My research focuses on practical machine learning and computer vision, with an emphasis on medical imaging, animal biometrics, accessible communication, multimodal systems, and creative AI.
Published Research
IEEE 5th International Conference on Computing and Machine Intelligence
Pre-trained Under Noise: A Framework for Robust Bone Fracture Detection in Medical Imaging
This study investigated how well pre-trained deep-learning models classify bone fractures in X-ray images when image quality varies. I developed a framework using transfer learning and controlled noise augmentation to simulate differences in medical imaging equipment and evaluate model robustness.
The results provide practical insight into model generalizability across healthcare settings, contributing to efforts to reduce healthcare disparities with more dependable AI-assisted diagnostics.
IEEE 5th International Conference on Computing and Machine Intelligence
Segmenting Photographs of Bat Wings Using U-Net Provides a Foundation for Recognizing Individuals
Working with the Ohio Bat Lab, I used a U-Net neural network to segment collagen-elastin bundle patterns in bat-wing photographs. The project explores a non-invasive alternative to physical banding for identifying individual bats.
Bat ecologists were able to use the segmented images to identify individuals accurately, establishing a foundation for a fully automated, photo-based recognition system. I have also presented the lab's AI research progress and FPV drone-tagging work at BatFest.
As part of my assistantship in the Applied Machine Learning and Intelligence Lab, I created a computer-vision model that detects static American Sign Language hand signs. The demonstration below focuses on hand shapes; signs that depend on motion require a temporal gesture-recognition approach.
Submitted to IEEE 3rd International Conference on Computing and Machine Intelligence
Empirical Evaluation of Signal Preprocessing in Electrocardiography Signal Classification
Data preprocessing can substantially affect machine-learning performance, but selecting an appropriate method remains difficult. This paper empirically evaluated scaling algorithms and their effects on electrocardiography signal-classification models used in disease-detection systems.
The paper was written from Winter 2023 through Spring 2024. It was submitted to the conference but was not published because university funding was unavailable.
Submitted to IJCAI-25, AI, Arts and Creativity Track
Detrended Fluctuation Analysis as Fitness Criterion for Music Generation by Cellular Automata
This project combined one-dimensional cellular automata, genetic programming, and detrended fluctuation analysis to evolve musical rhythms. Two genotype-to-phenotype mappings were compared to study how different evolutionary representations explored the system's creative possibilities.
Open-Source Foundation Models and Multimodal AI on HPC
On the University of Cincinnati's Advanced Research Computing Center cluster, I deployed open-source large language models with vLLM, built deep computer-vision pipelines, and experimented with multilingual speech-to-text using OpenAI Whisper for foreign-language film translation and captioning.
I also helped evaluate pre-prompted open-source foundation models for structured faculty analysis. I designed and compared prompting strategies whose results informed an administrative review of the approach's viability.
Conference Reviewership
I have served as a peer reviewer for the following artificial intelligence and machine-learning conferences, evaluating submissions for originality, technical soundness, relevance, and presentation while providing constructive feedback to authors:
IEEE 5th International Conference on Computing and Machine Intelligence
The 34th International Joint Conference on Artificial Intelligence (IJCAI-25)
IEEE 3rd International Conference on Computing and Machine Intelligence