Term of Award

Spring 2025

Degree Name

Master of Science, Information Technology

Document Type and Release Option

Thesis (open access)

Copyright Statement / License for Reuse

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Department

Department of Information Technology

Committee Chair

Hayden Wimmer

Committee Member 1

Jongyeop Kim

Committee Member 2

Atef Shalan

Abstract

The rapid advancement of artificial intelligence has significantly influenced digital media, enabling both the detection and generation of synthetic content. This thesis, titled Artificial Intelligence for digital deception: A Study on Detection, Generation, and Evaluation, explores AI’s role in digital deception through three distinct studies focused on facial expression analysis for deepfake detection, machine learning-based spam classification on cloud platforms, and the evaluation of generative AI state-of-the-art text to video models. The first study investigates the effectiveness of facial expression analysis in distinguishing between deepfake and genuine videos. Using Noldus FaceReader 7, participant’s emotional responses were analyzed while viewing deep-fake and real videos. Results reveal significant differences in emotional responses, highlighting the potential of automated expression analysis for misinformation detection. The second study evaluates the performance of three machine learning models, random forest logistic regression, and decision tree are deployed on Microsoft Azure data bricks and Google Cloud vertex AI for spam classification. Through an analysis of inference time, classification performance, and cost efficiency, the study reveals the most cost-effective platform with the fastest inference time and higher accuracy of ML models. These findings provide valuable insights into selecting optimal cloud platforms for large-scale machine learning applications. The third study explores the capabilities of cutting-edge text-to-video models: Runway Gen-2, CogVideoX-2B, and CogVideoX-5B through mathematical evaluations (FID, FVD, and CLIP Score) and human perceptual assessments. The study emphasizes the importance of integrating human perception with mathematical evaluations for comprehensive assessments of generative AI models. Collectively, these three studies contribute to the broader understanding of AI's role in digital deception, evaluation, and generation. This thesis provides a holistic perspective on the challenges and advancements in AI-driven digital content by bridging the gap between AI-powered detection, cloud-based classification, and generative media assessments. The findings underscore the need for robust AI evaluation methodologies, balancing computational accuracy with human perception, to ensure ethical and effective deployment of AI technologies in digital deception landscapes.

Research Data and Supplementary Material

No

Download

Share

COinS