WCCI Performance Prediction Challenge

The challenge is over, but a new challenge is on-going using the same datasets, check it out!

BNN-DR2

Submitted by Radford Neal

Bayesian neural network with three hidden layers. The first layer is
linear, and reduces dimensionality to ten, the following layers allow
non-linear mappings from this ten dimensional space. For this BNN-DR2
version, the second hidden layer also directly looks at some selected
features and functions of features.

Dataset Balanced Error Test guess Guess error Test score Area Under Curve
Train Valid Test Train Valid Test
ada 0.1463 0.1833 0.1748 0.1652 0.0096 0.1843 0.9303 0.902 0.9102
gina 0 0.051 0.0467 0.0646 0.0179 0.0646 1 0.9848 0.9895
hiva 0.08 0.3438 0.2958 0.2685 0.0273 0.3227 0.9806 0.7028 0.7606
nova 0 0.054 0.0514 0.1051 0.0536 0.1051 1 0.9853 0.9885
sylva 0.0042 0.0033 0.0068 0.0074 0.0006 0.0073 0.9999 0.9993 0.9991
Overall 0.0461 0.1271 0.1151 0.1222 0.0218 0.1368 0.9822 0.9148 0.9296

This entry is a complete valid challenge entry.