Neural Networks for Pattern Recognition
Neural Networks for Pattern Recognition
| Hi-index | 0.00 |
What is the difference between matter and anti-matter? A. I. Insight's winning solution on the Particle Physics Task for the 2004 KDD Cup demonstrates how an accurate predictive model can be formulated without knowledge of the content of the data. Information on the data was not available for the modeling, including a description on the outcome to be predicted. In other words, an 80 x 150,000 grid of numbers with the header "Particle Physics" was all that was given to the 500+ registrants of this competition. Key steps in creating the winning model were interactive analysis of the variables, detection of interactions, a powerful self-organizing neural network, and customization of the 4 different error criteria.