Exemplary Results of Producing Valid Approximations of Grain-size Data Using Constrained ANN

Published: 22 November 2017| Version 1 | DOI: 10.17632/cwpjkxcwcs.1
Tim Berthold


The given files describe approximations for grain-size data produced by an constrained feed-forward network. The given data serve as supplementary material for a publication with title 'Valid Approximation of Spatially Distributed Grain Size Distributions -- A Priori Information Encoded to a Feedforward Network'. The files ending with 'approximation' contain (discretized) approximations produced by the given models (e.g. 1-5-1-c_approximation.txt). For the 3-15-5-1-c model, there are approximations of the 5 marked samples as described in the paper. The files starting with 3-5-1-c_weight_evolution contain exemplary evolution of the weights in the second layer of the 1-5-1-c model during the training process. The files are ascii files, which contain two columns separated by a tab.



Artificial Neural Networks, Approximation of Function