Supporting DNN Safety Analysis and Retraining through Heatmap-based Unsupervised Learning

in IEEE Transactions on Reliability (2021), 70(4), 1641-1657

Deep neural networks (DNNs) are increasingly im- portant in safety-critical systems, for example in their perception layer to analyze images. Unfortunately, there is a lack of methods to ensure the ... [more ▼]

Deep neural networks (DNNs) are increasingly im- portant in safety-critical systems, for example in their perception layer to analyze images. Unfortunately, there is a lack of methods to ensure the functional safety of DNN-based components. We observe three major challenges with existing practices regarding DNNs in safety-critical systems: (1) scenarios that are underrepresented in the test set may lead to serious safety violation risks, but may, however, remain unnoticed; (2) char- acterizing such high-risk scenarios is critical for safety analysis; (3) retraining DNNs to address these risks is poorly supported when causes of violations are difficult to determine. To address these problems in the context of DNNs analyzing images, we propose HUDD, an approach that automatically supports the identification of root causes for DNN errors. HUDD identifies root causes by applying a clustering algorithm to heatmaps capturing the relevance of every DNN neuron on the DNN outcome. Also, HUDD retrains DNNs with images that are automatically selected based on their relatedness to the identified image clusters. We evaluated HUDD with DNNs from the automotive domain. HUDD was able to identify all the distinct root causes of DNN errors, thus supporting safety analysis. Also, our retraining approach has shown to be more effective at improving DNN accuracy than existing approaches. [less ▲]

Supporting Families and Children Beyond COVID-19: Social protection in Southern and Eastern Europe and Central Asia

Report (2020)

Supporting Grocery Shopping to Achieve a Healthy and Sustainable Diet – How Developing a Behavioural Theory Informs Dynamic Smartphone Applications

Doctoral thesis (2021)

Health and sustainability are becoming increasingly important in current lifestyles. In this context healthy and sustainable grocery shopping is one key aspect to facilitate a balanced and environmentally ... [more ▼]

Health and sustainability are becoming increasingly important in current lifestyles. In this context healthy and sustainable grocery shopping is one key aspect to facilitate a balanced and environmentally friendly diet. Many people are interested in changing their habits to become healthier and to consider their impact on the environment through the choices they make. But many do not consider where a healthy and sustainable diet starts. In other words, people frequently have a vague idea that grocery shopping is an important aspect of a healthy and sustainable lifestyle, but they lack sufficient knowledge and action plans to act accordingly. Therefore, the observable behaviour in many cases shows what is called the intention-action/behaviour gap, the attitude-behaviour gap, or the knowing-doing gap (Ajzen, 2016; Grunert, 2011; Hoek, Pearson, James, Lawrence, & Friel, 2017; de Schutter, 2015; Bailey & Harper, 2015). To break this deadlock people, who are interested in such a lifestyle change, need the required information and support to create appropriate action plans to lead them through their grocery shopping without incurring excessive cognitive impact. The risk of such cognitive strain is that people give up easily on their good intentions and fall back into old unhealthy or environmentally impacting habits. Smartphones are ubiquitous and therefore could potentially solve many of these problems, but the design of suitable applications is mostly ad-hoc and not based on thorough modelling. On the other hand, existing behavioural models are considered to be too static and not up to the task of dynamically assessing and influencing behaviour as would be required by a smartphone-based intervention (Riley, et al., 2011; Spruijt-Metz & Nilsen, 2014). To address these problems this work proposes three major contributions: first, a novel comprehensive model of behaviour built on well-established theories used in psychology and the social science. The novelty is the consistent integration of well-proven pre-existing theories into one single comprehensive model that aims to capture the benefits and tries to overcome the limitations of each base theory. Based on this model, the second contribution of this work is the evaluation of motivation and intention to buy healthy and sustainable groceries. It has been found that health is more important than sustainability in this regard, and that health-related goals are easier to act on than sustainability related goals resulting in a bigger intention-action gap for sustainable grocery shopping. To address these issues, the third major contribution of this work is a model-derived design framework for smartphone-based interventions that provides comprehensive guidelines for developing applications to assess and support a specific behaviour, such as grocery shopping, while at the same time aiming at addressing a superordinate issue, such as health and sustainability. [less ▲]

Supporting non-target identification by adding hydrogen deuterium exchange MS/MS capabilities to MetFrag

in Analytical and bioanalytical chemistry (2019), 411(19), 4683-4700

Liquid chromatography coupled with high-resolution mass spectrometry (LC-HRMS) is increasingly popular for the non-targeted exploration of complex samples, where tandem mass spectrometry (MS/MS) is used ... [more ▼]

Liquid chromatography coupled with high-resolution mass spectrometry (LC-HRMS) is increasingly popular for the non-targeted exploration of complex samples, where tandem mass spectrometry (MS/MS) is used to characterize the structure of unknown compounds. However, mass spectra do not always contain sufficient information to unequivocally identify the correct structure. This study investigated how much additional information can be gained using hydrogen deuterium exchange (HDX) experiments. The exchange of “easily exchangeable” hydrogen atoms (connected to heteroatoms), with predominantly [M+D]+ ions in positive mode and [M-D]− in negative mode was observed. To enable high-throughput processing, new scoring terms were incorporated into the in silico fragmenter MetFrag. These were initially developed on small datasets and then tested on 762 compounds of environmental interest. Pairs of spectra (normal and deuterated) were found for 593 of these substances (506 positive mode, 155 negative mode spectra). The new scoring terms resulted in 29 additional correct identifications (78 vs 49) for positive mode and an increase in top 10 rankings from 80 to 106 in negative mode. Compounds with dual functionality (polar head group, long apolar tail) exhibited dramatic retention time (RT) shifts of up to several minutes, compared with an average 0.04 min RT shift. For a smaller dataset of 80 metabolites, top 10 rankings improved from 13 to 24 (positive mode, 57 spectra) and from 14 to 31 (negative mode, 63 spectra) when including HDX information. The results of standard measurements were confirmed using targets and tentatively identified surfactant species in an environmental sample collected from the river Danube near Novi Sad (Serbia). The changes to MetFrag have been integrated into the command line version available at http://c-ruttkies.github.io/MetFrag and all resulting spectra and compounds are available in online resources and in the Electronic Supplementary Material (ESM). [less ▲]

Supporting Security Protocols on CAN-Based Networks

in 2017 IEEE 18th International Conference on Industrial Technology (ICIT2017) (2017)

The ever-increasing variety of services built on top of the Controller Area Network (CAN), along with the recent discovery of vulnerabilities in CAN-based automotive systems (some of them demonstrated in ... [more ▼]

The ever-increasing variety of services built on top of the Controller Area Network (CAN), along with the recent discovery of vulnerabilities in CAN-based automotive systems (some of them demonstrated in practice) stimulated a renewed attention to security-oriented enhancements of the CAN protocol. The issue is further compounded nowadays because, unlike in the past, security can no longer be enforced by physical bus segregation. This paper describes how CAN XR, a recently proposed extension of the CAN data-link layer, can effectively support the distributed calculation of arbitrary binary Boolean functions, which are the foundation of most security protocols, without necessarily disclosing their operands on the bus. The feasibility of the approach is then shown through experimental evaluation and by confirming its applicability to a shared key generation protocol proposed in literature. [less ▲]

Supporting Stress Testing in Real-Time Systems with Constraint Programming

Doctoral thesis (2015)

Failures in safety-critical Real-Time Embedded Systems (RTES) could result in catastrophic consequences for the system itself, its users, and the environment. Therefore, these systems are subject to ... [more ▼]

Failures in safety-critical Real-Time Embedded Systems (RTES) could result in catastrophic consequences for the system itself, its users, and the environment. Therefore, these systems are subject to strict performance requirements specifying constraints on real-time properties such as task deadlines, response time and CPU usage. Lately, RTES have been shifting towards multi-threaded application design, highly configurable operating systems, and multi-core architectures for computing platforms. The concurrent nature of their operating environment also entails that the order of external events triggering RTES tasks is often unpredictable. Such complexity in the system architecture, concurrency, and environment renders performance testing increasingly challenging. Specifically, computing input combinations that are intended to violate performance requirements, i.e., stress testing, is one of the preferred ways for verifying RTES performance. These input combinations are referred to as stress test cases, and, upon execution, are predicted to result in worst-case scenarios with respect to a performance requirement. In RTES, stress test cases are usually characterized by sequences of arrival times for aperiodic tasks in the subject system. Generating stress test cases is challenging because it is hard to predict how a particular sequence of arrival times will affect the system performance, and because the set of all arrival times for aperiodic tasks quickly grows as the system size increases. For this reason, search strategies based on Genetic Algorithms (GA) have been used to find stress test cases with high chances of violating performance requirements. For practical use, software testing has to accommodate time and budget constraints. In the context of stress testing, it is essential to investigate the trade-off between the time needed to generate test cases (efficiency), their capability to reveal scenarios that violate performance requirements (effectiveness), and to cover different scenarios where these violations arise (diversity). Even though GA are efficient, and capable of finding diverse solutions, they explore only part of the search space, and their effectiveness depends on configuration parameters. This aspect justifies considering alternative strategies, such as Constraint Programming (CP), that explore the search space completely. Furthermore, to enable effective industrial application, stress testing has to be capable of seamless integration in the development cycle of companies. Therefore, it is both important to capture specific system and contextual properties in a conceptual model, and to map such conceptual model in a standard Model Driven Engineering (MDE) language such as UML/MARTE. In this thesis, we address the challenges above by presenting a practical approach, based on CP, to support performance stress testing in RTES. Specifically, we make the following contributions: (1) a conceptual model, mapped to UML/MARTE, which captures the abstractions required to generate stress test cases, (2) a constraint optimization model to generate such test cases, and (3) a combined GA+CP stress testing strategy that achieves a practical trade-off between efficiency, effectiveness and diversity. The validation of our work shows that (1) the conceptual model can be applied with a reasonable overhead in an industrial settings, (2) CP is able to effectively identify worst-case scenarios with respect to task deadlines, response time, and CPU usage, and (3) the combined GA+CP strategy is more likely than GA and CP in isolation to scale to large and complex systems. The work in this thesis opens up the exploration of further directions, involving the use of multi-objective optimization to generate stress test cases that simultaneously exercise different performance properties of the system, and of MiniMax analysis to derive design and configuration guidelines that minimize the risk to violate performance requirements at runtime. [less ▲]

Supporting the Verification of Compliance to Safety Standards via Model-Driven Engineering: Approach, Tool-Support and Empirical Validation

in Information and Software Technology (2013), 55(1), 836-864

Supporting Youth in African Countries to Advance Local Economies and Community Health: The SDG Lab on Microfinance for Youth and Clean Water

Report (2018)

In the sub-Saharan African countries, a large number of young adults enters the labour market each year. Scarcity of regular employment opportunities and the wish to become an entrepreneur lead many young ... [more ▼]

In the sub-Saharan African countries, a large number of young adults enters the labour market each year. Scarcity of regular employment opportunities and the wish to become an entrepreneur lead many young people to start their own business. However, young people are often not able to become regular microloan customers due to both higher risks associated with young age and lack of experience with managing finances. If microfinance products should become accessible to young people, the loans need to be accompanied by non-financial services, i.e., financial advice and mentoring. In order to advance local economies and community health, we see two distinct problems around microfinance products for young adults. First, microfinance products combined with non-financial services are not sustainable, i.e. additional external funds are needed that render these microloan products unprofitable in the long run. Second, if improvements in community water, development, and health are envisaged, then new microfinance products need to be designed to serve the purpose of supporting the SDG goals of clean water and sanitation for all. We used an existing initiative of microfinance for young entrepreneurs and applied the social innovation lab methodology to gather experts in relevant fields. The SDG lab, co-sponsored by Future Earth and Appui au Développement Autonome Microfinance Luxembourg and hosted by the University of Luxembourg, first addressed the problem of sustainability of microfinance products for young entrepreneurs. Second, the SDG lab defined actors, processes, and goals to design microfinance products for young people to support the SDG goals of clean water and sanitation for all. [less ▲]

Suppression of Decoherence in Fast-Atom Diffraction at Surfaces

in Physical Review Letters (2008), 101(25),

Scattering of fast neutral atoms with keV kinetic energies at alkali-halide surfaces under grazing angles displays intriguing diffraction patterns. The surprisingly strong persistence of quantum coherence ... [more ▼]

Scattering of fast neutral atoms with keV kinetic energies at alkali-halide surfaces under grazing angles displays intriguing diffraction patterns. The surprisingly strong persistence of quantum coherence despite the impulsive interaction with an environment at solid state density and elevated temperatures raises fundamental questions such as to the suppression of decoherence and of the quantum-to-classical crossover. We present an ab initio simulation of the quantum diffraction of fast helium beams at a LiF (100) surface in the < 110 > direction and compare with recent experimental diffraction data. From the quantitative reconstruction of diffraction images the vertical LiF-surface reconstruction, or buckling, can be determined. [less ▲]

Suppression of the cooperative Jahn-Teller distortion and its effect on the Raman octahedra-rotation modes of TbMn1-xFexO3

in Physical Review. B (2018), 97(14),

This work reports the changes in structure and lattice dynamics induced by substituting the Jahn-Teller-active Mn3+ ion by the Jahn-Teller-inactive Fe3+ in TbMn1-xFexO3 over the full composition range ... [more ▼]

This work reports the changes in structure and lattice dynamics induced by substituting the Jahn-Teller-active Mn3+ ion by the Jahn-Teller-inactive Fe3+ in TbMn1-xFexO3 over the full composition range. The structural analysis reveals that the amplitude of the 0 (pure 0.5, where it is completely suppressed. We then correlate this evolution with the behavior of the Raman modes across the solid solution. In particular, we show that the Raman modes associated with the rotation of octahedra, whose wave number is commonly considered to scale linearly with the tilt angles in orthorhombic Pnma perovskites are also sensitive to the amplitude of the Jahn-Teller distortion. [less ▲]