Manel Abdellatif

About

Manel Abdellatif is a researcher specializing in the testing, verification, and safety assurance of Deep Reinforcement Learning (DRL) systems, with a particular focus on their deployment in safety-critical environments such as autonomous driving, robotics, and healthcare. Her work addresses one of the most pressing challenges in modern AI: ensuring that DRL agents behave reliably and safely when exposed to real-world conditions. Abdellatif's most influential contribution, "A Search-Based Testing Approach for Deep Reinforcement Learning Agents," has garnered 46 citations since its 2023 publication, establishing her as a notable voice in the emerging field of AI quality assurance. By applying search-based techniques to systematically expose vulnerabilities in DRL agents, she offers a rigorous methodology for uncovering failure scenarios before deployment. Complementing this work, her SMARLA framework introduces a proactive safety monitoring approach, enabling real-time oversight of DRL agents operating in high-stakes domains. Through these contributions, Abdellatif bridges the gap between cutting-edge machine learning and software engineering principles, providing the research community with practical tools to make autonomous, learning-based systems more trustworthy. Her growing body of work positions her as an important contributor to the responsible development of AI.

Research Focus

Key Achievements

Top Papers

1. 1
   A Search-Based Testing Approach for Deep Reinforcement Learning Agents

   46 citations · 2023
2. 2
   SMARLA: A Safety Monitoring Approach for Deep Reinforcement Learning Agents

   7 citations · 2024
3. 3
   A Search-Based Testing Approach for Deep Reinforcement Learning Agents

   7 citations · 2022
4. 4
   SMARLA: A Safety Monitoring Approach for Deep Reinforcement Learning Agents

   2 citations · 2023

Key Collaborators

Contact & Links