Cited by 3,088Imbalanced Data Classification TechniquesAdvanced Neural Network ApplicationsMachine Learning and Data Classification

The DOI matches its Crossref record; 77 of the 83 references it cites were found in a registry or an index.

A systematic study of the class imbalance problem in convolutional neural networks

Open access

Buda, Mateusz, Maki, Atsuto, Mazurowski, Maciej A.·2018·Neural Networks

DOI 10.1016/j.neunet.2018.07.011

Open this paper in CoChat — free account Open PDF

Abstract

In this study, we systematically investigate the impact of class imbalance on classification performance of convolutional neural networks (CNNs) and compare frequently used methods to address the issue. Class imbalance is a common problem that has been comprehensively studied in classical machine learning, yet very limited systematic research is available in the context of deep learning. In our study, we use three benchmark datasets of increasing complexity, MNIST, CIFAR-10 and ImageNet, to investigate the effects of imbalance on classification and perform an extensive comparison of several methods to address the issue: oversampling, undersampling, two-phase training, and thresholding that compensates for prior class probabilities. Our main evaluation metric is area under the receiver operating characteristic curve (ROC AUC) adjusted to multi-class tasks since overall accuracy metric is associated with notable difficulties in the context of imbalanced data. Based on results from our experiments we conclude that (i) the effect of class imbalance on classification performance is detrimental; (ii) the method of addressing class imbalance that emerged as dominant in almost all analyzed scenarios was oversampling; (iii) oversampling should be applied to the level that completely eliminates the imbalance, whereas the optimal undersampling ratio depends on the extent of imbalance; (iv) as opposed to some classical machine learning models, oversampling does not cause overfitting of CNNs; (v) thresholding should be applied to compensate for prior class probabilities when overall number of properly classified cases is of interest.

Integrity signals

Registry

Crossref · 10.1016/j.neunet.2018.07.011

Cites

77 of 77 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 1×
“And for all the cases, the classes chosen to be minority were the ones with the lowest label value.”
Background
Very Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 1×
“A standard CNN architecture comprises fully connected layers and a number of blocks consisting of convolution kernels, activation function layer and max pooling [22, 23, 24].”
Method
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
Scikit-learn: Machine Learning in Python
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Müller, A., Nothman, J., Louppe, G., et al. (2012) · arXiv · doi:10.48550/arxiv.1201.0490 · Cited by 71,016 · cited 1×
“We used a speciﬁc implementation of the ROC AUC available in scikit-learn python package [65].”
Method
Gradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 2×
“Images The CNN model that we use for MNIST is the modern version of LeNet-5 [52].”
Method
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380 · cited 1×
“In our study, we used three benchmark datasets: MNIST [52], CIFAR-10 [53] and ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2012 [54].”
Method
SMOTE synthetic minority over-sampling technique
Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002) · Journal of Artificial Intelligence Research · doi:10.1613/jair.953 · Cited by 32,865 · cited 3×
“For classical machine learning models it was shown that oversampling can cause overﬁtting, especially for minority classes [33].”
BackgroundMethod
Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
He, K., Zhang, X., Ren, S., & Sun, J. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.123 · Cited by 19,741 · cited 1×
“Network weights were initialized with Kaiming (also known as MSRA) initialization procedure [63].”
Method
Visualizing and understanding convolutional networks
Zeiler, M. D., & Fergus, R. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10590-1_53 · Cited by 15,839 · cited 1×
“This property allows CNNs to learn hierarchical representations [2].”
Background
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate ShiftPreprint
Ioffe, S. (2024) · arXiv (Cornell University) · doi:10.57702/o9raffed · Cited by 15,273

Show all 83 references

Sorted by citations

Items 201–300 of 1,001

A Survey on Fuzzy Deep Neural Networks
Das, R., Sen, S., & Maulik, U. (2020) · ACM Computing Surveys · doi:10.1145/3369798 · Cited by 98
“Furthermore, a lot of problems arise from low sample size, result from noisy samples or heterogeneous samples, or results from severe class imbalance issues [Buda et al. 2018].”
Background
Early prediction of neoadjuvant chemotherapy response for advanced breast cancer using PET/MRI image deep learning
Choi, J. H., Kim, H., Kim, W., Lim, I., Lee, I., Byun, B. H., Noh, W. C., Seong, M.-K., Lee, S., Kim, B. I., et al. (2020) · Scientific Reports · doi:10.1038/s41598-020-77875-5 · Cited by 100
Imbalanced Continual Learning with Partitioning Reservoir Sampling
Kim, C. D., Jeong, J., & Kim, G. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58601-0_25 · Cited by 101
“Data processing methods such as oversampling or undersampling [5,8,65] explicitly simulate the input balance, while cost-sensitive approaches [38,85,18] adjust the update via regularizing the objective.”
Method
Detection and Classification of Knee Osteoarthritis
Cueva, J. H., Castillo, D., Espinós-Morató, H., Durán, D., Díaz, P., & Lakshminarayanan, V. (2022) · Diagnostics · doi:10.3390/diagnostics12102362 · Cited by 98
“The effect of class imbalance on classification is detrimental to the performance of the model [25].”
Background
Trustworthy Long-Tailed Classification
Li, B., Han, Z., Li, H., Fu, H., & Zhang, C. (2022) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52688.2022.00684 · Cited by 101
“This also prevents the ensembling methods from perceiving the wrong predictions and OOD samples, and is especially harmful for the tail classes where the predictions have averagely more errors than the head classes [5, 18].”
BackgroundMethod
Long-Tailed Multi-Label Visual Recognition by Collaborative Training on Uniform and Re-balanced Samplings
Guo, H., & Wang, S. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.01484 · Cited by 97
“Re-balanced sampling [4, 32, 1, 10, 53] and cost sensitive re-weighting [3, 6, 13, 43, 29, 19, 36] are the two typical kinds of data re-balancing methods.”
BackgroundMethod
Sequential deep learning from NTSB reports for aviation safety prognosis
Zhang, X., Srinivasan, P., & Mahadevan, S. (2021) · Safety Science · doi:10.1016/j.ssci.2021.105390 · Cited by 103
Fully automated carbonate petrography using deep convolutional neural networks
Koeshidayatullah, A., Morsilli, M., Lehrmann, D. J., Al-Ramadan, K. A., & Payne, J. L. (2020) · Marine and Petroleum Geology · doi:10.1016/j.marpetgeo.2020.104687 · Cited by 100
Multi-Objective Particle Swarm Optimization Based Preprocessing of Multi-Class Extremely Imbalanced Datasets
Priya, R. D., Sivaraj, R., Abraham, A., Pravin, T., Sivasankar, P., & Anitha, N. (2022) · International Journal of Uncertainty Fuzziness and Knowledge-Based Systems · doi:10.1142/s0218488522500209 · Cited by 95
A real-time electrical load forecasting and unsupervised anomaly detection framework
Wang, X., Yao, Z., & Papaefthymiou, M. C. (2022) · Applied Energy · doi:10.1016/j.apenergy.2022.120279 · Cited by 95
A review on emotion detection by using deep learning techniques
Chutia, T., & Baruah, N. (2024) · Artificial Intelligence Review · doi:10.1007/s10462-024-10831-1 · Cited by 125
Transfer learning in agriculture: a review
Hossen, M. I., Awrangjeb, M., Pan, S., & Mamun, A. A. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-024-11081-x · Cited by 126
“…typically trained with balanced data distribution whereas, in the existing works, most of the datasets exhibit class imbalance, which can lead to over-fitting problems primarily affecting small classes and restricting the generalisation performance of the models (Buda et al 2018; Yan and Han 2018).”
Learning to Predict Visual Attributes in the Wild
Pham, K., Kafle, K., Lin, Z., Ding, Z., Cohen, S. D., Tran, Q., & Shrivastava, A. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.01282 · Cited by 99
“These works can be divided into two main approaches: cost-sensitive learning [26, 67, 14] and resampling [55, 22, 9, 20, 47].”
Background
Toward Open-World Electroencephalogram Decoding Via Deep Learning: A comprehensive survey
Chen, X., Li, C., Liu, A., McKeown, M. J., Qian, R., & Wang, Z. J. (2022) · IEEE Signal Processing Magazine · doi:10.1109/msp.2021.3134629 · Cited by 99
“Methods for addressing class imbalance problem based on deep learning can be divided into two main categories [81].”
BackgroundMethodInfluential
Thinking like a naturalist: Enhancing computer vision of citizen science images by harnessing contextual data
Terry, J. C. D., Roy, H. E., & August, T. A. (2020) · Methods in Ecology and Evolution · doi:10.1111/2041-210x.13335 · Cited by 92
“Our approach to augmenting metadata (adding Gaussian noise to each element) was relatively basic and more targeted approaches to generating additional synthetic training data (Chawla et al. 2002) could lead to better results.”
Background
XGBoost in handling missing values for life insurance risk prediction
Rusdah, D. A., & Murfi, H. (2020) · SN Applied Sciences · doi:10.1007/s42452-020-3128-y · Cited by 92
Handling class imbalance in COVID-19 chest X-ray images classification: Using SMOTE and weighted loss
Chamseddine, E., Mansouri, N., Soui, M., & Abed, M. (2022) · Applied Soft Computing · doi:10.1016/j.asoc.2022.109588 · Cited by 112
“It has been demonstrated to be efficient, although it can result in overfitting problems [35,37].”
Influential
Discriminative kernel convolution network for multi-label ophthalmic disease detection on imbalanced fundus image dataset
Bhati, A., Gour, N., Khanna, P., & Ojha, A. (2023) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2022.106519 · Cited by 92
Deep learning guided image-based droplet sorting for on-demand selection and analysis of single cells and 3D cell cultures.
Anagnostidis, V., Sherlock, B. E., Metz, J., Mair, P., Hollfelder, F., & Gielen, F. (2020) · Figshare · doi:10.17863/cam.63685 · Cited by 91
Crop type mapping by using transfer learning
Nowakowski, A., Mrziglod, J., Spiller, D., Bonifácio, R., Ferrari, I., Mathieu, P.-P., García–Herranz, M., & Kim, D.-H. (2021) · International Journal of Applied Earth Observation and Geoinformation · doi:10.1016/j.jag.2021.102313 · Cited by 98
Comparison of BERT implementations for natural language processing of narrative medical documents
Turchin, A., Masharsky, S., & Žitnik, M. (2022) · Informatics in Medicine Unlocked · doi:10.1016/j.imu.2022.101139 · Cited by 107
A New Approach for Detecting Fundus Lesions Using Image Processing and Deep Neural Network Architecture Based on YOLO Model
Santos, C., de Aguiar, M. S., Welfer, D., & Belloni, B. (2022) · Sensors · doi:10.3390/s22176441 · Cited by 97
Classification of flower species by using features extracted from the intersection of feature selection methods in convolutional neural network models
Toğaçar, M., Ergen, B., & Cömert, Z. (2020) · Measurement · doi:10.1016/j.measurement.2020.107703 · Cited by 89
Multi-Hazard and Spatial Transferability of a CNN for Automated Building Damage Assessment
Valentijn, T., Margutti, J., den Homberg, M. J. C. V., & Laaksonen, J. T. (2020) · Remote Sensing · doi:10.3390/rs12172839 · Cited by 100
“To limit the decrease in performance due to the imbalance, two methods commonly used to combat class imbalance were implemented, namely resampling and cost-sensitive learning [65].”
Method
Balanced knowledge distillation for long-tailed learning
Zhang, S., Chen, C., Hu, X., & Peng, S. (2023) · Neurocomputing · doi:10.1016/j.neucom.2023.01.063 · Cited by 97
“Re-sampling methods [4, 3, 44, 6, 13] created a roughly balanced distribution by either over-sampling or under-sampling.”
BackgroundMethod
Adversarial Robustness under Long-Tailed Distribution
Wu, T., Liu, Z., Huang, Q., Wang, Y., & Lin, D. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.00855 · Cited by 92
“To tackle the long-tailed recognition problem, traditional re-balancing approaches include re-sampling [5, 15, 12, 36, 1] and re-weighting [7, 1].”
Method
Exploring deep neural networks via layer-peeled model: Minority collapse in imbalanced training
Fang, C., He, H., Long, Q., & Su, W. (2021) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2103091118 · Cited by 125
Adaptive cost-sensitive learning: Improving the convergence of intelligent diagnosis models under imbalanced data
Ren, Z., Zhu, Y., Kang, W., Fu, H. C., Niu, Q., Gao, D., Yan, K., & Hong, J. (2022) · Knowledge-Based Systems · doi:10.1016/j.knosys.2022.108296 · Cited by 93
A coarse-to-fine weakly supervised learning method for green plastic cover segmentation using high-resolution remote sensing images
Cao, Y., & Huang, X. (2022) · ISPRS Journal of Photogrammetry and Remote Sensing · doi:10.1016/j.isprsjprs.2022.04.012 · Cited by 91
Metabolic Syndrome and Development of Diabetes Mellitus: Predictive Modeling Based on Machine Learning Techniques
Perveen, S., Shahbaz, M., Keshavjee, K., & Guergachi, A. (2018) · IEEE Access · doi:10.1109/access.2018.2884249 · Cited by 87
What makes multi-class imbalanced problems difficult? An experimental study
Lango, M., & Stefanowski, J. (2022) · Expert Systems with Applications · doi:10.1016/j.eswa.2022.116962 · Cited by 87
FTGAN: A Novel GAN-Based Data Augmentation Method Coupled Time–Frequency Domain for Imbalanced Bearing Fault Diagnosis
Wang, H., Li, P., Lang, X., Tao, D., Ma, J., & Li, X. (2023) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2023.3234095 · Cited by 87
“According to the deﬁnition of an imbalanced dataset proposed in [16], the different imbalanced ratios of fault data are set artiﬁcially.”
Background
Convolutional neural networks based focal loss for class imbalance problem: a case study of canine red blood cells morphology classification
Pasupa, K., Vatathanavaro, S., & Tungjitnob, S. (2020) · Journal of Ambient Intelligence and Humanized Computing · doi:10.1007/s12652-020-01773-x · Cited by 91
Machine learning to detect marine animals in UAV imagery: effect of morphology, spacing, behaviour and habitat
Dujon, A. M., Ierodiaconou, D. A., Geeson, J. J., Arnould, J. P. Y., Allan, B. M., Katselidis, K. A., & Schofield, G. (2021) · Remote Sensing in Ecology and Conservation · doi:10.1002/rse2.205 · Cited by 96
An empirical evaluation of sampling methods for the classification of imbalanced data
Kim, M., & Hwang, K. (2022) · PLoS ONE · doi:10.1371/journal.pone.0271260 · Cited by 87
“Other studies evaluated the sampling method using linear discriminant analysis (LDA) [35, 36], random forests (RFs) [37], and convolutional neural networks [34].”
Method
Detecting Large Vessel Occlusion at Multiphase CT Angiography by Using a Deep Convolutional Neural Network
Stib, M. T., Vasquez, J., Dong, M. P., Kim, Y. H., Subzwari, S. S., Triedman, H., Wang, A., Wang, H.-L. C., Yao, A. D., Jayaraman, M. V., et al. (2020) · Radiology · doi:10.1148/radiol.2020200334 · Cited by 88
“For the LVO-negative cases, we used the undersampling technique to alleviate the class imbalance problem common to real-world machine learning problems (20).”
Method
A convolutional neural network to filter artifacts in spectroscopic MRI
Gurbani, S. S., Schreibmann, E., Maudsley, A. A., Cordova, J. S., Soher, B. J., Poptani, H., Verma, G., Barker, P. B., Shim, H., & Cooper, L. A. D. (2018) · Magnetic Resonance in Medicine · doi:10.1002/mrm.27166 · Cited by 84
“6, whereas in other domains in which deep learning is been applied the imbalance can be several orders of magnitude.(33) Based on the results of GradCAM, the user can glean some insight into how the network arrives at a decision.”
Method
Blind Reverberation Time Estimation Using a Convolutional Neural Network
Gamper, H., & Tashev, I. (2018) · International Workshop on Acoustic Signal Enhancement · doi:10.1109/iwaenc.2018.8521241 · Cited by 84
An Imbalanced Big Data Mining Framework for Improving Optimization Algorithms Performance
Hassib, E. M., El-Desouky, A. I., El‐kenawy, E. M., & Elghamrawy, S. M. (2019) · IEEE Access · doi:10.1109/access.2019.2955983 · Cited by 85
“So, a preprocessing phase is needed for solving the problem of imbalanced classes in such datasets before continuing toward the classification phase [18].”
Learning from Imbalanced Data: Integration of Advanced Resampling Techniques and Machine Learning Models for Enhanced Cancer Diagnosis and Prognosis
Gürcan, F., & Soylu, A. (2024) · Cancers · doi:10.3390/cancers16193417 · Cited by 84
“This diverse set of metrics ensures a thorough evaluation of model performance, addressing various aspects crucial for effective cancer diagnosis and prognosis [9,50,59,60].”
BackgroundMethod
Self-supervised Learning is More Robust to Dataset Imbalance
Liu, H., HaoChen, J. Z., Gaidon, A., & Ma, T. (2021) · arXiv · doi:10.48550/arxiv.2110.05025 · Cited by 84
“The performance of vanilla supervised methods degrades significantly on class-imbalanced datasets [Cui et al., 2019, Cao et al., 2019, Buda et al., 2018], posing challenges to practical applications such as instance segmentation [Tang et al., 2020] and depth estimation [Yang et al., 2021].”
BackgroundMethod
Seizure Detection Using Least Eeg Channels by Deep Convolutional Neural Network
Avcu, M. T., Zhang, Z., & Chan, D. W. S. (2019) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2019.8683229 · Cited by 84
“A common problem in CNN networks for seizure detection is that datasets are often imbalanced meaning interictal phases outnumber the ictal phases by a wide margin and it has been shown that imbalanced datasets lead to statistically signiﬁcant performance drop in CNN architectures [16].”
Background
Deep convolutional neural networks for multi-scale time-series classification and application to tokamak disruption prediction using raw, high temporal resolution diagnostic data
Churchill, R. M., Tobias, B., Zhu, Y., & DIII-D team (2020) · Physics of Plasmas · doi:10.1063/1.5144458 · Cited by 88
Deep learning a boon for biophotonics?
Pradhan, P., Guo, S., Ryabchykov, O., Popp, J., & Bocklitz, T. W. (2020) · Journal of Biophotonics · doi:10.1002/jbio.201960186 · Cited by 86
Resampling and data augmentation for short-term PV output prediction based on an imbalanced sky images dataset using convolutional neural networks
Nie, Y., Zamzam, A. S., & Brandt, A. R. (2021) · Solar Energy · doi:10.1016/j.solener.2021.05.095 · Cited by 90
A Novel Deep Class-Imbalanced Semisupervised Model for Wind Turbine Blade Icing Detection
Cheng, X., Shi, F., Liu, X., Zhao, M. Y., & Chen, S. (2021) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2021.3102514 · Cited by 87
Multi-site, Multi-domain Airway Tree Modeling
Zhang, M., Wu, Y., Zhang, H., Qin, Y., Zheng, H., Tang, W., Arnold, C. W., Pei, C., Yu, P., Yang, N., et al. (2023) · Medical Image Analysis · doi:10.1016/j.media.2023.102957 · Cited by 83
Mitigating Bias in Radiology Machine Learning: 2. Model Development
Zhang, K., Khosravi, B., Vahdati, S., Faghani, S., Nugen, F., Rassoulinejad-Mousavi, S. M., Moassefi, M., Jagtap, J. M. M., Singh, Y., Rouzrokh, P., et al. (2022) · Radiology Artificial Intelligence · doi:10.1148/ryai.220010 · Cited by 87
“Oversampling and under-sampling will prevent the model from seeing substantially more instances of one class over others during training, therefore reducing the risk of overfitting (17).”
Background
An ensemble imbalanced classification method based on model dynamic selection driven by data partition hybrid sampling
Gao, X., Ren, B., Zhang, H., Sun, B., Li, J., Xu, J., He, Y., & Li, K. (2020) · Expert Systems with Applications · doi:10.1016/j.eswa.2020.113660 · Cited by 81
Learning with Noisy Labels via Sparse Regularization
Zhou, X. H., Liu, X., Wang, C., Zhai, D., Jiang, J., & Ji, X. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00014 · Cited by 83
“We also consider two imbalance types: long-tailed imbalance[4] and step imbalance [2].”
Background
Segmentation of pancreatic ductal adenocarcinoma (PDAC) and surrounding vessels in CT images using deep convolutional neural networks and texture descriptors
Mahmoudi, T., Kouzahkanan, Z. M., Radmard, A. R., Kafieh, R., Salehnia, A., Davarpanah, A. H., Arabalibeik, H., & Ahmadian, A. (2022) · Scientific Reports · doi:10.1038/s41598-022-07111-9 · Cited by 81
BalanceFL: Addressing Class Imbalance in Long-Tail Federated Learning
Shuai, X., Shen, Y., Jiang, S., Zhao, Z., Yan, Z., & Xing, G. (2022) · ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN) · doi:10.1109/ipsn54338.2022.00029 · Cited by 90
“Re-sampling based approaches over-sample the minority classes or under-sample the majority classes to calibrate the imbalanced distribution [1, 10].”
Background
Assessment of Deep Learning Techniques for Land Use Land Cover Classification in Southern New Caledonia
Rousset, G., Despinoy, M., Konrad Schindler, B., & Mangeas, M. (2021) · Remote Sensing · doi:10.3390/rs13122257 · Cited by 85
“Neural networks do not perform well when trained with unbalanced data sets [41].”
Background
Label-Efficient Breast Cancer Histopathological Image Classification
Qi, Q., Li, Y., Wang, J., Zheng, H., Huang, Y., Ding, X., & Rohde, G. K. (2018) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2018.2885134 · Cited by 82
An integrated machine learning model for aircraft components rare failure prognostics with log-based dataset
Dangut, M. D., Skaf, Z., & Jennions, I. K. (2020) · ISA Transactions · doi:10.1016/j.isatra.2020.05.001 · Cited by 79
The effects of data sampling with deep learning and highly imbalanced big data
Johnson, J. M., & Khoshgoftaar, T. M. (2020) · Information Systems Frontiers · doi:10.1007/s10796-020-10022-7 · Cited by 81
“Generally, model performance degrades as the level of class imbalance increases and the relative size of the positive class decreases (Buda et al. 2018).”
BackgroundMethodResultInfluential
Deep learning approach for defective spot welds classification using small and class-imbalanced datasets
Dai, W., Li, D., Tang, D., Wang, H., & Peng, Y. (2022) · Neurocomputing · doi:10.1016/j.neucom.2022.01.004 · Cited by 79
Breast cancer detection in thermograms using a hybrid of GA and GWO based deep feature selection method
Pramanik, R., Pramanik, P., & Sarkar, R. (2023) · Expert Systems with Applications · doi:10.1016/j.eswa.2023.119643 · Cited by 79
Personalized Dialogue Generation with Diversified TraitsPreprint
Zheng, Y., Chen, G., Huang, M., Liu, S., & Zhu, X. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1901.09672 · Cited by 79
“4, The datasets used to train, validate and test each trait classifier were balanced using the random minority oversampling approach [4], i.”
Method
Transfer learning for predicting conversion from mild cognitive impairment to dementia of Alzheimer's type based on a three-dimensional convolutional neural network
(2020) · doi:10.1016/j.neurobiolaging.2020.12.005 · Cited by 78
“Further, setting the conversion time at 3 years provided a well-balanced data set between MCI-NC (N=222) and MCI-C (N=228) (Buda et al., 2018).”
Pixel‐level multicategory detection of visible seismic damage of reinforced concrete components
Miao, Z., Ji, X., Okazaki, T., & Takahashi, N. (2021) · Computer-Aided Civil and Infrastructure Engineering · doi:10.1111/mice.12667 · Cited by 78
“Data imbalance can extensively affect the training process (Buda et al., 2018).”
Background
Going Deeper than Tracking: A Survey of Computer-Vision Based Recognition of Animal Pain and Emotions
Broomé, S., Feighelstein, M., Zamansky, A., Lencioni, G. C., Andersen, P. H., Pessanha, F., Mahmoud, M., Kjellström, H., & Salah, A. A. (2022) · International Journal of Computer Vision · doi:10.1007/s11263-022-01716-3 · Cited by 93
“Modifying the loss function is another option in a deep learning setting (Buda et al., 2018); this is commonly done for the above mentioned long-tail distribution scenarios.”
BackgroundInfluential
On oversampling imbalanced data with deep conditional generative models
Fajardo, V. A., Findlay, D. M., Jaiswal, C., Yin, X., Houmanfar, R., Xie, H., Liang, J., She, X., & Emerson, D. B. (2020) · Expert Systems with Applications · doi:10.1016/j.eswa.2020.114463 · Cited by 81
Artificial Intelligence and the Medical Physicist: Welcome to the Machine
Avanzo, M., Trianni, A., Botta, F., Talamonti, C., Stasi, M., & Iori, M. (2021) · Applied Sciences · doi:10.3390/app11041691 · Cited by 81
“For overcoming class imbalance, under‐sampling or over‐sampling can be applied; the latter has been proven to be more effective [110].”
Background
Few-shot imbalanced classification based on data augmentation
Chao, X., & Zhang, L. (2021) · Multimedia Systems · doi:10.1007/s00530-021-00827-0 · Cited by 77
Adaptive Logit Adjustment Loss for Long-Tailed Visual Recognition
Zhao, Y. L., Chen, W., Tan, X., Huang, K., & Zhu, J. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i3.20258 · Cited by 81
“These works can be roughly divided into three paradigms: re-sampling the training data (Buda, Maki, and Mazurowski 2018; Chawla et al. 2002; Wallace et al. 2011), re-weighting the coefficients of loss formulations (Menon et al. 2013; Cui et al. 2019; Ren et al. 2018) and adjusting the logit (Cao et…”
Background
REVISE: A Tool for Measuring and Mitigating Bias in Visual Datasets
Wang, A., Liu, A., Zhang, R., Kleiman, A., Kim, L., Zhao, D., Shirai, I., Narayanan, A., & Russakovsky, O. (2022) · International Journal of Computer Vision · doi:10.1007/s11263-022-01625-5 · Cited by 95
“Since then, researchers have noted the under-representation of object classes (Buda et al., 2017; Liu et al., 2009; Oksuz et al., 2019; Ouyang et al., 2016; Salakhutdinov et al., 2011; J. Yang et al., 2014), object contexts (Choi et al., 2012; Rosenfeld et al., 2018), object sub-categories (Zhu et…”
Background
A novel multi-module neural network system for imbalanced heartbeats classification
Jiang, J. B., Zhang, H., Pi, D., & Dai, C. (2019) · Expert Systems with Applications X · doi:10.1016/j.eswax.2019.100003 · Cited by 79
Cloud-based federated learning implementation across medical centers
Rajendran, S., Obeid, J. S., Binol, H., D`Agostino, R., Foley, K., Zhang, W., Austin, P., Brakefield, J., Gurcan, M. N., & Topaloglu, U. (2021) · JCO Clinical Cancer Informatics · doi:10.1200/cci.20.00060 · Cited by 76
“Furthermore, the scarcity of data in underrepresented populations may lead to biases when training data do not sufficiently reflect the attributes of these populations.(5) Healthcare data quality and algorithmic challenges are also known barriers for ML.”
Background
Automatic contouring system for cervical cancer using convolutional neural networks
Rhee, D. J., Jhingran, A., Rigaud, B., Netherton, T., Cardenas, C. E., Zhang, L., Vedam, S. S., Kry, S. F., Brock, K. K., Shaw, W. V., et al. (2020) · Medical Physics · doi:10.1002/mp.14467 · Cited by 76
On the Value of Oversampling for Deep Learning in Software Defect Prediction
Yedida, R., & Menzies, T. (2021) · IEEE Transactions on Software Engineering · doi:10.1109/tse.2021.3079841 · Cited by 80
Hybrid deep learning model for accurate classification of solid waste in the society
Zhang, H., Cao, H., Zhou, Y., Gu, C., & Li, D. (2023) · Urban Climate · doi:10.1016/j.uclim.2023.101485 · Cited by 76
Transfer‐learning‐based Raman spectra identification
Zhang, R., Xie, H., Cai, S., Hu, Y., Liu, G., Hong, W., & Tian, Z. (2019) · Journal of Raman Spectroscopy · doi:10.1002/jrs.5750 · Cited by 75
Active Learning for Imbalanced Datasets
Aggarwal, U., Popescu, A., & Hudelot, C. (2020) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv45572.2020.9093475 · Cited by 81
“Also relevant are works from imbalanced learning [3], which we use to test their impact on results.”
BackgroundMethodInfluential
Improving the prediction of continuous integration build failures using deep learning
Saidani, I., Ouni, A., & Mkaouer, M. W. (2022) · Automated Software Engineering · doi:10.1007/s10515-021-00319-5 · Cited by 83
Improving Deep Learning Classifiers Performance via Preprocessing and Class Imbalance Approaches in a Plant Disease Detection Pipeline
Ojo, M. O., & Zahid, A. (2023) · Agronomy · doi:10.3390/agronomy13030887 · Cited by 75
Dual-Attention Generative Adversarial Networks for Fault Diagnosis Under the Class-Imbalanced Conditions
Wang, R., Chen, Z., Zhang, S., & Li, W. (2021) · IEEE Sensors Journal · doi:10.1109/jsen.2021.3131166 · Cited by 75
Brain tumor classification from MRI scans: a framework of hybrid deep learning model with Bayesian optimization and quantum theory-based marine predator algorithm
Ullah, M. S., Khan, M. A., Masood, A., Mzoughi, O., Saidani, O., & Alturki, N. M. (2024) · Frontiers in Oncology · doi:10.3389/fonc.2024.1335740 · Cited by 74
Deep Learning–based Prescription of Cardiac MRI Planes
Blansit, K., Retson, T., Masutani, E., Bahrami, N., & Hsiao, A. (2019) · Radiology Artificial Intelligence · doi:10.1148/ryai.2019180069 · Cited by 73
“This redefined the MVS localization as a classification task to reduce the difficulties that arise from imbalanced data (Fig E2 [supplement]) (21,22).”
Background
A weighted hybrid ensemble method for classifying imbalanced data
Zhao, J., Jin, J. B., Chen, S., Zhang, R., Yu, B., & Liu, Q. (2020) · Knowledge-Based Systems · doi:10.1016/j.knosys.2020.106087 · Cited by 72
REVISE: A Tool for Measuring and Mitigating Bias in Visual Datasets
Wang, A., Narayanan, A., & Russakovsky, O. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58580-8_43 · Cited by 73
Dynamic Prediction in Clinical Survival Analysis Using Temporal Convolutional Networks
Jarrett, D. P., Yoon, J., & van der Schaar, M. (2019) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2019.2929264 · Cited by 77
“effective than alternative methods, especially with binary class labels [60].”
Automated segmentation of RGB-D images into a comprehensive set of building components using deep learning
Czerniawski, T., & Leite, F. L. (2020) · Advanced Engineering Informatics · doi:10.1016/j.aei.2020.101131 · Cited by 73
Learning from Class-imbalanced Data with a Model-Agnostic Framework for Machine Intelligent Diagnosis
Wu, J., Zhao, Z., Sun, C., Yan, R., & Chen, X. (2021) · Reliability Engineering & System Safety · doi:10.1016/j.ress.2021.107934 · Cited by 71
Detection of Malfunctioning Photovoltaic Modules Based on Machine Learning Algorithms
Hwang, H. P.-C., Ku, C., & Chan, J. C.-C. (2021) · IEEE Access · doi:10.1109/access.2021.3063461 · Cited by 72
PML: Progressive Margin Loss for Long-tailed Age Classification
Deng, Z., Liu, H., Wang, Y., Wang, C., Yu, Z., & Sun, X. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.01036 · Cited by 74
“With the remarkable success achieved by data-driven CNNs [26, 32, 40, 19], deep models have witnessed that the generation capacity is limited especially for imbalanced and distributed data [6, 25].”
BackgroundMethod
Seismic fault detection using convolutional neural networks with focal loss
Wei, X., Zhang, C., Kim, S., Jing, K., Wang, Y., Xu, S., & Xie, Z.-Z. (2021) · Computers & Geosciences · doi:10.1016/j.cageo.2021.104968 · Cited by 73
Development of Deep Learning Methodology for Maize Seed Variety Recognition Based on Improved Swin Transformer
Bi, C., Hu, N., Zou, Y., Zhang, S., Xu, S., & Yu, H. (2022) · Agronomy · doi:10.3390/agronomy12081843 · Cited by 72
Understanding CNN fragility when learning with imbalanced data
Dablain, D. A., Jacobson, K. N., Bellinger, C., Roberts, M., & Chawla, N. V. (2023) · Machine Learning · doi:10.1007/s10994-023-06326-9 · Cited by 79
“Imbalanced learning is concerned with designing methods that allow classifiers to better generalize from training to test data on minority classes (Fernández et al., 2018; He & Garcia, 2009; Buda et al., 2018).”
Generalized Parametric Contrastive Learning
Cui, J., Zhong, Z., Tian, Z., Liu, S., Yu, B., & Jia, J. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3278694 · Cited by 79
“However the center learning also needs to be balanced, which has been explored in [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [55].”
BackgroundInfluential
Wind turbine blade surface inspection based on deep learning and UAV-taken images
Xu, D., Wen, C., & Liu, J. (2019) · Journal of Renewable and Sustainable Energy · doi:10.1063/1.5113532 · Cited by 69
Text Classification Using Novel Term Weighting Scheme-Based Improved TF-IDF for Internet Media Reports
Jiang, Z., Gao, B., He, Y., Han, Y. M., Doyle, P., & Zhu, Q. (2021) · Mathematical Problems in Engineering · doi:10.1155/2021/6619088 · Cited by 69
“Furthermore, the classification performance is heavily affected by the imbalanced distribution of the dataset in TC [31, 32].”
Background
Learning to See: Convolutional Neural Networks for the Analysis of Social Science Data
Torres, M., & Cantú, F. (2021) · Political Analysis · doi:10.1017/pan.2021.9 · Cited by 71
“It is also useful to make sure that all classes in the training set are representedby a similar number of examples (Buda, Maki, and Mazurowski 2018).”
Background
Mitotic nuclei analysis in breast cancer histopathology images using deep ensemble classifier
Sohail, A., Khan, A., Nisar, H., Tabassum, S., & Zameer, A. (2021) · Medical Image Analysis · doi:10.1016/j.media.2021.102121 · Cited by 72
Diabetes mellitus risk prediction in the presence of class imbalance using flexible machine learning methods
Sadeghi, S., Khalili, D., Ramezankhani, A., Mansournia, M. A., & Parsaeian, M. (2022) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-022-01775-z · Cited by 68
“To cope with the class imbalance problem, two main approaches have been established in the literature [15].”
BackgroundMethodInfluential
Tackling class imbalance in computer vision: a contemporary review
Saini, M., & Susan, S. (2023) · Artificial Intelligence Review · doi:10.1007/s10462-023-10557-6 · Cited by 79
Enhanced Stacked Denoising Autoencoder-Based Feature Learning for Recognition of Wafer Map Defects
Yu, J. (2019) · IEEE Transactions on Semiconductor Manufacturing · doi:10.1109/tsm.2019.2940334 · Cited by 72
“Methods of dealing with class imbalance are well studied for those typical learning models [55], [56], which can be divided into two main categories, i.”
Background
Goosegrass Detection in Strawberry and Tomato Using a Convolutional Neural Network
Sharpe, S. M., Schumann, A. W., & Boyd, N. S. (2020) · Scientific Reports · doi:10.1038/s41598-020-66505-9 · Cited by 67
EEG data augmentation: towards class imbalance problem in sleep staging tasks
Fan, J., Sun, C., Chen, C., Jiang, X., Liu, X., Zhao, X. F., Meng, L., Dai, C., & Chen, W. (2020) · Journal of Neural Engineering · doi:10.1088/1741-2552/abb5be · Cited by 67
KNN weighted reduced universum twin SVM for class imbalance learning
Ganaie, M. A., & Tanveer, M. (2022) · Knowledge-Based Systems · doi:10.1016/j.knosys.2022.108578 · Cited by 72

Authors: Buda, M., Maki, A., & Mazurowski, M. A.
Affiliations: Duke University (US)
KTH Royal Institute of Technology (SE)
Corresponding author: Buda, Mateusz
ORCID: Buda, Mateusz
Maki, Atsuto
Mazurowski, Maciej A.
Venue: Neural Networks
Publisher: Elsevier BV
Published: 2018-10
Volume: 106
Pages: 249-259
ISSN: 0893-6080 (ISSN-L)
DOI: 10.1016/j.neunet.2018.07.011
OpenAlex: W2767106145
Cited by: 3,088
References: 83
Citations by year: 2026: 459 · 2025: 507 · 2024: 371 · 2023: 405 · 2022: 405 · 2021: 380 · 2020: 308 · 2019: 189 · 2018: 58
FWCI: 131.58
Type: journal article
Language: en
Open access: green
arxiv.org
Licence: www.elsevier.com/tdm/userlicense/1.0/ (tdm)
www.elsevier.com/legal/tdmrep-license (tdm)
Locations: Neural Networks (publishedVersion)
PubMed (publishedVersion)
arXiv (Cornell University) (submittedVersion)
KTH Publication Database DiVA (KTH Royal Institute of Technology) (submittedVersion, other-oa)
arXiv (Cornell University)
Topics: Physical Sciences › Computer Science › Artificial Intelligence › Imbalanced Data Classification Techniques
Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
Keywords: class imbalance, convolutional neural networks, oversampling, undersampling, two-stage training, thresholding, area under the receiver operating characteristic, multi-class classification, overfitting, MNIST, CIFAR-10, ImageNet
MeSH terms: Machine Learning, Humans, Probability, ROC Curve, Neural Networks, Computer

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

A systematic study of the class imbalance problem in convolutional neural networks

Abstract

Overview

Integrity signals

Verification

Claims the paper makes

Claims other papers make about this one

Citations

Cites

Cited by

Related

Metadata

Record

About these checks