CS 677: Data and Information Fusion

Course Contents:

Multiple sources/sensors/modalities Fusion: Registration of data from similar sources, approaches to data fusion from similar and heterogeneous data

*Feature selection and combination methods: * Feature extraction, dimensionality reduction

Randomized generation and selection of feature sets,

Feature selection and combination approaches

Classifier Ensemble Generation, Selection and Combination approaches:

Classifier diversity, diversity versus fusion accuracy

Homogeneous and heterogeneous classifier ensembles generation, selection and fusion, Parallel and serial fusion approaches, Optimal fusion rule selection: theoretical and empirical approaches Classifier selection: local context, and classifier competence regions

Role of data distributions in classification

Context-aware clustering for estimating distributions

Optimal classifier ensemble generation based on estimated data distribution