Dataset Concordia (targetcl. 6)

Basic characteristics Concordia (targetcl. 6)

400

target objects

The Concordia CENPARMI handwritten digits (16x16 pixels per image). Used in "Neural-Network Classifiers for Recognizing Totally Unconstrained Handwritten Numerals" by Cho, Sung-Bae. Download mat-file with Prtools dataset.

3600

outlier objects

256

features

Unsupervised PCA Concordia (targetcl. 6)

On the left, the PCA scatterplot is shown, on the right the retained variance for varying number of features.
On the left, the PCA scatterplot is shown of data rescaled to unit variance, on the right the retained variance.

Supervised Fisher Concordia (targetcl. 6)

On the left, the Fisher scatterplot is shown, on the right the ROC curve along this direction.

Results Concordia (targetcl. 6)

The experiments are performed using dd_tools. A rudimentary explanation of the classifiers is given in the classifier section.

536, 0 outliers, AUC (x100) 5x strat. 10-fold
Classifiers Preproc
none unit var PCA 95\%
Gauss 94.0 (0.0) 93.2 (0.0) 94.9 (0.0)
Min.Cov.Determinant NaN (0.0) NaN (0.0) NaN (0.0)
Mixture of Gaussians 96.0 (1.3) 93.8 (0.7) 96.0 (1.1)
Naive Parzen 88.8 (0.0) 88.8 (0.0) 94.7 (0.0)
Parzen 96.1 (0.0) 94.1 (0.0) 93.7 (0.0)
k-means 95.5 (0.6) 93.3 (0.1) 92.0 (0.8)
1-Nearest Neighbors 96.1 (0.0) 94.0 (0.0) 93.7 (0.0)
k-Nearest Neighbors 96.1 (0.0) 94.0 (0.0) 93.7 (0.0)
Nearest-neighbor dist 91.7 (0.0) 82.0 (0.0) 87.1 (0.0)
Principal comp. 95.1 (0.0) 94.2 (0.0) 93.6 (0.0)
Self-Organ. Map 96.7 (0.3) 93.9 (0.2) 94.2 (0.1)
Auto-enc network NaN (0.0) NaN (0.0) 89.1 (0.1)
MST 96.1 (0.0) 94.0 (0.0) 94.6 (0.0)
L_1-ball 66.7 (0.0) 66.7 (0.0) 89.9 (0.0)
k-center 91.6 (2.8) 90.3 (0.9) 88.1 (3.8)
Support vector DD 96.0 (0.0) 55.8 (0.4) 93.9 (0.0)
Minimax Prob. DD 96.2 (0.0) 93.9 (0.0) 93.9 (0.0)
LinProg DD 95.0 (0.0) 92.7 (0.0) 90.2 (0.0)
Lof DD 97.4 (0.0) 95.2 (0.0) 93.6 (0.0)
Lof range DD 98.3 (0.0) 96.3 (0.0) 95.5 (0.0)
Loci DD 92.8 (0.0) 90.6 (0.0) 87.7 (0.0)

Classifier projection spaces The first classifier projection spaces are obtained by computing the classifier label disagreements (setting the threshold on 10% target error) and applying an MDS on the resulting distance matrix between classifiers:



Original



Unit variance



PCA mapped

Classifier projection spaces The second versions of the classifier projection spaces are obtained by computing the classifier ranking disagreements and applying an MDS on the resulting distance matrix between classifiers:



Original



Unit variance



PCA mapped