4 papers
2nd-order Updates with 1st-order Complexity
Michael F. Zimmer
It has long been a goal to efficiently compute and use second order information on a function () to assist in numerical approximations. Here it is shown how, using only basic ph…
Neograd: Near-Ideal Gradient Descent
Michael F. Zimmer
The purpose of this paper is to improve upon existing variants of gradient descent by solving two problems: (1) removing (or reducing) the plateau that occurs while minimizing the…
Speedup from a different parametrization within the Neural Network algorithm
Michael F. Zimmer
A different parametrization of the hyperplanes is used in the neural network algorithm. As demonstrated on several autoencoder examples it significantly outperforms the usual param…
A row space method for solving a system of linear equations
Michael F. Zimmer
A new algorithm is presented for computing a direct solution to a system of consistent linear equations. It produces a minimum norm particular solution, a generalized inverse (of t…