Abstract
Active learning aims to reduce the amount of labeled data while maximizing machine learning models’ performances. Currently, there is sparse research on the potential of an optimal active learning strategy. Therefore, we propose a non-myopic oracle policy that accesses the true labels of the data pool to approximate an optimal active learning strategy. We evaluate how the hyperparameters of this oracle policy influence its performance and empirically demonstrate that it is an upper baseline for common active learning strategies while being faster than a state-of-the-art oracle policy. For the sake of reproducibility, all the code related to our research is publicly available on our GitHub repository at https://github.com/ies-research/non-myopic-oracle-policy.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Benavoli, A., Corani, G., Mangili, F.: Should we really use post-hoc tests based on mean-ranks? J. Mach. Learn. Res. 17(1), 152–161 (2016)
Chapelle, O.: Active learning for Parzen window classifier. In: International Conference on Artificial Intelligence and Statistics, pp. 49–56. Bridgetown, Barbados (2005)
Chaudhuri, A., Kakde, D., Sadek, C., Gonzalez, L., Kong, S.: The mean and median criteria for kernel bandwidth selection for support vector data description. In: International Conference on Data Mining Workshops, pp. 842–849. New Orleans, LA (2017)
Gissin, D., Shalev-Shwartz, S.: Discriminative active learning. arXiv:1907.06347 (2019)
Huang, S.J., Jin, R., Zhou, Z.H.: Active learning by querying informative and representative examples. In: Advance in Neural. Information Processing System Vancouver, BC (2010)
Koshorek, O., Stanovsky, G., Zhou, Y., Srikumar, V., Berant, J.: On the limits of learning to actively learn semantic representations. In: Conference Comput. Nat. Lang. Learn. Hong Kong (2019)
Kottke, D., et al.: scikit-activeml: a library and toolbox for active learning algorithms. Preprints (2021)
Kottke, D., Krempl, G., Lang, D., Teschner, J., Spiliopoulou, M.: Multi-class probabilistic active learning. In: European Conference on Artificial Intelligence, pp. 586–594. The Hague, Netherlands (2016)
Kumar, P., Gupta, A.: Active learning query strategies for classification, regression, and clustering: a survey. J. Comput. Sci. Technol. 35(4), 913–945 (2020). https://doi.org/10.1007/s11390-020-9487-4
Lewis, D.D., Gale, W.A.: A sequential algorithm for training text classifiers. In: International Conference Research Development in Information Retrieval, pp. 3–12. Dublin, Ireland (1994)
Nguyen, V.-L., Destercke, S., Hüllermeier, E.: Epistemic uncertainty sampling. In: Kralj Novak, P., Šmuc, T., Džeroski, S. (eds.) DS 2019. LNCS (LNAI), vol. 11828, pp. 72–86. Springer, Cham (2019). https://doi.org/10.1007/978-3-030-33778-0_7
Seung, H.S., Opper, M., Sompolinsky, H.: Query by committee. In: Conference on Learnimg Theory, pp. 287–294. Pittsburgh, PA (1992)
Vanschoren, J., van Rijn, J.N., Bischl, B., Torgo, L.: OpenML: networked science in machine learning. SIGKDD Explor. 15(2), 49–60 (2013)
Zhou, Y., Renduchintala, A., Li, X., Wang, S., Mehdad, Y., Ghoshal, A.: Towards Understanding the Behaviors of Optimal Deep Active Learning Algorithms. In: International Conference on Artificial Intelligence and Statistics, pp. 1486–1494. Virtual (2021)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Sandrock, C., Herde, M., Kottke, D., Sick, B. (2023). Exploring the Potential of Optimal Active Learning via a Non-myopic Oracle Policy. In: Bifet, A., Lorena, A.C., Ribeiro, R.P., Gama, J., Abreu, P.H. (eds) Discovery Science. DS 2023. Lecture Notes in Computer Science(), vol 14276. Springer, Cham. https://doi.org/10.1007/978-3-031-45275-8_18
Download citation
DOI: https://doi.org/10.1007/978-3-031-45275-8_18
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-45274-1
Online ISBN: 978-3-031-45275-8
eBook Packages: Computer ScienceComputer Science (R0)