Narrow your search
Listing 1 - 4 of 4
Sort by

Dissertation
Master thesis : Evaluating and improving the robustness of machine learning models, using mixed-integer optimization techniques
Authors: --- --- --- ---
Year: 2019 Publisher: Liège Université de Liège (ULiège)

Loading...
Export citation

Choose an application

Bookmark

Abstract

Artificial neural networks are able to reach the highest accuracy on a great variety of complex visual tasks. Their impressive performances, often surpassing humans, attract a lot of interest. 

But their opaque nature makes them considered as distrusted black-box models by experts.

In 2013, Szegedy et al. discovered that images can be slightly modified to cause the models to classify them differently. The adversary creating the modified image can even choose the new class. These modified images, called adversarial examples, draw even more distrust on these models.

In this thesis, we will present methods to evaluate the robustness of a model against such examples, among which one based on mixed integer linear programming and others based on relaxations of it.
We will also present algorithms to train models to be more robust.
Finally, we will empirically evaluate models trained with these algorithms.


Book
Trustworthiness in Mobile Cyber Physical Systems
Authors: --- ---
Year: 2021 Publisher: Basel, Switzerland MDPI - Multidisciplinary Digital Publishing Institute

Loading...
Export citation

Choose an application

Bookmark

Abstract

Computing and communication capabilities are increasingly embedded in diverse objects and structures in the physical environment. They will link the ‘cyberworld’ of computing and communications with the physical world. These applications are called cyber physical systems (CPS). Obviously, the increased involvement of real-world entities leads to a greater demand for trustworthy systems. Hence, we use "system trustworthiness" here, which can guarantee continuous service in the presence of internal errors or external attacks. Mobile CPS (MCPS) is a prominent subcategory of CPS in which the physical component has no permanent location. Mobile Internet devices already provide ubiquitous platforms for building novel MCPS applications. The objective of this Special Issue is to contribute to research in modern/future trustworthy MCPS, including design, modeling, simulation, dependability, and so on. It is imperative to address the issues which are critical to their mobility, report significant advances in the underlying science, and discuss the challenges of development and implementation in various applications of MCPS.


Book
Trustworthiness in Mobile Cyber Physical Systems
Authors: --- ---
Year: 2021 Publisher: Basel, Switzerland MDPI - Multidisciplinary Digital Publishing Institute

Loading...
Export citation

Choose an application

Bookmark

Abstract

Computing and communication capabilities are increasingly embedded in diverse objects and structures in the physical environment. They will link the ‘cyberworld’ of computing and communications with the physical world. These applications are called cyber physical systems (CPS). Obviously, the increased involvement of real-world entities leads to a greater demand for trustworthy systems. Hence, we use "system trustworthiness" here, which can guarantee continuous service in the presence of internal errors or external attacks. Mobile CPS (MCPS) is a prominent subcategory of CPS in which the physical component has no permanent location. Mobile Internet devices already provide ubiquitous platforms for building novel MCPS applications. The objective of this Special Issue is to contribute to research in modern/future trustworthy MCPS, including design, modeling, simulation, dependability, and so on. It is imperative to address the issues which are critical to their mobility, report significant advances in the underlying science, and discuss the challenges of development and implementation in various applications of MCPS.

Keywords

Technology: general issues --- AUTOSAR --- DAG --- runnable scheduling --- control-scheduling codesign --- lagrange multiplier --- Mobile Cyber-Physical Systems (MCPS) --- industry --- Mobile Sink Groups (MSG) --- group mobility --- real-time data delivery --- Intelligent Transportation System (ITS) --- deep neural network --- Red Light Runner (RLR) --- dynamic signal control --- intersection safety --- mixed-criticality --- power-aware --- real-time scheduling --- DVFS --- security taxonomies --- event-based systems --- mobile cyber physical systems --- security flaws --- searchable encryption --- PEKS --- forward privacy --- trusted execution environment --- SGX --- data quality --- large-scale --- high-dimensionality --- linear discriminant analysis --- random projection --- bootstrapping --- controller area network bus --- authentication --- authenticity --- resiliency --- sustainability --- formal verification --- model checking --- in-vehicle network --- model compression --- adversarial robustness --- weight pruning --- adversarial training --- distillation --- embedded system --- secure AI --- AUTOSAR --- DAG --- runnable scheduling --- control-scheduling codesign --- lagrange multiplier --- Mobile Cyber-Physical Systems (MCPS) --- industry --- Mobile Sink Groups (MSG) --- group mobility --- real-time data delivery --- Intelligent Transportation System (ITS) --- deep neural network --- Red Light Runner (RLR) --- dynamic signal control --- intersection safety --- mixed-criticality --- power-aware --- real-time scheduling --- DVFS --- security taxonomies --- event-based systems --- mobile cyber physical systems --- security flaws --- searchable encryption --- PEKS --- forward privacy --- trusted execution environment --- SGX --- data quality --- large-scale --- high-dimensionality --- linear discriminant analysis --- random projection --- bootstrapping --- controller area network bus --- authentication --- authenticity --- resiliency --- sustainability --- formal verification --- model checking --- in-vehicle network --- model compression --- adversarial robustness --- weight pruning --- adversarial training --- distillation --- embedded system --- secure AI


Book
Trustworthiness in Mobile Cyber Physical Systems
Authors: --- ---
Year: 2021 Publisher: Basel, Switzerland MDPI - Multidisciplinary Digital Publishing Institute

Loading...
Export citation

Choose an application

Bookmark

Abstract

Computing and communication capabilities are increasingly embedded in diverse objects and structures in the physical environment. They will link the ‘cyberworld’ of computing and communications with the physical world. These applications are called cyber physical systems (CPS). Obviously, the increased involvement of real-world entities leads to a greater demand for trustworthy systems. Hence, we use "system trustworthiness" here, which can guarantee continuous service in the presence of internal errors or external attacks. Mobile CPS (MCPS) is a prominent subcategory of CPS in which the physical component has no permanent location. Mobile Internet devices already provide ubiquitous platforms for building novel MCPS applications. The objective of this Special Issue is to contribute to research in modern/future trustworthy MCPS, including design, modeling, simulation, dependability, and so on. It is imperative to address the issues which are critical to their mobility, report significant advances in the underlying science, and discuss the challenges of development and implementation in various applications of MCPS.

Listing 1 - 4 of 4
Sort by