A reasonable way to factor a non-negative matrix $A$ into two non-negative matrices $M$ and $N$ can be done by minimizing the squared Frobenius norm, $||A - (M^2) (N^2)||_F^2$, where $M^2$ is the operation of squaring each entry of the matrix $M$ element wise (likewise with $N^2$). The final factorization of $A$ is given by $A = WH$, where $W = M^2$ and $H = N^2$, and squaring is done element-wise as before.

I tried to search for information about this method, but I could not find anything. I am specifically interested in how this method relates to non-negative matrix factorization (NMF), and the pros and cons of factorizing a matrix by the method described above over using NMF.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.