Multidimensional Perspective to Data Preprocessing for Model Cognition Verity

Advances in systems analysis, software engineering, and high performance computing book series, Journal Year: 2024, Volume and Issue: unknown, P. 15 - 57

Published: May 14, 2024

Reliable data analysis depends on effective preparation, especially since AI-driven business intelligence unbiased and error-free for decision-making. However, developing a reliable dataset is difficult task that requires expertise. Due to the costly damage negligible error in can cause system, good understanding of processes quality transformation necessary. Data varies properties, which determines how it generated, errors it, transformations needs undergo before fed into model. Also, most used analytics sourced from public stores without means verify its or what further steps need be taken preprocessing optimal performance. This chapter provides detailed description practical scientific procedures generate develop different models scenarios. highlights tools techniques clean prepare performance prevent unreliable outcomes.

Language: Английский

---

Minimum Variance Weighted Broad Cascade Network Structure for Imbalanced Classification

Published: Jan. 1, 2025

Broad learning system (BLS) are widely used due to their speed and versatility. Despite efficiency, minority class samples' accuracy is sometimes overlooked when dealing with severely imbalanced rate data. Traditional weighted BLS only considers the number of samples, such a fixed weighting leads poor classification performance. In addition, original does not take into account dispersion after its random data mapping. To solve aforementioned concerns, this study presents minimum variance broad cascade network. By incorporating constraint reduced category information cascading feature nodes enhancement at each level, network may extract enough valuable from while accounting for dispersion. We use support vector describe hyperplane distribution in order further investigate data's distribution. From there, we develop boundary strategy concentrate on samples whose boundaries hard discern. Extensive comparative validation 20 real-world datasets confirms that our method has significant advantages problems.

Language: Английский

---

Long-term policy guidance for sustainable energy transition in Nigeria: A deep learning-based peak load forecasting with econo-environmental scenario analysis

Energy, Journal Year: 2025, Volume and Issue: unknown, P. 135707 - 135707

Published: March 1, 2025

Language: Английский

---

Improvement of Bank Fraud Detection Through Synthetic Data Generation with Gaussian Noise

Technologies, Journal Year: 2025, Volume and Issue: 13(4), P. 141 - 141

Published: April 4, 2025

Bank fraud detection faces critical challenges in imbalanced datasets, where fraudulent transactions are rare, severely impairing model generalization. This study proposes a Gaussian noise-based augmentation method to address class imbalance, contrasting it with SMOTE and ADASYN. By injecting controlled perturbations into the minority class, our approach mitigates overfitting risks inherent interpolation-based techniques. Five classifiers, including XGBoost convolutional neural network (CNN), were evaluated on augmented datasets. achieved superior performance noise-augmented data (accuracy: 0.999507, AUC: 0.999506), outperforming These results underscore noise’s efficacy enhancing accuracy, offering robust alternative conventional oversampling methods. Our findings emphasize pivotal role of strategies optimizing classifier for financial data.

Language: Английский

---

Retraining and evaluation of machine learning and deep learning models for seizure classification from EEG data

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: May 2, 2025

Electroencephalography (EEG) is one of the most used techniques to perform diagnosis epilepsy. However, manual annotation seizures in EEG data a major time-consuming step analysis process EEGs. Different machine learning models have been developed automated detection from large gap observed between initial accuracies and those clinical practice. In this work, we reproduced assessed accuracy number models, including deep networks, for Benchmarking included three different datasets training testing, manually annotated local patient further testing. Random forest convolutional neural network achieved best results on public data, but reduction was testing with especially network. We expect that retrained available work will contribute integration as tools improve settings.

Language: Английский

---

A novel framework for efficient dominance-based rough set approximations using K-dimensional (K-D) tree partitioning and adaptive recalculations techniques

Engineering Applications of Artificial Intelligence, Journal Year: 2025, Volume and Issue: 154, P. 110993 - 110993

Published: May 8, 2025

Language: Английский

---

A multimodal data generation method for imbalanced classification with dual-discriminator constrained diffusion model and adaptive sample selection strategy

Information Fusion, Journal Year: 2024, Volume and Issue: 117, P. 102843 - 102843

Published: Dec. 5, 2024

Language: Английский

---

DDP-DAR: Network Intrusion Detection Based on Denoising Diffusion Probabilistic Model and Dual-Attention Residual Network

Neural Networks, Journal Year: 2024, Volume and Issue: 184, P. 107064 - 107064

Published: Dec. 19, 2024

Language: Английский

---

Handling imbalance dataset issue in insider threat detection using machine learning methods

Computers & Electrical Engineering, Journal Year: 2024, Volume and Issue: 120, P. 109726 - 109726

Published: Oct. 1, 2024

Language: Английский

---

Evaluation of Synthetic Data Generation Models for Balancing Multiclass Metabolomic Profiles

Published: Sept. 13, 2024

Language: Английский

---