriminant NMF