demo_deciders_and_grags.m File Reference

Detailed Description

Different ways to use per-spectrum prediction aggregation (grag), and decision thresholds (decider)

Generates 4 situations: 2 datasets x 2 analyses.

Datasets (she5trays.mat)

1. Chemicals (class level 1)
2. Non-transformed/transformed (class level 2)

Analyses

1. Classifier output is post-processed through decider -> grag_classes_vote
2. Classifier output is post-processed through grag_classes_mean -> decider

In the first case, the decider selects which spectra will be allowed to vote, whereas in the second case, the posterior probabilities (supports) per spectrum are averaged before the decider accepts or rejects a single per-group prediction.

See also

gridsearch, decider, grag_mean, grag_classes_vote

Definition in file demo_deciders_and_grags.m.

Go to the source code of this file.