Machine Learning New Math – Deep Layer Linear Algebra. Solve problems in reduced coeffient space. Since quadratic matrices are too large. Represent the problem in two or three layers with reduction of parameters. // WOW Per

Machine Learning Complex Numbers guess – How many layers does it take to represent complex numbers? layer1 x = sqrt(-1) to y representation in layer2. One layer number can represent complex if its quadratic. Then two layers to reduce the coeffients greatly. x = model(X) and x->X.

Machine Learning – 3 layer Fourier Series Coeffients. Since you can iterate 1 sample with a perceptron of 3 layers to get the x = X. A fourier version with 3 layers can be developed. Not just one layer of complex coeffients. Could complex coffient vector be 2 non linear layers?