Out-of-distribution detection is not all you need

J Guérin, K Delmas, R Ferreira… - Proceedings of the AAAI …, 2023 - ojs.aaai.org
The usage of deep neural networks in safety-critical systems is limited by our ability to
guarantee their correct behavior. Runtime monitors are components aiming to identify …

Smarla: A safety monitoring approach for deep reinforcement learning agents

A Zolfagharian, M Abdellatif, LC Briand… - IEEE Transactions on …, 2024 - ieeexplore.ieee.org
Deep Reinforcement Learning (DRL) has made significant advancements in various fields,
such as autonomous driving, healthcare, and robotics, by enabling agents to learn optimal …

Evaluation of runtime monitoring for UAV emergency landing

J Guerin, K Delmas, J Guiochet - … International Conference on …, 2022 - ieeexplore.ieee.org
To certify UAV operations in populated areas, risk mitigation strategies-such as Emergency
Landing (EL)-must be in place to account for potential failures. EL aims at reducing ground …

Unifying evaluation of machine learning safety monitors

J Guerin, RS Ferreira, K Delmas… - 2022 IEEE 33rd …, 2022 - ieeexplore.ieee.org
With the increasing use of Machine Learning (ML) in critical autonomous systems, runtime
monitors have been developed to detect prediction errors and keep the system in a safe …

Monitizer: automating design and evaluation of neural network monitors

M Azeem, M Grobelna, S Kanav, J Křetínský… - … on Computer Aided …, 2024 - Springer
The behavior of neural networks (NNs) on previously unseen types of data (out-of-
distribution or OOD) is typically unpredictable. This can be dangerous if the network's output …

Uncertainty quantification for deep neural networks: An empirical comparison and usage guidelines

M Weiss, P Tonella - Software Testing, Verification and …, 2023 - Wiley Online Library
Deep neural networks (DNN) are increasingly used as components of larger software
systems that need to process complex data, such as images, written texts, audio/video …

MarMot: Metamorphic Runtime Monitoring of Autonomous Driving Systems

J Ayerdi, A Iriarte, P Valle, I Roman… - ACM Transactions on …, 2024 - dl.acm.org
Autonomous Driving Systems (ADSs) are complex Cyber-Physical Systems (CPSs) that
must ensure safety even in uncertain conditions. Modern ADSs often employ Deep Neural …

Robust traffic sign recognition against camera failures

M Atif, A Ceccarelli, T Zoppi, M Gharib… - IEEE Open Journal of …, 2022 - ieeexplore.ieee.org
Failures of the vehicle camera may compromise the correct acquisition of frames, that are
subsequently used by autonomous driving tasks. A clear understanding of the behavior of …

Metamorphic runtime monitoring of autonomous driving systems

J Ayerdi, A Iriarte, P Valle, I Roman… - arXiv preprint arXiv …, 2023 - arxiv.org
Autonomous Driving Systems (ADSs) are complex Cyber-Physical Systems (CPSs) that
must ensure safety even in uncertain conditions. Modern ADSs often employ Deep Neural …

Inherent Diverse Redundant Safety Mechanisms for AI-based Software Elements in Automotive Applications

M Pitale, A Abbaspour, D Upadhyay - arXiv preprint arXiv:2402.08208, 2024 - arxiv.org
This paper explores the role and challenges of Artificial Intelligence (AI) algorithms,
specifically AI-based software elements, in autonomous driving systems. These AI systems …