Training deep neural networks on imbalanced data sets

By Shoujin Wang, Jia Wu, Longbing Cao, Qinxue Meng, Paul J. Kennedy. IEEE International Joint Conference on Neural Network, 2016

452 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Electrical and Electronic Engineering, Media Technology

View Paper: https://www.semanticscholar.org/paper/a0d86c44f2843a483dfffbfc03dda230bbaad4cc

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Electrical and Electronic Engineering

In Media Technology

Citations in the corpus

Cost-Sensitive Learning of Deep Feature Representations From Imbalanced Data (2015, background): “After the submission of this work for review, we note that a number of new approaches have recently been proposed to incorporate class-specific costs in the deep networks [30, 31, 32].”
DeepSMOTE: Fusing Deep Learning and SMOTE for Imbalanced Data (2021, background): “Mean False Error [73] and Focal Loss [74] are two of the most popular approaches based on this principle.”
Remix: Rebalanced Mixup (2020, background): “Re-Weighting Re-weighting (cost-sensitive learning) focuses on tuning cost or loss to redefine the importance of each class or sample [6,34,15,3].”
MFC-GAN: Class-imbalanced dataset classification using Multiple Fake Class Generative Adversarial Network (2019, methodology): “SMOTE can also lead to over-generalization with high variance [18].”
Data Augmentation for Deep-Learning-Based Electroencephalography. (2020, methodology): “Beside resampling techniques, another way to deal with imbalanced data is cost sensitive learning [104].”
Deep Imbalanced Learning for Face Recognition and Attribute Prediction (2018, background): “Alternatives [34], [35] simply tune the networks to maximize a class-balanced accuracy measure.”
Addressing Class Imbalance in Federated Learning (2020, methodology): “Other cost-sensitive methods need speciﬁc information of minority classes, e.g., MFE Loss (Wang et al. 2016) regards minority classes as positive classes, and calculates false positive and false negative to generate a new loss form.”
Imbalanced Deep Learning by Minority Class Incremental Rectification (2018, methodology): “A few recent studies [41,62,63,64,65] have exploited classic strategies in singlelabel deep learning.”
Sequential Recommender Systems: Challenges, Progress and Prospects (2019, background): “Deep neural networks [23] have natural strength to model and capture the comprehensive relations over different entities (e.g., users, items, interactions) in a sequence, and thus they nearly dominate SRSs in the past few years.”
A systematic study of the class imbalance problem in convolutional neural networks (2017, background): “New kinds of loss function for neural networks training were also developed [27].”
SleepEEGNet: Automated sleep stage scoring with sequence to sequence deep learning approach (2019, methodology): “We extended the proposed loss functions, mean false error (MFE) and mean squared false error (MSFE), in [29] for the multi-class classiﬁcation task.”