4 papers
Efficient and Scalable Kernel Matrix Approximations using Hierarchical Decomposition
Keerthi Gaddameedi, Severin Reiz, Tobias Neckel +1
With the emergence of Artificial Intelligence, numerical algorithms are moving towards more approximate approaches. For methods such as PCA or diffusion maps, it is necessary to co…
Neural Nets with a Newton Conjugate Gradient Method on Multiple GPUs
Severin Reiz, Tobias Neckel, Hans-Joachim Bungartz
Training deep neural networks consumes increasing computational resource shares in many compute centers. Often, a brute force approach to obtain hyperparameter values is employed.…
Fast Approximation of the Gauss-Newton Hessian Matrix for the Multilayer Perceptron
Chao Chen, Severin Reiz, Chenhan Yu +2
We introduce a fast algorithm for entry-wise evaluation of the Gauss-Newton Hessian (GNH) matrix for the fully-connected feed-forward neural network. The algorithm has a precomputa…
Geometry-Oblivious FMM for Compressing Dense SPD Matrices
Chenhan D. Yu, James Levitt, Severin Reiz +1
We present GOFMM (geometry-oblivious FMM), a novel method that creates a hierarchical low-rank approximation, "compression," of an arbitrary dense symmetric positive definite (SPD)…