Bootstrapping, AdaBoosting, Uncertainty Sampling for Genre Classification of Fine Art Paintings

The project has split itself into two clear directions, each with its own challenges. One of them explores Adaboost in more detail, using different weak learners and comparing their performance on different datasets, and seeing how uncertainty-based active sampling for labeling compares with random sampling for labeling. The other explores the problem of paintings’ images genre classification, selection of features and comparision of classification methods. These problems are connected because if both subparts work out well, then we could see how uncertainty-based active adaboosting works on the paintings’ genre clasification problem.