A review of ensemble learning and data augmentation models for class imbalanced problems: Combination, implementation and evaluation

Expert Systems with Applications, Journal Year: 2023, Volume and Issue: 244, P. 122778 - 122778

Published: Dec. 10, 2023

Class imbalance (CI) in classification problems arises when the number of observations belonging to one class is lower than other. Ensemble learning combines multiple models obtain a robust model and has been prominently used with data augmentation methods address problems. In last decade, strategies have added enhance ensemble methods, along new such as generative adversarial networks (GANs). A combination these applied many studies, evaluation different combinations would enable better understanding guidance for application domains. this paper, we present computational study evaluate prominent benchmark CI We general framework that evaluates 9 Our objective identify most effective improving performance on imbalanced datasets. The results indicate can significantly improve find traditional synthetic minority oversampling technique (SMOTE) random (ROS) are not only selected problems, but also computationally less expensive GANs. vital development novel handling

Language: Английский

---

Generative Adversarial Networks (GANs) in Medical Imaging: Advancements, Applications, and Challenges

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 35728 - 35753

Published: Jan. 1, 2024

Generative Adversarial Networks are a class of artificial intelligence algorithms that consist generator and discriminator trained simultaneously through adversarial training. GANs have found crucial applications in various fields, including medical imaging. In healthcare, contribute by generating synthetic images, enhancing data quality, aiding image segmentation, disease detection, synthesis. Their importance lies their ability to generate realistic facilitating improved diagnostics, research, training for professionals. Understanding its applications, algorithms, current advancements, challenges is imperative further advancement the imaging domain. However, no study explores recent state-of-the-art development To overcome this research gap, extensive study, we began exploring vast array imaging, scrutinizing them within research. We then dive into prevalent datasets pre-processing techniques enhance comprehension. Subsequently, an in-depth discussion GAN elucidating respective strengths limitations, provided. After that, meticulously analyzed results experimental details some cutting-edge obtain more comprehensive understanding Lastly, discussed diverse encountered future directions mitigate these concerns. This systematic review offers complete overview encompassing application domains, models, analysis, challenges, directions, serving as valuable resource multidisciplinary studies.

Language: Английский

---

Evaluating the Impact of Data Augmentation on Predictive Model Performance

Published: Feb. 21, 2025

In supervised machine learning (SML) research, large training datasets are essential for valid results. However, obtaining primary data in analytics (LA) is challenging. Data augmentation can address this by expanding and diversifying data, though its use LA remains underexplored. This paper systematically compares techniques their impact on prediction performance a typical task: of academic outcomes. Augmentation demonstrated four SML models, which we successfully replicated from previous LAK study based AUC values. Among 21 techniques, SMOTE-ENN sampling performed the best, improving average 0.01 approximately halving time compared to baseline models. addition, 99 combinations chaining found minor, although statistically significant, improvements across models when adding noise (+0.014). Notably, some significantly lowered predictive or increased fluctuation related random chance. paper's contribution twofold. Primarily, our empirical findings show that provide most reliable applications SML, computationally more efficient than deep generation methods with complex hyperparameter settings. Second, community may benefit validating recent through independent replication.

Language: Английский

---

Prediction of Students’ Academic Performance in the Programming Fundamentals Course Using Long Short-Term Memory Neural Networks

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 5882 - 5898

Published: Jan. 1, 2024

In recent years, there has been evidence of a growing interest on the part universities to know in advance academic performance their students and allow them establish timely strategies avoid desertion failure. One biggest challenges predicting student is presented course "Programming Fundamentals" Computer Science, Software Engineering, Information Systems Engineering careers Peruvian for high dropout rates. The objective this research was explore efficiency Long-Short Term Memory Networks (LSTM) field Educational Data Mining (EDM) predict during seventh, eighth, twelfth, sixteenth weeks semester, which allowed us identify at risk failing course. This compares several predictive models, such as Deep Neural Network (DNN), Decision Tree (DT), Random Forest (RF), Logistic Regression (LR), Support Vector Classifier (SVM), K-Nearest Neighbor (KNN). A major challenge machine learning algorithms face class imbalance dataset, resulting over-fitting available data and, consequently, low accuracy. We use Generative Adversarial (GAN) Synthetic Minority Over-sampling Technique (SMOTE) balance needed our proposal. From experimental results based accuracy, precision, recall, F1-Score, superiority model verified concerning better classification, with 98.3% accuracy week 8 using LSTM-GAN, followed by DNN-GAN 98.1%

Language: Английский

---

Enhancing and improving the performance of imbalanced class data using novel GBO and SSG: A comparative analysis

Neural Networks, Journal Year: 2024, Volume and Issue: 173, P. 106157 - 106157

Published: Feb. 2, 2024

Language: Английский

---

A histogram SMOTE-based sampling algorithm with incremental learning for imbalanced data classification

Information Sciences, Journal Year: 2024, Volume and Issue: 686, P. 121193 - 121193

Published: July 25, 2024

Language: Английский

---

Transfer learning in agriculture: a review

Artificial Intelligence Review, Journal Year: 2025, Volume and Issue: 58(4)

Published: Jan. 25, 2025

Language: Английский

---

Prospectivity modelling of critical mineral deposits using a generative adversarial network with oversampling and positive-unlabelled bagging

Ore Geology Reviews, Journal Year: 2023, Volume and Issue: 162, P. 105665 - 105665

Published: Sept. 14, 2023

The demand for critical minerals is rapidly increasing worldwide, yet future global supply remains uncertain due to the difficulty in discovering new deposits using traditional methods. To increase success rate of exploration projects these vital resources, use artificial intelligence continuously big and complex data analysis. This study proposes a machine learning-based framework that tackles common problems associated with exploring mineral deposits, such as shortage known occurrences, challenges selecting negative samples barren regions, unbalanced training data. Our combines an improved generative adversarial network positive unlabelled learning enhance efficiency. test performance framework, we create prospectivity maps mafic-ultramafic intrusion-hosted mineralisation cobalt, chromium, nickel Gawler Craton, South Australia. models are trained on carefully selected set independent features based conceptual model derived from open-access data, resulting high stable performance. show strong spatial correlation between probabilities occurrences predict potential greenfield regions exploration. demonstrate significantly higher accuracy compared conventional approach standard random forest classifier reveal geophysical play crucial role mapping prospective minerals. Overall, our has by providing more accurate efficient identifying mining operations.

Language: Английский

---

Predicting diabetes in adults: identifying important features in unbalanced data over a 5-year cohort study using machine learning algorithm

BMC Medical Research Methodology, Journal Year: 2024, Volume and Issue: 24(1)

Published: Sept. 27, 2024

Language: Английский

---

Ensemble Synthesized Minority Oversampling-Based Generative Adversarial Networks and Random Forest Algorithm for Credit Card Fraud Detection

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 89694 - 89710

Published: Jan. 1, 2023

The recent increase in credit card fraud is rapidly has caused huge monetary losses for individuals and financial institutions. Most frauds are conducted online by illegally obtaining payment credentials through data breaches, phishing, or scamming. Many solutions have been suggested to address the problem transactions. However, high class imbalance major challenge that faces existing construct an effective detection model. of techniques used overestimate distribution minority class, resulting highly overlapped noisy unrepresentative features, which cause either overfitting imprecise learning. In this study, a model (CCFDM) proposed based on ensemble learning generative adversarial network (GAN) assisted Ensemble Synthesized Minority Oversampling (ESMOTE-GAN). Multiple subsets were extracted using under-sampling SMOTE was applied generate less skewed sets prevent GAN from modeling noise. These train diverse models synthesized subsets. A set Random Forest classifiers then trained ESMOTE-GAN technique. probabilistic outputs combined weighted voting scheme decision-making. results show achieved 1.9%, 3.2% improvements overall performance rate, respectively, with 0% false alarm rate. Due massive number transactions, even tiny positive rate can overwhelm analysis team. Thus, improved reduced cost needed manual analysis.

Language: Английский

---