Abstract.

In this paper we consider the task of balancing a training sample in the training of an image classifier based on a convolutional neural network. Аn active scheme of selection (thinning out) of informative elements in the learning process is proposed to teach the lightweight classifier in the conditions of a strong imbalance of the training samples, An experimental verification using the example of the problem of classifying images of handwritten figures and zones of road sign detection shows a stable advantage of the proposed scheme in comparison with random selection.

Keywords:

pattern recognition, convolutional neural networks, sampling, active learning, curriculum learning.

PP. 199-204.

DOI: 10.14357/20790279180523

References

1. He H., Garcia E.A. Learning from imbalanced data // IEEE Transactions on Knowledge and Data Engineering – 2009. – V. 21 – № 9 – P.1263–1284.
2. Buda M., Maki A., Mazurowski M.A. A systematic study of the class imbalance problem in convolutional neural networks / – 2017. – P.1–23.
3. Settles B. Active learning literature survey // Madison.–University of Wisconsin. –2010. – 11 p.
4. Dasgupta S. Analysis of a greedy active learning strategy // Advances in Neural Information Procesing Systems.–Saul.–2014.
5. Elman J.L. Learning and development in neural networks: the importance of starting small // Cognition.–V.48.–№1.–P.71– 99.–1993.
6. Bengio Y., Louradour J., Collobert R. e.a. Curriculum learning // Proceedings of the 26th Annual International Conference on Machine Learning, ser. ICML ’09, Montreal, Quebec, Canada.–ACM.–2009.–P. 41–48.
7. Huang C., Li Y., Loy C.C., Tang X. Learning Deep Representation for Imbalanced Classification // 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, -2016. P. 5375-5384.
8. Yu H., Zhang Z., Qin Z., Wu H., Li D., Zhao J., Lu X. Loss Rank Mining: A General Hard Example Mining Method for Real-time Detectors // – 2018. Available at: https://arxiv.org/abs/1804.04606 (accessed July 6, 2018).
9. LeCun Y., Bottou L., Bengio Y., e.a. Gradientbased learning applied to document recognition // Proc. IEEE.–86(11).–P. 2278–2324.–1998.
10. Prun V.E., Postnikov V.V., Sadekov R.N., Sholomov D.L. “Development of Active Safety Software of Road Freight Transport, Aimed at Improving Inter-City Road Safety, Based on Stereo Vision Technologies and Road Scene Analysis” // Proceedings of the Scientific-Practical Conference “Research and Development – 2016”, Springer, Cham, pp.209-218. – 2017. ISBN 978-3-319-62869-1
11. Prun V., Polevoy D., Postnikov V. Forward rectification: spatial image normalization for a video from a forward facing vehicle camera // Proc. SPIE 10341, Ninth International Conference on Machine Vision (ICMV 2016), T. 10341, C. 103410W