The Cluster of Excellence: 'Machine Learning: New Perspectives for Science' @NeurIPS2020

The cluster is a joint initiative of @unituebingen , the @maxplanckinstituteforintel3078 and the Leibniz Institute IWM, funded by the German Science Foundation. Our goal is to develop machine learning algorithms to advances science. We have a diverse and young team of researchers, who tell you more about their research in this video.

[nguyenngocly]

You can swap what is adjustable in neural nets. Using fixed dot products (enacted with fast transforms) and adjustable (parametric) activation functions. Eg. fi(x)=ai.x x<0, fi(x)=bi.x x>=0, i=0 to n-1. The fast Walsh Hadamard transform is a good choice. To stop the first transform taking a spectrum you can apply a random fixed pattern of sign flips to the input data. The net then is: sign flips, transform, activation functions, transform, activation