|
I have a possibly new algorithm for optimizing efficiently when the search space is the set of orthogonal matrices (the manifold of orthogonal matrices actually). I'm looking for applications beyond PCA and variants thereof, and since I'm not very familiar with current neural network literature I thought perhaps someone here could enlighten me. |
I think you have to provide more information. Can't see any applications beyond PCA. Effectively deep-learning is all non-linear... but why is the method good? Linear methods are v important for web scale data ...