Performance Analysis Of Class Rebalancing Self-Training Framework For Imbalanced Semi-Supervised Learning
Abstract
This research aims to analyze the effectiveness of
the Class-Rebalancing Self-Training (CReST) method in semisupervised
learning (SSL) on class-imbalanced data. The study
uses the CIFAR 10 long-tailed dataset to test the performance of
SSL with CReST using Python programming language on the
Google Colab platform. The results showed that CReST
effectively reduces pseudo-labels in the majority class and
increases recall in the minority class, with the best performance
achieved at Generation 16. However, there was a decrease in
Average Accuracy Recall per Class after Generation 16. The
study suggests addressing the over-sampling issue and exploring
the application of the CReST framework in other areas of
machine learning and AI.
Kata kunci— CReST, Semi-Supervised Learning, imbalance
data, pseudo label, Semi-Supervised Learning Generation
References
X. Zhu and A. B. Goldberg,
learning,= Synthesis lectures on artificial
intelligence and machine learning, vol. 3, no. 1, pp. 1-130,
T. O. Ayodele,
algorithms,= New advances in machine learning, vol. 3, pp.
-48, 2010.
M. C. du Plessis and M. Sugiyama,
learning of class balance under class-prior change
by distribution matching,= Neural Networks, vol. 50, pp.
-119, 2014.
C. Wei, K. Sohn, C. Mellina, A. L. Yuille, and F.
Yang,
Framework for Imbalanced Semi-Supervised Learning,=
CoRR, vol. abs/2102.09559, 2021, [Online]. Available:
https://arxiv.org/abs/2102.09559
in machine learning,= The Journal of Machine Learning
Research, vol. 8, pp. 1825-1832, 2007.
A. Krizhevsky, G. Hinton, and others,
multiple layers of features from tiny images,= 2009.
T. Wu, X. Wang, Q. Ye, and Y. Liu,
recognition in large-scale datasets,= Proceedings of the
European Conference on Computer Vision (ECCV), pp. 816-
, 2018.
E. Bisong and E. Bisong,
Building machine learning and deep learning models on
google cloud platform: a comprehensive guide for beginners,
pp. 59-64, 2019.
K. Sohn, R. Roelofs, B. Zoph, Q. v Le, and J. Shlens,
Consistency and Confidence,= arXiv preprint
arXiv:2001.07685, 2020.
S. Zagoruyko and N. Komodakis,
Networks,= in Proceedings of the IEEE Conference on
Computer Vision and Pattern Recognition, 2016, pp. 778-
S. Ioffe and C. Szegedy,
Accelerating deep network training by reducing internal
covariate shift,= arXiv preprint arXiv:1502.03167, 2015.
A. L. Maas, A. Y. Hannun, and A. Y. Ng,
nonlinearities improve neural network acoustic models,=
Proc. ICML, vol. 30, p. 3, 2013.
Z. Zhou, L. Chen, J. Sun, and X. Wang,
Rebalancing for Long-Tailed Recognition,= IEEE Trans
Pattern Anal Mach Intell, 2020.
X. Zhang, Z. Fang, Y. Wen, Z. Li, and Y. Qiao,
Nov. 2016, [Online]. Available:
http://arxiv.org/abs/1611.08976
..
