ItemBook Chapter

GAN-Based Metaheuristic Techniques for Data Generation and Imbalance Data Control

Title
GAN-Based Metaheuristic Techniques for Data Generation and Imbalance Data Control
Creator
Vineetha, K.R.; Resmi, K.R.; Amrutha, K.; Blessie, E. Chandra; Omanakuttan, Midhun
Description
In cybersecurity today, the power of features such as threat detection, anomaly spotting, and predictive analytics depends heavily on having abundant, properly dispersed datasets. The actual datasets often fall short, suffering both from a lack of volume and skewed class distribution, for example, a flood of routine network activity records overshadowing the infrequent but vital records of malicious behavior. The performance of data-driven models hinges on access to abundant, well-distributed data. However, real-world datasets frequently exhibit inadequate sample sizes and pronounced class imbalances, limiting the viability of complex models. This chapter proposes a novel strategy for generating synthetic data and effectively managing class imbalance, leveraging the integration of Generative Oppositional Networks (GANs) and sophisticated metaheuristic optimization techniques. Rather than settling for fixed GAN architectures, our approach progressively enhances a dynamic GAN framework by deploying a metaheuristic search to identify optimal network topologies, antidote scaling factors, and training schedules. This iterative calibration enables the model to adaptively respond to the imbalance and ensures a richer, balanced synthetic training environment. This adaptive optimization addresses common GAN training pitfalls, mode collapse, and instability while consistently producing synthetic samples that are both precise and varied. What sets the framework apart is its built-in ability to detect and over-represent minority classes, intelligently augmenting the dataset to correct class imbalance without falling back on naive duplication. By equipping GANs with metaheuristic reasoning, the study seeks to elevate data synthesis beyond current limits, generating more robust and impartial machine learning models in any domain where data collection is limited or systematically skewed. 2026 selection and editorial matter, E. Chandra Blessie, Pethuru Raj, and B. Sundaravadivazhagan; individual chapters, the contributors.
Source
Generative Adversarial Networks for Cybersecurity:: Protecting Data and Networks;pp.220-234
Date
01-01-2026
Publisher
CRC Press
Coverage
Vineetha K.R., Department of Computer Science, CHRIST (Deemed to be University), Bengaluru, Karnataka, India; Resmi K.R., Department of Computer Science, CHRIST (Deemed to be University), Bengaluru, Karnataka, India; Amrutha K., Department of Computer Science, CHRIST (Deemed to be University), Bengaluru, Karnataka, India; Blessie E.C., School of Innovation, KG College of Arts and Science, Tamil Nadu, Coimbatore, India; Omanakuttan M., Computing Department, International College of Dundee Scotland, United Kingdom
Rights
Restricted Access; Hardcopy may be available in the library
Relation
ISBN: 978-100365299-1; 978-104109801-0; 978-104110023-2;
Format
online
Language
English
Type
Book chapter

Citation

Vineetha, K.R.; Resmi, K.R.; Amrutha, K.; Blessie, E. Chandra; Omanakuttan, Midhun, “GAN-Based Metaheuristic Techniques for Data Generation and Imbalance Data Control,” CHRIST (Deemed To Be University) Institutional Repository, accessed June 18, 2026, https://archives.christuniversity.in/items/show/24455.

Output Formats