Research Notes for Multi-Objective Bias Learning.
By
Abraham Bagherjeiran
1 Planning
2 Experiments
Experiments for thesis research.
2.1 Clustering ROC Curves
Experiment performs
clustering? on
ROC curves
Hausdorff Distance and
Matching Distance. Clusters of
ROC curves are compared to those using standard performance data. Results show
Contingency Table?s for cluster information.
2.2 Classification of ROC Curves
Experiment uses
Distance Matrix? to evaluate the
Hausdorff and
Matching Distance Function?s. Computes
Contingency Table?s for
predict? the dataset of a
ROC curve given a set of
Cross Validation folds.
See if we can use
ROC curves as signatures to investigate properties of algorithms when applied to datasets.
2.4 Regression for Optimization
From several
ROC, form a predictive model of the
ROC space and investigate its use for optimization.
3 Reports
3.1 Report on ROC Curve Analysis
Evaluation of distance functions for analyzing
ROC curves. Report is
here.
Comments
No comments for this document