Extreme Value Theory for Open Set Classification -- GPD and GEV Classifiers
This addresses the challenge of robust classification in dynamic environments for machine learning applications, but it is incremental as it builds on existing extreme value theory methods.
The paper tackles the problem of open set classification where unknown classes may appear after training, showing that the extreme value machine can fail when known and unknown class geometries differ, and proposes two new algorithms based on extreme value theory that demonstrate effectiveness on LETTER and MNIST datasets.
Classification tasks usually assume that all possible classes are present during the training phase. This is restrictive if the algorithm is used over a long time and possibly encounters samples from unknown classes. The recently introduced extreme value machine, a classifier motivated by extreme value theory, addresses this problem and achieves competitive performance in specific cases. We show that this algorithm can fail when the geometries of known and unknown classes differ. To overcome this problem, we propose two new algorithms relying on approximations from extreme value theory. We show the effectiveness of our classifiers in simulations and on the LETTER and MNIST data sets.