Enhancing Ai Ethics Through Integrating Aif360 With Generative Ai

Author

Christina Ieisha Hastings, Prairie View A&M University

Date of Award

12-2024

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Degree Discipline

Electrical Engineering

Abstract

In recent years, Artificial Intelligence (AI) has brought about significant positive changes in people's lives with respect to convenience and efficiency. The primary aim of AI is to emulate human intelligence in machines, creating a system capable of thinking and replicating human cognitive functions. AI encompasses tasks that mirror human behaviors, such as visual perception, speech recognition, and language translation.

The realm of AI ethics is an evolving and interdisciplinary field dedicated to addressing ethical considerations in AI. The establishment of ethical standards for AI is vital for developing morally upright AI systems and ensuring ethical conduct in AI applications. Fairness AI strives to recognize and mitigate bias throughout the entire life cycle of AI technique development, spanning data curation and preparation, modeling, evaluation, and deployment. However, two concerns have been raised regarding current efforts: first, existing endeavors seem to lack a thorough and comprehensive evaluation of the effectiveness of techniques in mitigating bias; second, existing work has not fully explored enhancing bias mitigation techniques via generative AI.

This dissertation conducts a comprehensive evaluation of the effectiveness of reweighing samples with the addition of AI generated data in addressing bias associated with traditional machine learning models, utilizing the AI Fairness 360 (AIF360) framework. The reweighing process is conducted with respect to privileged attributes such as sex and race. Subsequently, each traditional machine learning model undergoes classification tasks on both the original datasets and the new datasets resulting from reweighing samples.

In addition, this dissertation explores the potential of diffusion models to generate synthetic tabular data to improve AI fairness. The Tabular Denoising Diffusion Probabilistic Model (TabDDPM), a diffusion model adaptable to any tabular dataset and capable of handling various feature types, was utilized with different amounts of generated data for data augmentation. Additionally, reweighting samples from AIF360 was employed to further enhance AI fairness. Experimental results demonstrate that the synthetic data generated by TabDDPM improves fairness in binary classification.

Index Terms: Artificial Intelligence, AIF360, Generative AI, Reweighting.

Committee Chair/Advisor

Pamela Obiomon

Committee Member

Lijun Qian

Committee Member

Xishuang Dong

Committee Member

Xiangfang Li

Committee Member

Camille Gibson

Rights

Date of Digitization

4/1/2025

Contributing Institution

John B Coleman Library

City of Publication

Prairie View

MIME Type

Application/PDF

Recommended Citation

Hastings, C. I. (2024). Enhancing Ai Ethics Through Integrating Aif360 With Generative Ai. Retrieved from https://digitalcommons.pvamu.edu/pvamu-dissertations/111