International Conference on Machine Learning, Artificial Intelligence and Data Science

Hadiza Umar Yusuf

Biography

Hadiza Umar Yusuf is a highly accomplished researcher and educator in the fields of Machine Learning, Artificial Intelligence, and Cyber-Physical Systems. She is currently pursuing her Ph.D. in Computer and Information Sciences at the University of Michigan?Dearborn (2024?2028), where she is conducting empirical research on the falsification of AI-enabled cyber-physical systems under the supervision of Professor Khouloud Gaaloal. Her work focuses on categorizing CPS models, identifying verification challenges in AI systems, and proposing novel methods that combine stochastic optimization and reinforcement learning to enhance fault detection. Prior to this, she earned an MS in Engineering Artificial Intelligence from Carnegie Mellon University (2022?2024) in Kigali, Rwanda, and a B.Tech in Mathematics with a minor in Statistics from the Federal University of Technology, Minna, Nigeria (2014?2018).

Hadiza has been recognized with numerous honors and scholarships, including the CMU-Africa Student Excellent Award, Mastercard Foundation Scholars Impact Award, and accolades as the Best Female Graduating Student and Silver Medalist at the National Mathematical Competition for University Students in Abuja. Her research experience spans diverse projects, including prosthetics design for low-resource environments, adaptive education for African languages using large language models, polyphonic piano transcription, DTMF signal demodulation with neural networks, and AI applications for clean energy. She has published multiple research articles and maintains active contributions on GitHub, GitLab, and LinkedIn.

In addition to her research, Hadiza has extensive teaching experience. She has served as a teaching assistant at Carnegie Mellon University and the College of Education Katsina-Ala, an e-tutor at the Centre for Open Distance & eLearning, and an instructor at Merrion IT, mentoring students in mathematics, programming, graphic design, and web development. She has also participated in various talks and webinars, including the Ignite Research Academy, and volunteered at prestigious events such as the African Business Heroes Grand Finale, ICLR 2023, and STEM Women Lead Series in Kigali.

Hadiza is proficient in Python, MATLAB, Simulink, PyTorch, TensorFlow, and other ML frameworks, and is familiar with C/C++, OpenCV, NLTK, spaCy, Git, and cloud platforms including AWS, Azure, and GCP. Outside of her academic pursuits, she enjoys swimming, drawing, painting, and traveling, reflecting her well-rounded and dynamic personality.

Research Interest

Machine Learning for Cyber-Physical Systems, Reinforcement Learning,Trustworthy Artificial Intelligence,Software Verification and Testing

Abstract

Emerging Trends in Verification of AI-Enabled Cyber-Physical Systems" Under the AI in autonomous systems : Cyber-Physical Systems (CPS) are increasingly augmented with AI components to enhance control and decision-making in safety-critical domains. However, the integration of neural networks into CPS introduces verification challenges that traditional falsification methods fail to address effectively. Existing tools demonstrate strong performance on classical controllers, yet their efficacy diminishes significantly for AI-driven controllers due to undecidability and complex dynamics. My research investigates reinforcement learning? based falsification techniques as a means to systematically explore high-dimensional state spaces, identify potential safety violations, and benchmark verification methods across a wide range of CPS models. This work contributes toward closing the verification gap in AI-enabled CPS and advancing trustworthy system design.