Non-negative matrix factorization (NMF or NNMF) [ ] has been widely used as a general method for dimensional reduction and feature extraction on non-negative data. The main difference between NMF and other factorization methods, such as SVD, is the nonnegativity, which allows only additive combinations of intrinsic ‘parts’, i.e. the hidden features. This is demonstrated in [ ], where NMF learns parts of faces and a face is naturally represented as an additive linear combination of different parts. Indeed, negative combinations are not as intuitive or natural as positive combinations.
展开▼