Two-Stage Sampling: A Framework for Imbalanced Classification With Overlapped Classes

Firdausanti, Neni Alya; Fatyanosa, Tirana Noor; Data, Mahendra; Mendonça, Israel; Aritsugi, Masayoshi

doi:10.1109/bigdata55660.2022.10020788

Cited by 3 publications

(9 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To address these issues effectively, a combination of both strategies has been found to be more successful in overcoming these challenges, as demonstrated in previous studies [25,41]. In our previous research [28], combining an oversampling method, SWIM, with undersampling methods like Tomek Links and Edited Nearest Neighbors (ENN) significantly enhances classification outcomes in imbalanced situations involving overlapping classes.…”

Section: Related Workmentioning

confidence: 84%

“…The SWIM framework proves to be effective in handling extremely imbalanced problems. Based on our previous study [28], combining SWIM with undersampling methods, such as Tomek Links [29] and Edited Nearest Neighbors (ENN) [30], shows superiority in improving the classification performance on Imbalanced cases with overlapped classes. However, it still requires adequate study addressing the issue arising from noise data.…”

Section: Introductionmentioning

confidence: 94%

“…The state-of-the-art oversampling framework, namely the SWIM, generates synthetic samples regarding the information of the majority class and has demonstrated its effectiveness in managing highly imbalanced problems. Our previous study demonstrates that enhancing the representation of the majority class, by utilizing undersampling techniques to remove overlapped data, results in a better generation of synthetic data in the SWIM framework and leads to subsequent improvement in classification performance [28]. However, the problem arising from noise data remains to be solved and needs to be addressed.…”

Section: Related Workmentioning

confidence: 99%

“…In this section, we introduced a framework to improve the synthetic data generation process of the SWIM by adding a mechanism to eliminate the noise data in the majority class. While our previous study focused on removing overlapped data between the classes using undersampling methods [28], this study highlights noise detection methods and eliminates them to provide a noise-free majority class and improve the quality of the synthetic data generation process. Identifying and removing noise data from the majority class enhances the accuracy of features used by the SWIM framework, leading to improved representation of synthetic data.…”

Section: Noise-free Sampling With Majority (Nf-swim) Frameworkmentioning

confidence: 99%

“…A study by [32] demonstrates an improvement of SMOTE by addressing the noisy and borderline examples problem in imbalance classification. Additionally, our previous study [28] shows that a better representation of the majority class can improve the SWIM framework to generate synthetic minority data and improve classification performance. Therefore, these inspire our study to modify the SWIM framework by integrating a procedure to detect and remove the possible noise data in the majority class.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Noise-free sampling with majority framework for an imbalanced classification problem

Firdausanti,

Mendonça,

Aritsugi

2024

Knowl Inf Syst

Self Cite

View full text Add to dashboard Cite

Class imbalance has been widely accepted as a significant factor that negatively impacts a machine learning classifier's performance. One of the techniques to avoid this problem is to balance the data distribution by using sampling-based approaches, in which synthetic data is generated using the probability distribution of classes. However, this process is sensitive to the presence of noise in the data, in which the boundaries between the majority class and the minority class are blurred. Such phenomena shift the algorithm's decision boundary away from an ideal outcome. In this work, we propose a framework that tackles two primary objectives: first, to address class distribution imbalance by synthetically increasing the data of a minority class; and second, to devise an efficient noise reduction technique that improves the class balance algorithm. The proposed framework focuses its capability towards removing noisy elements from the majority class, and by doing so, provides more accurate information to the subsequent synthetic data generator algorithm. Experimental results show that our framework is capable of improving the prediction accuracy of eight classifiers from 7.78% up to 67.45% for eleven datasets tested.

show abstract

Section: Related Workmentioning

confidence: 84%

Section: Introductionmentioning

confidence: 94%

Section: Related Workmentioning

confidence: 99%

Section: Noise-free Sampling With Majority (Nf-swim) Frameworkmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Noise-free sampling with majority framework for an imbalanced classification problem

Firdausanti,

Mendonça,

Aritsugi

2024

Knowl Inf Syst

Self Cite

View full text Add to dashboard Cite

show abstract

A Population-Based Hybrid Approach for Hyperparameter Optimization of Neural Networks

et al. 2023

Self Cite

View full text Add to dashboard Cite

Hyperparameter optimization is a fundamental part of Auto Machine Learning (AutoML) and it has been widely researched in recent years; however, it still remains as one of the main challenges in this area. Motivated by the need of faster and more accurate hyperparameter optimization algorithms we developed HyperBRKGA, a new population-based approach for hyperparameter optimization. HyperBRKGA combines the Biased Random Key Genetic Algorithm with an Exploitation Method in order to search the hyperparameter space more efficiently than other commonly used hyperparameter optimization algorithms, such as Grid Search, Random Search, CMA-ES or Bayesian Optimization. We develop and test two different alternatives for this Exploitation Method: Random Walk and Bayesian Walk. We also discuss and implement other schemes, such as a Training Data Reduction Strategy and a Diversity Control strategy, in order to further improve the efficacy of our method. We performed several computational experiments on 8 different datasets to assess the effectiveness of the proposed approach. Results showed that HyperBRKGA could find hyperparameter configurations that outperformed in terms of predictive quality the baseline methods in 6 out of 8 datasets while showing a reasonable execution time. Lastly, we conducted an ablation study and showed that the addition of every component was relevant to achieving high quality results.

show abstract

Noise-Free Sampling with Majority for Imbalanced Classification Problem

Firdausanti,

Mendonça,

Aritsugi

2023

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Two-Stage Sampling: A Framework for Imbalanced Classification With Overlapped Classes

Cited by 3 publications

References 24 publications

Noise-free sampling with majority framework for an imbalanced classification problem

Noise-free sampling with majority framework for an imbalanced classification problem

A Population-Based Hybrid Approach for Hyperparameter Optimization of Neural Networks

Noise-Free Sampling with Majority for Imbalanced Classification Problem

Contact Info

Product

Resources

About