Leveraging GANs for synthetic tabular data generation: a platform-centric solution to data scarcity
Muhammad Mobeen, Awais Afzal, Reeda Saeed, Tayyaba Arshad, Muhammad Imran Asad · IET conference proceedings. · 2024
In an era dominated by data-driven decision-making, the insufficiency of high-quality and diverse datasets presents a profound challenge for researchers, businesses, and organizations. The scarcity of data, particularly in certain domains, constrains the application of Artificial Intelligence (AI) and machine learning, hindering progress and innovation. This project endeavours to alleviate this bottleneck through the development of Synthium AI, a platform that provides easy access to advanced Synthetic Tabular Data Generation Deep Neural Network Models. Synthium AI leverages Generative Adversarial Networks (GANs) to perform Synthetic Tabular Data Generation (STDG), focusing on a wide range of tabular datasets. By democratizing access to high-quality synthetic data, Synthium AI empowers stakeholders to overcome the challenges posed by data scarcity, facilitating innovation and progress in various domains. As AI applications become increasingly prevalent, the ability to generate synthetic datasets emerges as an enabler, bridging the gap created by data limitations and fostering advancements in AI research and applications. While the current implementation focuses on tabular data, future work will extend Synthium AI's capabilities to address challenges in computer vision and image processing.