Target Robust Discriminant Analysis
arXiv:1806.09463 · doi:10.1007/978-3-030-73973-7_1
Abstract
In practice, the data distribution at test time often differs, to a smaller or larger extent, from that of the original training data. Consequentially, the so-called source classifier, trained on the available labelled data, deteriorates on the test, or target, data. Domain adaptive classifiers aim to combat this problem, but typically assume some particular form of domain shift. Most are not robust to violations of domain shift assumptions and may even perform worse than their non-adaptive counterparts. We construct robust parameter estimators for discriminant analysis that guarantee performance improvements of the adaptive classifier over the non-adaptive source classifier.
10 pages, no figures, 2 tables, 2 lemma's, 1 theorem. arXiv admin note: substantial text overlap with arXiv:1706.08082 Accepted to the IAPR Joint International Workshops on Statistical + Structural and Syntactic Pattern Recognition (S+SSPR 2020). The final authenticated publication will soon be available online