Location
Innovation Lab Atrium
Start Date
4-3-2025 10:00 AM
End Date
4-3-2025 11:30 AM
Presentation Date
3 April 2025, 10:00am - 11:30am
Biography
Shane Cairns is a second-year PhD student in Computer Science and a Kummer Innovation and Entrepreneurship Doctoral Fellow at Missouri S&T. With a passion for advancing artificial intelligence, their research focuses on lifelong machine learning and explainability, aiming to develop systems that learn continuously and transparently over time. Shane holds a bachelor's degree in computer science from S&T and is committed to pushing the boundaries of machine learning to create robust, interpretable AI solutions. Outside of research, they enjoy soccer, chess, and hiking.
Meeting Name
2025 - Miners Solving for Tomorrow Research Conference
Department(s)
Computer Science
Second Department
Electrical and Computer Engineering
Document Type
Poster
Document Version
Final Version
File Type
event
Language(s)
English
Rights
© 2025 The Authors, All rights reserved
On the Robustness of Adaptive Resonance Theory Neural Networks
Innovation Lab Atrium
Comments
Advisor: Donald C. Wunsch
Abstract:
In this 5-minute presentation, we demonstrate how to generate adversarial examples targeting adaptive resonance theory predictive mapping (ARTMAP) neural networks, assessing ARTMAP's vulnerability to both white box and black box attacks. We reveal that ARTMAP, like other neural networks, is susceptible to white box adversarial perturbations, where full model access is assumed. We also explore its robustness under black box conditions with limited information. To counter these threats, we propose an adversarial training strategy tailored to ARTMAP's match-based learning. Experiments show this approach significantly enhances ARTMAP's resilience to FSGM attacks, with optimal results achieved when the perturbation strength (epsilon) is known in advance.