-
Robustness Assessment of a Runway Object Classifier for Safe Aircraft Taxiing
Authors:
Yizhak Elboher,
Raya Elsaleh,
Omri Isac,
Mélanie Ducoffe,
Audrey Galametz,
Guillaume Povéda,
Ryma Boumazouza,
Noémie Cohen,
Guy Katz
Abstract:
As deep neural networks (DNNs) are becoming the prominent solution for many computational problems, the aviation industry seeks to explore their potential in alleviating pilot workload and in improving operational safety. However, the use of DNNs in this type of safety-critical applications requires a thorough certification process. This need can be addressed through formal verification, which pro…
▽ More
As deep neural networks (DNNs) are becoming the prominent solution for many computational problems, the aviation industry seeks to explore their potential in alleviating pilot workload and in improving operational safety. However, the use of DNNs in this type of safety-critical applications requires a thorough certification process. This need can be addressed through formal verification, which provides rigorous assurances -- e.g.,~by proving the absence of certain mispredictions. In this case-study paper, we demonstrate this process using an image-classifier DNN currently under development at Airbus and intended for use during the aircraft taxiing phase. We use formal methods to assess this DNN's robustness to three common image perturbation types: noise, brightness and contrast, and some of their combinations. This process entails multiple invocations of the underlying verifier, which might be computationally expensive; and we therefore propose a method that leverages the monotonicity of these robustness properties, as well as the results of past verification queries, in order to reduce the overall number of verification queries required by nearly 60%. Our results provide an indication of the level of robustness achieved by the DNN classifier under study, and indicate that it is considerably more vulnerable to noise than to brightness or contrast perturbations.
△ Less
Submitted 28 June, 2024; v1 submitted 8 January, 2024;
originally announced February 2024.
-
Surrogate Neural Networks Local Stability for Aircraft Predictive Maintenance
Authors:
Mélanie Ducoffe,
Guillaume Povéda,
Audrey Galametz,
Ryma Boumazouza,
Marion-Cécile Martin,
Julien Baris,
Derk Daverschot,
Eugene O'Higgins
Abstract:
Surrogate Neural Networks are nowadays routinely used in industry as substitutes for computationally demanding engineering simulations (e.g., in structural analysis). They allow to generate faster predictions and thus analyses in industrial applications e.g., during a product design, testing or monitoring phases. Due to their performance and time-efficiency, these surrogate models are now being de…
▽ More
Surrogate Neural Networks are nowadays routinely used in industry as substitutes for computationally demanding engineering simulations (e.g., in structural analysis). They allow to generate faster predictions and thus analyses in industrial applications e.g., during a product design, testing or monitoring phases. Due to their performance and time-efficiency, these surrogate models are now being developed for use in safety-critical applications. Neural network verification and in particular the assessment of their robustness (e.g., to perturbations) is the next critical step to allow their inclusion in real-life applications and certification. We assess the applicability and scalability of empirical and formal methods in the context of aircraft predictive maintenance for surrogate neural networks designed to predict the stress sustained by an aircraft part from external loads. The case study covers a high-dimensional input and output space and the verification process thus accommodates multi-objective constraints. We explore the complementarity of verification methods in assessing the local stability property of such surrogate models to input noise. We showcase the effectiveness of sequentially combining methods in one verification 'pipeline' and demonstrating the subsequent gain in runtime required to assess the targeted property.
△ Less
Submitted 5 June, 2024; v1 submitted 11 January, 2024;
originally announced January 2024.
-
The Habitable Zone of Inhabited Planets
Authors:
Jorge I. Zuluaga,
Juan F. Salazar,
Pablo Cuartas-Restrepo,
German Poveda
Abstract:
In this paper we discuss and illustrate the hypothesis that life substantially alters the state of a planetary environment and therefore, modifies the limits of the HZ as estimated for an uninhabited planet. This hypothesis lead to the introduction of the Habitable Zone for Inhabited planets (hereafter InHZ), defined here as the region where the complex interaction between life and its abiotic env…
▽ More
In this paper we discuss and illustrate the hypothesis that life substantially alters the state of a planetary environment and therefore, modifies the limits of the HZ as estimated for an uninhabited planet. This hypothesis lead to the introduction of the Habitable Zone for Inhabited planets (hereafter InHZ), defined here as the region where the complex interaction between life and its abiotic environment is able to produce plausible equilibrium states with the necessary physical conditions for the existence and persistence of life itself. We support our hypothesis of an InHZ with three theoretical arguments, multiple evidences coming from observations of the Earth system, several conceptual experiments and illustrative numerical simulations. Conceptually the diference between the InHZ and the Abiotic HZ (AHZ) depends on unique and robust properties of life as an emergent physical phenomenon and not necesarily on the particular life forms bearing in the planet. Our aim here is to provide conceptual basis for the development of InHZ models incorporating consistently life-environment interactions. Although previous authors have explored the effects of life on habitability there is a gap in research develo** the reasons why life should be systematically included at determining the HZ limits. We do not provide here definitive limits to the InHZ but we show through simple numerical models (as a parable of an inhabited planet) how the limits of the AHZ could be modified by including plausible interactions between biota and its environment. These examples aim also at posing the question that if limits of the HZ could be modified by the presence of life in those simple dynamical systems how will those limits change if life is included in established models of the AHZ.
△ Less
Submitted 18 May, 2014;
originally announced May 2014.