Target-oriented augmentation privacy-protection domain adaptation for imbalanced ECG beat classification

Liqiang, Yuan; Siyal, Mohammed Yakoob

doi:10.1016/j.bspc.2023.105308

Cited by 4 publications

(4 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similarly, lower classification performance for S and V class samples is also reported in [16] and [51] where data augmentation methods are used to tackle imbalanced data issues. In comparison to [16], [40], [51], our proposed method achieved a better performance score for the most crucial class of S and V beats. These results substantiate previous findings in literature [52].…”

Section: Comparisonmentioning

confidence: 82%

“…The classification results of [40] presented in Table V depict that generating minority samples from minority class samples via SMOTE causes over-fitting on minority class samples [52] and, therefore, a lower sensitivity score is achieved for S and V class samples in [40]. Similarly, lower classification performance for S and V class samples is also reported in [16] and [51] where data augmentation methods are used to tackle imbalanced data issues. In comparison to [16], [40], [51], our proposed method achieved a better performance score for the most crucial class of S and V beats.…”

Section: Comparisonmentioning

confidence: 96%

“…In this study, class weights are computed for each batch during training, and a weighted cross entropy loss function is used to train a CNN model for arrhythmia classification. In another study, the issue of imbalanced data in ECG classification is addressed by synthesizing minority class samples using a target-oriented augmentation method [16]. The data augmentation method has also been adopted by Yong Xia [51] to tackle the challenge of imbalanced data in arrhythmia detection.…”

Section: Handling Imbalanced Datamentioning

confidence: 99%

“…However, the classification model expects a balanced distribution of samples among different classes, and thus the existence of imbalanced data significantly deteriorates the model performance [6], [14], [15]. During training, the scarce representation of minority classes leads the classification model to learn an imbalanced and biased representation and thus perform poorly on minority class samples [16], [17]. Therefore, in arrhythmia detection where rare events are of vital importance, imbalanced data problems should be considered to characterize minority-class samples adequately.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Deep Representation Learning With Sample Generation and Augmented Attention Module for Imbalanced ECG Classification

Zubair,

Woo,

Lim

et al. 2024

IEEE J. Biomed. Health Inform.

View full text Add to dashboard Cite

Developing an efficient heartbeat monitoring system has become a focal point in numerous healthcare applications. Specifically, in the last few years, heartbeat classification for arrhythmia detection has gained considerable interest from researchers. This paper presents a novel deep representation learning method for the efficient detection of arrhythmic beats. To mitigate the issues associated with the imbalanced data distribution, a novel re-sampling strategy is introduced. Unlike the existing oversampling methods, the proposed technique transforms majority-class samples into minority-class samples with a novel translation loss function. This approach assists the model in learning a more generalized representation of crucially important minority class samples. Moreover, by exploiting an auxiliary feature, an augmented attention module is designed that focuses on the most relevant and target-specific information. We adopted an inter-patient classification paradigm to evaluate the proposed method. The experimental results of this study on the MIT-BIH arrhythmia database clearly indicate that the proposed model with augmented attention mechanism and over-sampling strategy significantly learns a balanced deep representation and improves the classification performance of vital heartbeats.

show abstract