Virtual Computational Chemistry Laboratory

Input data Output results Example List of key words

Training neural network with PARTITION=2 (random sets) option.

Functional scheme of partition procedure and process of neural network training. Updated from article Tetko et al., 1998.

The gray circles and left-most rectangle represent data samples (HPLC data in original article). There are 12 data samples in he initial training set, each sample is identified by its position in the rectangles. Each sample can belong only to a learning or validation set (designated by empty or black circles, respectively) for a given neural network. Because of random partitioning between the learning and validation sets, each sample on average participates an equal number of times to both these sets. After ensemble learning, results are calculated for each sample both to learning and validation sets. The leave-one-out results for the whole data set are also calculated as proposed in Tetko et al., 1995.

The trained networks can be used to predict the test set (if available) or the calculated models can be saved using MODELS=1 option and further retrieved to predict new data.

See FAQ if you have questions. How to cite this applet? Are you looking for a new job in chemoinformatics?

Copyright 2001 -- 2016 All rights reserved.