Corruption Robust Image Classification with a new Activation Function. Our proposed Activation Function is inspired by the Human Visual System and a classic signal processing fix for data corruption.

The project is to implement the Error Back-Propagation (EBP) training algorithm for a multi-layer perceptron (MLP) 4-2-4 encoder.

This is the recent work of my on the importance and application of mathematical function around its Hilbert function theory on artificial intelligence algorithms. The main motivation was the desire of improving the convergence rate and learning rate of various learning algorithms via Generalized Gaussian Radial Basis Function.