Abstract
Image classification by semi-supervised learning has recently become a hot spot, and the Co-Training framework is an important method of semi-supervised image classification. In the traditional Co-Training structure, the sub-networks will generate pseudo-labels for each other, and these pseudo-labels will further be used as a supervisory signal for model training. However, the pseudo-labels will hurt classification performance because of their low accuracy and unbalanced distribution. In this article, we are trying to solve the preceding two problems by designing the Balanced Module (BM) and Gaussian Mixture Module (GMM), and propose BAPS (the Balanced and Accurate Pseudo-labels for Semi-supervised image classification). In BM, the two sub-networks jointly predict the unlabeled images, then select the pseudo-labels with a high-confidence threshold to perform the balancing operation to obtain the initial samples with balanced distribution of each category. In GMM, referring to the common practice of the Learning from Noise Labels task, we use GMM to fit the loss distribution of images with pseudo-labels output by BM, then clean samples and noise samples are divided based on the observation that the loss of correctly labeled images is generally smaller than that of wrongly labeled ones. Through BM and GMM, pseudo-labels with balanced distribution and high accuracy are obtained for the subsequent model training process. Our model has achieved better classification accuracy than most state-of-the-art semi-supervised image classification algorithms on the CIFAR-10/100 and SVHN datasets, and further ablation experiments demonstrate the effectiveness of our BAPS. The source code of BAPS will be available at https://github.com/zhaojianaaa.
- [1] . 2018. MobileNetV2: Inverted residuals and linear bottlenecks. arXiv preprint arXiv:1801.04381 (2018).Google Scholar
- [2] . 2019. Squeeze-and-excitation networks. IEEE Transactions on Pattern Analysis and Machine Intelligence 42, 8 (2019), 2011–2023.Google Scholar
Digital Library
- [3] . 2017. Learning transferable architectures for scalable image recognition. arXiv preprint arXiv:1707.07012 (2017).Google Scholar
- [4] . 2017. Dual path networks. arXiv preprint arXiv:1707.01629 (2017).Google Scholar
- [5] . 2017. PolyNet: A pursuit of structural diversity in very deep networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’17). 3900–3908.Google Scholar
Cross Ref
- [6] . 2020. YOLOv4: Optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934 (2020).Google Scholar
- [7] . 2019. Libra R-CNN: Towards balanced learning for object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’19). 821–830.Google Scholar
Cross Ref
- [8] . 2017. SegNet: A deep convolutional encoder-decoder architecture for image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence 39, 12 (2017), 2481–2495.Google Scholar
- [9] . 2017. Fully convolutional instance-aware semantic segmentation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’17). 2359–2367.Google Scholar
Cross Ref
- [10] . 2015. Training deep neural networks on noisy labels with bootstrapping. In Proceedings of the International Conference on Learning Representations (ICLR’15). 1–11.Google Scholar
- [11] . 2004. Learning from imbalanced data sets with boosting and data generation: The DataBoost-IM approach. ACM SIGKDD Explorations Newsletter 6, 1 (2004), 30–39.Google Scholar
Digital Library
- [12] . 2016. Evolutionary undersampling boosting for imbalanced classification of breast cancer malignancy. Applied Soft Computing 38 (2016), 714–726.Google Scholar
Digital Library
- [13] . 2014. A resampling ensemble algorithm for classification of imbalance problems. Neurocomputing 143 (2014), 57–67.Google Scholar
Digital Library
- [14] . 2019. An anti-noise ensemble algorithm for imbalance classification. Intelligent Data Analysis 23, 6 (2019), 1205–1217.Google Scholar
Cross Ref
- [15] . 1998. Combining labeled and unlabeled data with co-training. In Proceedings of the 11th Annual Conference on Computational Learning Theory. 92–100.Google Scholar
Digital Library
- [16] . 2018. Deep Co-Training for semi-supervised image recognition. In Proceedings of the European Conference on Computer Vision (ECCV’18). 1–17.Google Scholar
Digital Library
- [17] . 2018. Tri-net for semi-supervised deep learning. In Proceedings of the 27th International Joint Conference on Artificial Intelligence (IJCAI’18). 2014–2020.Google Scholar
Digital Library
- [18] . 2021. Weighted pseudo labeled data and mutual learning for semi-supervised classification. IEEE Access 9 (2021), 36522–36534.Google Scholar
Cross Ref
- [19] . 2017. Temporal ensembling for semi-supervised learning. In Proceedings of the International Conference on Learning Representations (ICLR’17). 1–13.Google Scholar
- [20] . 2017. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. In Proceedings of the Conference on Neural Information Processing Systems (NIPS’17). 1–16.Google Scholar
- [21] . 2020. Unsupervised data augmentation for consistency training. In Proceedings of the Conference on Neural Information Processing Systems (NIPS’20). 1–19.Google Scholar
- [22] . 2020. FixMatch: Simplifying semi-supervised learning with consistency and confidence. In Proceedings of the Conference on Neural Information Processing Systems (NIPS’20). 1–21.Google Scholar
- [23] . 2018. Mixup: Beyond empirical risk minimization. In Proceedings of the International Conference on Learning Representations (ICLR’18). 1–13.Google Scholar
- [24] . 2019. MixMatch: A holistic approach to semi-supervised learning. In Proceedings of the Conference on Neural Information Processing Systems (NIPS’19). 1–14.Google Scholar
- [25] . 2020. ReMixMatch: Semi-supervised learning with distribution alignment and augmentation anchoring. In Proceedings of the International Conference on Learning Representations (ICLR’20). 1–13.Google Scholar
- [26] . 2014. Generative adversarial nets. In Advances in Neural Information Processing Systems 27, 2672-2680.Google Scholar
Digital Library
- [27] . 2016. Improved techniques for training GANs. In Proceedings of the Conference on Neural Information Processing Systems (NIPS’16). 1–11.Google Scholar
- [28] . 2016. Semi-supervised learning with generative adversarial networks. In Proceedings of the International Conference on Machine Learning (ICML’16). 1–3.Google Scholar
- [29] . 2021. Improving unsupervised image clustering with robust learning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’21). 12278–12287.Google Scholar
Cross Ref
- [30] . 2020. SELF: Learning to filter noisy labels with self-ensembling. In Proceedings of the International Conference on Learning Representations (ICLR’20). 1–15.Google Scholar
- [31] . 2019. Unsupervised label noise modeling and loss correction. In Proceedings of the International Conference on Machine Learning (ICML’19). 1–12.Google Scholar
- [32] . 2020. DIVIDEMIX: Learning with noisy labels as semi-supervised learning. In Proceedings of the International Conference on Learning Representations (ICLR’20). 1–14.Google Scholar
- [33] . 2018. MentorNet: Learning data-driven curriculum for very deep neural networks on corrupted labels. In Proceedings of the International Conference on Machine Learning (ICML’18). 1–10.Google Scholar
- [34] . 2017. Decoupling “when to update” from “how to update.” In Proceedings of the Conference on Neural Information Processing Systems (NIPS’17). 1–19.Google Scholar
- [35] . 2018. Co-teaching: Robust training of deep neural networks with extremely noisy labels. In Proceedings of the Conference on Neural Information Processing Systems (NIPS’18). 1–11.Google Scholar
- [36] . 2021. Improving calibration for long-tailed recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’21). 16489–16498.Google Scholar
Cross Ref
- [37] . 2021. CReST: A class-rebalancing self-training framework for imbalanced semi-supervised learning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’21). 10857–10866.Google Scholar
Cross Ref
- [38] 2017. A closer look at memorization in deep networks. In Proceedings of the International Conference on Machine Learning (ICML’17). 233–242.Google Scholar
- [39] . 2019. Understanding and utilizing deep neural networks trained with noisy labels. In Proceedings of the International Conference on Machine Learning (ICML’19). 1062–1070.Google Scholar
- [40] . 2009. Learning Multiple Layers of Features from Tiny Images. Technical Report. University of Toronto.Google Scholar
- [41] . 2011. Reading digits in natural images with unsupervised feature learning. In Proceedings of the NIPS Workshop on Deep Learning and Unsupervised Feature Learning.Google Scholar
- [42] . 2016. Wide residual networks. arXiv preprint arXiv:1605.07146 (2016).Google Scholar
- [43] . 2016. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR’16). 770–778.Google Scholar
Cross Ref
- [44] . 2019. RandAugment: Practical automated data augmentation with a reduced search space. arXiv preprint arXiv:1909.13719 (2019).Google Scholar
- [45] . 2021. SelfMatch: Combining contrastive self-supervision and consistency for semi-supervised learning. arXiv preprint arXiv:2101.06480 (2021).Google Scholar
- [46] . 2019. Virtual adversarial training: A regularization method for supervised and semi-supervised learning. IEEE Transactions on Pattern Analysis and Machine Intelligence 41, 8 (2019), 1979–1993.Google Scholar
Cross Ref
- [47] . 2021. CCS-GAN: A semi-supervised generative adversarial network for image classification. Visual Computer. Published online, 2021.Google Scholar
- [48] . 2020. Semi-supervised robust deep neural networks for multi-label image classification. Pattern Recognition 100 (2020), 107164.Google Scholar
Digital Library
- [49] . 2021. Attention-based label consistency for semi-supervised deep learning based image classification. Neurocomputing 453 (2021), 731–741.Google Scholar
Digital Library
Index Terms
Balanced and Accurate Pseudo-Labels for Semi-Supervised Image Classification
Recommendations
Semi-Supervised Deep Learning Using Pseudo Labels for Hyperspectral Image Classification
Deep learning has gained popularity in a variety of computer vision tasks. Recently, it has also been successfully applied for hyperspectral image classification tasks. Training deep neural networks, such as a convolutional neural network for ...
Semi-supervised multi-label classification using incomplete label information
Highlights- An inductive semi-supervised method called Smile is proposed for multi-label classification using incomplete label information.
AbstractClassifying multi-label instances using incompletely labeled instances is one of the fundamental tasks in multi-label learning. Most existing methods regard this task as supervised weak-label learning problem and assume sufficient ...
Self-labeling techniques for semi-supervised time series classification: an empirical study
An increasing amount of unlabeled time series data available render the semi-supervised paradigm a suitable approach to tackle classification problems with a reduced quantity of labeled data. Self-labeled techniques stand out from semi-supervised ...






Comments