Mathematical Modeling, AI Security, and Trustworthy Cyber-Physical Systems

Background

Artificial intelligence has rapidly evolved from predictive machine learning models into autonomous intelligent systems capable of reasoning, planning, decision-making, and interacting with external tools. These advances have introduced new opportunities across engineering, healthcare, cybersecurity, robotics, scientific computing, and critical infrastructure. However, they have also created significant security and trust challenges. Modern AI systems remain vulnerable to adversarial attacks, prompt injection, data poisoning, model manipulation, privacy leakage, insecure tool invocation, and autonomous decision failures. Mathematical modeling plays a fundamental role in understanding these risks, designing resilient architectures, evaluating robustness, and developing trustworthy AI systems capable of operating safely in complex environments. The symposium, which serves as a specialized session of the 4th International Conference on Applied Physics and Mathematical Modeling (CONF-APMM 2026), will focus on mathematics and AI security.

Goal/Rationale

The objective of this symposium is to bring together researchers and practitioners working at the intersection of mathematical modeling, artificial intelligence, cybersecurity, and engineering to explore methods for designing secure, robust, and trustworthy AI systems.

Recent advances in large language models, generative AI, agentic AI, reinforcement learning, optimization, and intelligent autonomous systems have transformed numerous scientific and engineering disciplines. At the same time, these technologies introduce novel attack surfaces that traditional cybersecurity methods cannot adequately address. Mathematical modeling offers powerful techniques for understanding AI behavior, quantifying uncertainty, modeling adversarial threats, optimizing defenses, and evaluating system resilience.

The symposium seeks contributions that combine theoretical advances with practical applications addressing AI robustness, adversarial machine learning, AI governance, secure autonomous agents, mathematical optimization, simulation, and trustworthy intelligent systems. Particular emphasis will be placed on interdisciplinary research that bridges applied mathematics, computer science, cybersecurity, engineering, and physics.

Scope

Topics of interest include, but are not limited to:

• AI Security
• Secure Artificial Intelligence Systems
• Trustworthy AI
• Agentic AI Security
• Autonomous Intelligent Systems
• Mathematical Modeling of AI Security
• Adversarial Machine Learning
• Prompt Injection Detection
• Secure Large Language Models (LLMs)
• Retrieval-Augmented Generation (RAG) Security
• AI Risk Assessment
• AI Governance and Compliance
• AI Red Teaming
• Explainable and Interpretable AI
• Federated Learning Security
• Privacy-Preserving Machine Learning
• AI Safety
• Cyber-Physical Systems Security
• Optimization Methods for AI
• Simulation of Intelligent Systems
• Reinforcement Learning Security
• Critical Infrastructure Protection
• Digital Twins and Secure Modeling
• Secure Multi-Agent Systems
• Mathematical Foundations of Trustworthy AI