Accepted version, 375.3Kb
This file wil be publicly accessible from 2021-11-28
Request a copy
Request a copy
Wang, H., Gouk, H., Frank, E., Pfahringer, B., & Mayo, M. (2020). A comparison of machine learning methods for cross-domain few-shot learning. In M. Gallagher, N. Moustafa, & E. Lakshika (Eds.), AI 2020: Advances in Artificial Intelligence. AI 2020. Lecture Notes in Computer Science (Vol. LNAI 12576, pp. 445–457). Cham, Switzerland: Springer. https://doi.org/10.1007/978-3-030-64984-5_35
Permanent Research Commons link: https://hdl.handle.net/10289/14027
We present an empirical evaluation of machine learning algorithms in cross-domain few-shot learning based on a fixed pre-trained feature extractor. Experiments were performed in five target domains (CropDisease, EuroSAT, Food101, ISIC and ChestX) and using two feature extractors: a ResNet10 model trained on a subset of ImageNet known as miniImageNet and a ResNet152 model trained on the ILSVRC 2012 subset of ImageNet. Commonly used machine learning algorithms including logistic regression, support vector machines, random forests, nearest neighbour classification, naïve Bayes, and linear and quadratic discriminant analysis were evaluated on the extracted feature vectors. We also evaluated classification accuracy when subjecting the feature vectors to normalisation using p-norms. Algorithms originally developed for the classification of gene expression data—the nearest shrunken centroid algorithm and LDA ensembles obtained with random projections—were also included in the experiments, in addition to a cosine similarity classifier that has recently proved popular in few-shot learning. The results enable us to identify algorithms, normalisation methods and pre-trained feature extractors that perform well in cross-domain few-shot learning. We show that the cosine similarity classifier and ℓ² -regularised 1-vs-rest logistic regression are generally the best-performing algorithms. We also show that algorithms such as LDA yield consistently higher accuracy when applied to ℓ² -normalised feature vectors. In addition, all classifiers generally perform better when extracting feature vectors using the ResNet152 model instead of the ResNet10 model.
© 2020 Springer Nature Switzerland AG. This is the author's accepted version. The final publication is available at Springer via dx.doi.org/10.1007/978-3-030-64984-5_35