Abstract
In this manuscript, a deep learning-based solution to the step-imbalance problem has been investigated for multi-class image annotation task where the number of training images from some of the rare classes is extremely low. Step-imbalance is a complex sub-problem of the popular class-imbalance problems where there is a steep (stair-like) disparity amongst the sample frequencies for the majority, medium and minority classes. Contrasting to the classical solutions to class- imbalance and long-tailed distribution problems, here there is a huge gap in the sample frequencies between majority and minority classes thus forming a staircase function in the overall class frequency distribution. Moreover, the pro- posed methodology focuses on a robust solution by operating under an extreme scarcity of labeled images from the minority classes (for example, below 20 images per class). Due to this, the existing neural solutions based on cost-sensitisation or generative oversampling are ineffective as they rely on the availability of sufficient minority examples in mitigating the effect of a relative and skewed ‘class-imbalance ratio’ measure. This situation is prevalent in the real-life appli- cations of computer vision, remote sensing and allied domains having severe scarcity of minority examples along with a wide gap between the major-minor class frequencies. To work under a scarce environment, an intensity-based split- ting technique has been explored to automatically extract synthetic samples for oversampling the images from the minority classes devoid of any training. In par- allel, a siamese network-based undersampling technique has been investigated for selective fusion of non-contrasting images from the majority classes. In overall, a 6–19% improvement over the existing approaches in terms of precision, recall and F1-score has been observed for the proposed technique while experimenting with CIFAR-10, Natural Images and ASCD datasets.
Similar content being viewed by others
Data Availability
The datasets are publicly available.
References
Wang M, Hua X (2011) Active learning in multimedia annotation and retrieval: A survey. ACM Trans Intell Syst Technol (TIST) 2(2):1–21
Mottaghi, R, Chen X, Liu X, Cho NG, Lee SW, Fidler S, Urtasun R, Yuille A (2014) The role of context for object detection and semantic segmentation in the wild. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 891–898
Chen H, Wang Y, Guo T, Xu C, Deng Y, Liu Z, Ma S, Xu C, Gao W (2021) Pre-trained image processing transformer. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp 12299–12310
Krizhevsky A, Sutskever I, Hinton G (2017) Imagenet classification with deep convolutional neural networks. Commun ACM 60(6):84–90
Shao L, Zhu F, Li X (2014) Transfer learning for visual categorization: A survey. IEEE Trans Neural Networks Learn Syst 26(5):1019–1034
Guo Y, Liu Y, Georgiou T, Lew M (2018) A review of semantic segmentation using deep neural networks. Intl J Multimed Inform Retriev 7:87–93
Kotsiantis S, Sotiris B, Zaharakis I, Pintelas P (2007) Supervised machine learning: A review of classification techniques. Emerg Artif Intell Appl Comp Eng 160(1):3–24
Das S, Datta S, Chaudhuri B (2018) Handling data irregularities in classification: Foundations, trends, and future challenges. Pattern Recogn 81:674–693
Johnson J, Khoshgoftaar T (2019) Survey on deep learning with class imbalance. J Big Data 6(1):1–54
Wang L, Zhang L, Qi X, Yi Z (2022) Deep attention-based imbalanced image classification. IEEE Trans Neural Networks Learn Syst 33(8):3320–3330
Buda M, Maki A, Mazurowski M (2018) A systematic study of the class imbalance problem in convolutional neural networks. Neural Netw 106:249–259
Yang Lu, Jiang He, Song Qing (1837) Jun Guo 2022 A survey on long-tailed visual recognition. Intl J Comp Vision 130(7):1872
Masko D, Hensman P (2015) The impact of imbalanced training data for convolutional neural networks
Lee H, Park M, Kim J (2016) Plankton classification on imbalanced large scale database via convolutional neural networks with transfer learning. In: 2016 IEEE international conference on image processing (ICIP), pp 3713–3717. IEEE
Chawla N, Bowyer K, Hall L, Kegelmeyer W (2002) Smote: synthetic minority over-sampling technique. J Artif Intell Res 16:321–357
Roy S, Haut J, Paoletti M, Dubey S, Plaza A (2022) Generative adversarial minor- ity oversampling for spectral–spatial hyperspectral image classification. IEEE Transactions Geoscience Remote Sensing 60:1–15
Cui Y, Jia M, Lin T, Song Y, Belongie S (2019) Class-Balanced Loss Based on Effective Number of Samples. IEEE Computer Society, Los Alamitos, CA, USA
Li B, Han Z, Li H, Fu H, Zhang C (2022) Trustworthy Long-Tailed Classification. IEEE Computer Society, Los Alamitos, CA, USA
Wu J, Song L, Zhang Q, Yang M, Yuan J (2021) Forestdet: Large-vocabulary long-tailed object detection and instance segmentation. In: IEEE Transactions on Multimedia, pp 3693–3705. IEEE
Suh MK, Seo SW (2023) Long-tailed recognition by mutual information maximization between latent features and ground-truth labels. International Conference on Machine Learning 32770–32782
Liu B, Li H, Kang H, Hua G, Vasconcelos N (2021) Gistnet: a geometric structure transfer network for long-tailed recognition. In: Proceedings of the IEEE/CVF international conference on computer vision, pp 8209–8218
Buda M, Maki A, Mazurowski MA (2018) A systematic study of the class imbalance problem in convolutional neural networks. Neural Networks (106):248–259. Elsevier
Krizhevsky A, Hinton G (2009) Learning multiple layers of features from tiny images. Toronto, ON, Canada
Roy P, Ghosh S, Bhattacharya S, Pal U (2018) Effects of degradations on deep neural network architectures ar**v preprint ar**v:1807.10108
Kalita I, Roy M (2020) Deep neural network-based heterogeneous domain adaptation using ensemble decision making in land cover classification. IEEE Trans Artif Intell 1(2):167–180
Wang S, Liu W, Wu J, Cao L, Meng Q, Kennedy PJ (2016) Training deep neural networks on imbalanced data sets. In: 2016 international joint conference on neural networks (IJCNN), 4368–4374. IEEE
Lin T, Goyal P, Girshick R, He K (2020) Focal loss for dense object detection. IEEE Trans Pattern Anal Mach Intell 42(2):318–327
Ding W, Huang DY, Chen Z, Yu X, Lin W (2017) Facial action recognition using very deep networks for highly imbalanced class distribution. In: 2017 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC), pp 1368–1372. IEEE
Wang H, Cui Z, Chen Y, Avidan M, Abdallah A, Kronzer A (2018) Predicting hospital readmission via cost-sensitive deep learning. IEEE/ACM Trans Comput Biol Bioinf 15(6):1968–1978
Khan S, Hayat M, Bennamoun M, Sohel F, Togneri R (2018) Cost-sensitive learning of deep feature representations from imbalanced data. IEEE Trans Neural Netw Learn Syst 29(8):3573–3587
Pouyanfar S, Tao Y, Mohan A, Tian H, Kaseb AS, Gauen K, Dailey R, Aghajanzadeh S, Lu YH, Chen SC (2018) Dynamic sampling in convolutional neural networks for imbalanced data classification. In: 2018 IEEE conference on multimedia information processing and retrieval (MIPR), pp 112–117
Huang C, Li Y, Loy CC, Tang X (2016) Learning deep representation for imbalanced classification. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 5375–5384
Dong Q, Gong S, Zhu X (2018) Imbalanced deep learning by minority class incremental rectification. IEEE Trans Pattern Anal Mach Intell 41(6):1367–1381
Zhao L, Shang Z, Tan J, Zhou M, Zhang M, Gu D, Zhang T, Tang Y (2022) Siamese networks with an online reweighted example for imbalanced data learning. Pattern Recogn 132:108947
Kaelbling LP, Littman ML, Moore AW (1996) Reinforcement learning: A survey. J Artif Intell Res 237–285
Goodfellow I, Bengio Y, Courville A (2016) Deep learning. MIT press
Zhang Y, Shuai L, Ren Y, Chen H (2018) Image classification with category centers in class imbalance situation. In: 2018 33rd youth academic annual conference of Chinese association of automation (YAC), pp 359–363. IEEE
Ando S, Huang CY (2017) Deep over-sampling framework for classifying imbalanced data. Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2017, Skopje, Macedonia, Proceedings, Part I 10. Springer, pp 770–785
Wang Y, Yao Q, Kwok J, Ni L (2019) Few-shot learning: A survey. CoRR
Chawla NV, Bowyer KW, Hall LO, Kegelmeyer WP (2002) SMOTE: synthetic minority over-sampling technique. J Artif Intell Res 16: 321–357
Khairy M, Mahmoud TM, Abd-El-Hafeez T (2024) The effect of rebalancing techniques on the classification performance in cyberbullying datasets. Neural Comp Appl (36):1049–1065. Springer
Eliwa EHI, El Koshiry AM, Abd El-Hafeez T, Farghaly HM (2023) Utilizing convolutional neural networks to classify monkeypox skin lesions. Sci Rep 13. Nature Publishing Group, UK, London
Chicco D (2021) Siamese Neural Networks: An Overview. Springer, US, New York, NY
Parisot S (2022) Esperan¸ca, Pedro, M, McDonagh, Steven, Madarasz, Tamas, J, Yang, Li Y, and Henguo Z. Long-tail Recognition via Compositional Knowledge Transfer. IEEE Computer Society, Los Alamitos, CA, USA
Funding
This work is supported with the project grant (with No. TDP/DR- ISHTI CPS/L2M/SL/2023/0007) from IITI DRISHTI CPS Foundation under the aegis of National Mission on Interdisciplinary Cyber-Physical System (NMICPS), Department of Science and Technology, Government of India.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Ethical Approval
Not applicable.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Below is the link to the electronic supplementary material.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Vemulapalli, V.M., Chakraborty, S. & Korra, S.B. An intensity-based deep approach to mitigate step-imbalance problem under extreme paucity of images from rare classes. Multimed Tools Appl (2024). https://doi.org/10.1007/s11042-024-19303-8
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1007/s11042-024-19303-8