Search

The degree of inter-rater agreement is usually assessed through (Formula presented.) -type coefficients and the extent of agreement is then characterized by comparing the value of the adopted coefficient against a benchmark scale. Through two motivating examples, it is displayed the different behavior of some (Formula presented.) -type coefficients due to asymmetric distribution of marginal frequencies over categories. In order to investigate the robustness of four (Formula presented.) -type coefficients for nominal and ordinal classifications and of an inferential benchmarking procedure that, differently from straightforward benchmarking, does not neglect the influence of the experimental conditions, an extensive Monte Carlo simulation study has been conducted. The robustness has been investigated for several scenarios, differing for sample size, rating scale dimension, number of raters, frequency distribution of rater classifications, pattern of agreement across raters. Simulation results reveal an higher paradoxical behavior of Fleiss kappa and Conger kappa with ordinal rather than nominal classifications; the coefficients robustness improves with increasing sample size and number of raters for both nominal and ordinal classifications whereas robustness improves with rating scale dimension only for nominal classifications. By identifying the scenarios (ie, minimum sample size, number of raters, rating scale dimension) with acceptable robustness, this study provides guidelines about the design of robust agreement studies.

Le reti neurali profonde (DNNs) sono potenti modelli predittivi, che superano le capacità umane in una varietà di task. Imparano sistemi decisionali complessi e flessibili dai dati a disposizione e raggiungono prestazioni eccezionali in molteplici campi di apprendimento automatico, dalle applicazioni dell'intelligenza artificiale, come il riconoscimento di immagini, parole e testi, alle scienze più tradizionali, tra cui medicina, fisica e biologia. Nonostante i risultati eccezionali, le prestazioni elevate e l’alta precisione predittiva non sono sufficienti per le applicazioni nel mondo reale, specialmente in ambienti critici per la sicurezza, dove l'utilizzo dei DNNs è fortemente limitato dalla loro natura black-box. Vi è una crescente necessità di comprendere come vengono eseguite le predizioni, fornire stime di incertezza, garantire robustezza agli attacchi avversari e prevenire comportamenti indesiderati. Anche le migliori architetture sono vulnerabili a piccole perturbazioni nei dati di input, note come attacchi avversari: manipolazioni malevole degli input che sono percettivamente indistinguibili dai campioni originali ma sono in grado di ingannare il modello in predizioni errate. In questo lavoro, dimostriamo che tale fragilità è correlata alla geometria del manifold dei dati ed è quindi probabile che sia una caratteristica intrinseca delle predizioni dei DNNs. Questa condizione suggerisce una possibile direzione al fine di ottenere robustezza agli attacchi: studiamo la geometria degli attacchi avversari nel limite di un numero infinito di dati e di pesi per le reti neurali Bayesiane, dimostrando che, in questo limite, sono immuni agli attacchi avversari gradient-based. Inoltre, proponiamo alcune tecniche di training per migliorare la robustezza delle architetture deterministiche. In particolare, osserviamo sperimentalmente che ensembles di reti neurali addestrati su proiezioni casuali degli input originali in spazi basso-dimensionali sono più resistenti agli attacchi. Successivamente, ci concentriamo sul problema dell'interpretabilità delle predizioni delle reti nel contesto delle saliency-based explanations. Analizziamo la stabilità delle explanations soggette ad attacchi avversari e dimostriamo che, nel limite di un numero infinito di dati e di pesi, le interpretazioni Bayesiane sono più stabili di quelle fornite dalle reti deterministiche. Confermiamo questo comportamento in modo sperimentale nel regime di un numero finito di dati. Infine, introduciamo il concetto di attacco avversario alle sequenze di amminoacidi per protein Language Models (LM). I modelli di Deep Learning per la predizione della struttura delle proteine, come AlphaFold2, sfruttano le architetture Transformer e il loro meccanismo di attention per catturare le proprietà strutturali e funzionali delle sequenze di amminoacidi. Nonostante l'elevata precisione delle predizioni, perturbazioni biologicamente piccole delle sequenze di input, o anche mutazioni di un singolo amminoacido, possono portare a strutture 3D sostanzialmente diverse. Al contempo, i protein LMs sono insensibili alle mutazioni che inducono misfolding o disfunzione (ad esempio le missense mutations). In particolare, le predizioni delle coordinate 3D non rivelano l'effetto di unfolding indotto da queste mutazioni. Pertanto, esiste un'evidente incoerenza tra l'importanza biologica delle mutazioni e il conseguente cambiamento nella predizione strutturale. Ispirati da questo problema, introduciamo il concetto di perturbazione avversaria delle sequenze proteiche negli embedding continui dei protein LMs. Il nostro metodo utilizza i valori di attention per rilevare le posizioni degli amminoacidi più vulnerabili nelle sequenze di input. Le mutazioni avversarie sono biologicamente diverse dalle sequenze di riferimento e sono in grado di alterare in modo significativo le strutture 3D. Deep Neural Networks (DNNs) are powerful predictive models, exceeding human capabilities in a variety of tasks. They learn complex and flexible decision systems from the available data and achieve exceptional performances in multiple machine learning fields, spanning from applications in artificial intelligence, such as image, speech and text recognition, to the more traditional sciences, including medicine, physics and biology. Despite the outstanding achievements, high performance and high predictive accuracy are not sufficient for real-world applications, especially in safety-critical settings, where the usage of DNNs is severely limited by their black-box nature. There is an increasing need to understand how predictions are performed, to provide uncertainty estimates, to guarantee robustness to malicious attacks and to prevent unwanted behaviours. State-of-the-art DNNs are vulnerable to small perturbations in the input data, known as adversarial attacks: maliciously crafted manipulations of the inputs that are perceptually indistinguishable from the original samples but are capable of fooling the model into incorrect predictions. In this work, we prove that such brittleness is related to the geometry of the data manifold and is therefore likely to be an intrinsic feature of DNNs’ predictions. This negative condition suggests a possible direction to overcome such limitation: we study the geometry of adversarial attacks in the large-data, overparameterized limit for Bayesian Neural Networks and prove that, in this limit, they are immune to gradient-based adversarial attacks. Furthermore, we propose some training techniques to improve the adversarial robustness of deterministic architectures. In particular, we experimentally observe that ensembles of NNs trained on random projections of the original inputs into lower dimensional spaces are more resilient to the attacks. Next, we focus on the problem of interpretability of NNs’ predictions in the setting of saliency-based explanations. We analyze the stability of the explanations under adversarial attacks on the inputs and we prove that, in the large-data and overparameterized limit, Bayesian interpretations are more stable than those provided by deterministic networks. We validate this behaviour in multiple experimental settings in the finite data regime. Finally, we introduce the concept of adversarial perturbations of amino acid sequences for protein Language Models (LMs). Deep Learning models for protein structure prediction, such as AlphaFold2, leverage Transformer architectures and their attention mechanism to capture structural and functional properties of amino acid sequences. Despite the high accuracy of predictions, biologically small perturbations of the input sequences, or even single point mutations, can lead to substantially different 3d structures. On the other hand, protein language models are insensitive to mutations that induce misfolding or dysfunction (e.g. missense mutations). Precisely, predictions of the 3d coordinates do not reveal the structure-disruptive effect of these mutations. Therefore, there is an evident inconsistency between the biological importance of mutations and the resulting change in structural prediction. Inspired by this problem, we introduce the concept of adversarial perturbation of protein sequences in continuous embedding spaces of protein language models. Our method relies on attention scores to detect the most vulnerable amino acid positions in the input sequences. Adversarial mutations are biologically diverse from their references and are able to significantly alter the resulting 3D structures.

In this letter, the authors have assessed the robustness of an Economic Nonlinear Model-Predictive Controller (ENMPC) aimed at maximizing the power production of wind turbines. The scope of this letter is to quantify the sensitivity of this type of controller concerning wind conditions, climate, wind speed prediction unavailability, and aerodynamic performance degradation. A power production controller's robustness is crucial for the wind turbine industry due to the extreme variability of external conditions and the wear caused by long-term continuous operativity. Model-Predictive controllers are, in principle, more prone to robustness issues concerning standard controllers, a fact that limits their adoption on actual wind turbines. The analysis is performed with the fully-aeroelastic solver OpenFAST considering a wide set of realistic load cases. It is demonstrated that the ENMPC previously developed is robust to wind prediction unavailability and change in wind turbulence intensity. Conversely, it is not robust to the modelling error due to aerodynamic degradation. Indeed, a reduction in generated power concerning the reference controller is observed, especially for operating region two and end-life blades. Finally, a significant increase in power production is achieved considering the external temperature variation thanks to the ENMPC's direct handling of the generator temperature constraint.

We introduce the Robust Logistic Zero-Sum Regression (RobLZS) estimator, which can be used for a two-class problem with high-dimensional compositional covariates. Since the log-contrast model is employed, the estimator is able to do feature selection among the compositional parts. The proposed method attains robustness by minimizing a trimmed sum of deviances. A comparison of the performance of the RobLZS estimator with a non-robust counterpart and with other sparse logistic regression estimators is conducted via Monte Carlo simulation studies. Two microbiome data applications are considered to investigate the stability of the estimators to the presence of outliers. Robust Logistic Zero-Sum Regression is available as an R package that can be downloaded at https://github.com/giannamonti/RobZS.

Microbiome data analysis often relies on the identification of a subset of potential biomarkers associated with a clinical outcome of interest. Robust ZeroSum regression, an elastic-net penalized compositional regression built on the least trimmed squares estimator, is a variable selection procedure capable to cope with the high dimensionality of these data, their compositional nature, and, at the same time, it guarantees robustness against the presence of outliers. The necessity of discovering “true” effects and to improve clinical research quality and reproducibility has motivated us to propose a two-step robust compositional knockoff filter procedure, which allows selecting the set of relevant biomarkers, among the many measured features having a nonzero effect on the response, controlling the expected fraction of false positives. We demonstrate the effectiveness of our proposal in an extensive simulation study, and illustrate its usefulness in an application to intestinal microbiome analysis.

Resilience and the efficiency of transportation systems are crucial for the economic development of geographical areas, and network analysis applied to railways can provide insight into the importance of branch lines and their impacts on the entire system. This paper explores the behavior of the ERC measure, a local robustness measure, on the railway network in Lombardy, Italy, and analyzes the impacts of deactivating stations or journeys on the network’s robustness. Changes in the topological properties of the network were studied by simulating potential external disturbances and analyzing the impact of deleting the most connected stations or railway lines. The numerical results show how the measures provided effectively identify critical stations and journeys within the network structure and outperform classical topological metrics. Since ERC measures take into account all of the alternative paths present in the network, they can provide valuable information for rerouting traffic along alternative paths in case of failures or disruptions. The paper’s original contribution lies in demonstrating the effectiveness of the ERC measure in identifying critical stations and journeys within the network structure.

The main goal of this paper is to show that Bayesian optimization could be regarded as a general framework for the data driven modelling and solution of problems arising in water distribution systems. Hydraulic simulation, both scenario based, and Monte Carlo is a key tool in modelling in water distribution systems. The related optimization problems fall in a simulation/optimization framework in which objectives and constraints are often black-box. Bayesian Optimization (BO) is characterized by a surrogate model, usually a Gaussian process, but also a random forest and increasingly neural networks and an acquisition function which drives the search for new evaluation points. These modelling options make BO nonparametric, robust, flexible and sample efficient particularly suitable for simulation/optimization problems. A defining characteristic of BO is its versatility and flexibility, given for instance by different probabilistic models, in particular different kernels, different acquisition functions. These characteristics of the Bayesian optimization approach are exemplified by the two problems: cost/energy optimization in pump scheduling and optimal sensor placement for early detection on contaminant intrusion. Different surrogate models have been used both in explicit and implicit control schemes. Showing that BO can drive the process of learning control rules directly from operational data. BO can also be extended to multi-objective optimization. Two algorithms have been proposed for multi-objective detection problem using two different acquisition functions.

This paper focuses on the control of microgrids where both gas and electricity are provided to the final customer, i.e., multi-carrier microgrids. Hence, these microgrids include thermal and electrical loads, renewable energy sources, energy storage systems, heat pumps, and combined heat and power units. The parameters characterizing the multi-carrier microgrid are subject to several disturbances, such as fluctuations in the provision of renewable energy, variability in the electrical and thermal demand, and uncertainties in the electricity and gas pricing. With the aim of accounting for the data uncertainties in the microgrid, we propose a Robust Model Predictive Control (RMPC) approach whose goal is to minimize the total economical cost, while satisfying comfort and energy requests of the final users. In the related literature various RMPC approaches have been proposed, focusing either on electrical or on thermal microgrids. Only a few contributions have addressed the robust control of multi-carrier microgrids. Consequently, we propose an innovative RMPC algorithm that employs on an uncertainty set-based method and that can provide better performance compared with deterministic model predictive controllers applied to multi-carrier microgrids. With the aim of mitigating the conservativeness of the approach, we define suitable robustness factors and we investigate the effects of such factors on the robustness of the solution against variations of the uncertain parameters. We show the effectiveness of the proposed RMPC approach by applying it to a realistic residential multi-carrier microgrid and comparing the obtained results with the ones of a baseline robust method. Postprint accepted for publication in IEEE Transactions on Automation Science and Engineering (T-ASE). How to cite: R. Carli, G. Cavone, T. Pippia, B. De Schutter, and M. Dotoli, ”Robust Optimal Control for Demand Side Management of Multi-Carrier Microgrids,” IEEE Transactions on Automation Science and Engineering, vol. 19, no. 3, pp. 1338-1351, July 2022. DOI: https://doi.org/10.1109/TASE.2022.3148856

Despite the enormous success of artificial neural networks (ANNs) in many disciplines, the characterization of their computations and the origin of key properties such as generalization and robustness remain open questions. Recent literature suggests that robust networks with good generalization properties tend to be biased toward processing low frequencies in images. To explore the frequency bias hypothesis further, we develop an algorithm that allows us to learn modulatory masks highlighting the essential input frequencies needed for preserving a trained network's performance. We achieve this by imposing invariance in the loss with respect to such modulations in the input frequencies. We first use our method to test the low-frequency preference hypothesis of adversarially trained or data-augmented networks. Our results suggest that adversarially robust networks indeed exhibit a low-frequency bias but we find this bias is also dependent on directions in frequency space. However, this is not necessarily true for other types of data augmentation. Our results also indicate that the essential frequencies in question are effectively the ones used to achieve generalization in the first place. Surprisingly, images seen through these modulatory masks are not recognizable and resemble texture-like patterns.

Social distancing has been enacted in order to mitigate the spread of COVID-19. Like many authors, we adopt the classic epidemic SIR model, where the infection rate is the control variable. Its differential flatness property yields ele mentary closed-form formulae for open-loop social distancing scenarios, where, for instance, the increase of the number of uninfected people may be taken into account. Those formulae might therefore be useful to decision makers. A feedback loop stemming from model-free control leads to a remarkable robustness with respect to severe uncertainties and mismatches. Although an identification procedure is presented, a good knowledge of the recovery rate is not necessary for our control strategy., Comment: 10th Vienna International Conference on Mathematical Modelling (MATHMOD 2022) -- July 27-29, 2022 -- Vienna, Austria

The issue of vulnerability and robustness in networks have been addressed by several methods. The goal is to identify which are the critical components (i.e., nodes/edges) whose failure impairs the functioning of the network and how much this impacts the ensuing increase in vulnerability. In this paper we consider the drop in the network robustness as measured by the increase in vulnerability of the perturbed network and compare it with the original one. Traditional robustness metrics are based on centrality measures, the loss of efficiency and spectral analysis. The approach proposed in this paper sees the graph as a set of probability distributions and computes, specifically the probability distribution of its node to node distances and computes an index of vulnerability through the distance between the node-to-node distributions associated to original network and the one obtained by the removal of nodes and edges. Two such distances are proposed for this analysis: Jensen–Shannon and Wasserstein, based respectively on information theory and optimal transport theory, which are shown to offer a different characterization of vulnerability. Extensive computational results, including two real-world water distribution networks, are reported comparing the new approach to the traditional metrics. This modelling and algorithmic framework can also support the analysis of other networked infrastructures among which power grids, gas distribution and transit networks.

In this work, we propose a novel robustness measure for networks, which we refer to as Effective Resistance Centrality of a vertex (or an edge), defined as the relative drop of the Kirchhoff index due to deletion of this vertex (edge) from the network. Indeed, we provide a local robustness measure, able to catch which is the effect of either a specific vertex or a specific edge on the network robustness. The validness of this new measure is illustrated on some typical graphs and on a wide variety of well-known model networks. Furthermore, we analyse the topology of the US domestic flight connections. In particular, we investigate the role that airports play in maintaining the structure of the entire network.

International audience; Reversible computation is a computing paradigm where execution can progress backwards as well as in the usual, forward direction. It has found applications in many areas of computer science, such as circuit design, programming languages, simulation, modelling of chemical reactions, debugging and robotics. In this article, we give an overview of reversible computation focusing on its use in robotics. We present an example of programming industrial robots for assembly operations where we combine classical AI planning with reversibility and embodied AI to increase robustness and versatility of industrial robots.

In this article, we study the dynamic resilient containment control problem for continuous-time multirobot systems (MRSs), i.e., the problem of designing a local interaction protocol that drives a set of robots, namely the followers, toward a region delimited by the positions of another set of robots, namely the leaders, under the presence of adversarial robots in the network. In our setting, all robots are anonymous, i.e., they do not recognize the identity or class of other robots. We consider as adversarial all those robots that intentionally or accidentally try to disrupt the objective of the MRS, e.g., robots that are being hijacked by a cyber–physical attack or have experienced a fault. Under specific topological conditions defined by the notion of (r,s)-robustness, our control strategy is proven to be successful in driving the followers toward the target region, namely a hypercube, in finite time. It is also proven that the followers cannot escape the moving containment area despite the persistent influence of anonymous adversarial robots. Numerical results with a team of 44robots are provided to corroborate the theoretical findings.

In this paper we describe a scan-matching based registration algorithm for tracking moving objects which falls in the emerging area that predicates the integration between robotics and big data applications. The scan matching approaches track paths of a mobile object by comparing maps of the environment seen by the object during its movement. Algorithms described in this paper are hybrid, i.e. they compare maps by using first a genetic pre-alignment based on a novel metrics, and then performing a finer alignment using a deterministic approach. This kind of hybridization is, indeed, not new. However, the novel metrics used in this paper leads to important new properties, namely to correct arbitrary rotational errors and to cover larger search spaces. The proposed algorithm is experimentally compared to other approaches, and better performance in terms of accuracy and robustness are reported. Finally, our algorithm is also very fast thanks to the genetic pre-alignment task and the novel metrics we propose.

In the last decade, microbial-based biotechnological processes are paving the way toward sustainability as they implemented the use of renewable feedstocks. Nonetheless, the viability and competitiveness of these processes are often limited due to harsh conditions such as: the presence of feedstock-derived inhibitors including weak acids, non-uniform nature of the substrates, osmotic pressure, high temperature, extreme pH. These factors are detrimental for microbial cell factories as a whole, but more specifically the impact on the cell’s membrane is often overlooked. The plasma membrane is a complex system involved in major biological processes, including establishing and maintaining transmembrane gradients, controlling uptake and secretion, intercellular and intracellular communication, cell to cell recognition and cell’s physical protection. Therefore, when designing strategies for the development of versatile, robust and efficient cell factories ready to tackle the harshness of industrial processes while delivering high values of yield, titer and productivity, the plasma membrane has to be considered. Plasma membrane composition comprises diverse macromolecules and it is not constant, as cells adapt it according to the surrounding environment. Remarkably, membrane-specific traits are emerging properties of the system and therefore it is not trivial to predict which membrane composition is advantageous under certain conditions. This review includes an overview of membrane engineering strategies applied toSaccharomyces cerevisiaeto enhance its fitness under industrially relevant conditions as well as strategies to increase microbial production of the metabolites of interest.

International audience; Approximate computing (AxC) is increasingly emerging as a new design paradigm to produce more efficient computation systems by judiciously reducing the computation quality. In particular, AxC has been successfully applied to integrated circuits (ICs), in the last years. Hence, concerning the test of such new class of ICs, namely approximate ICs (AxICs), new challenges--as well as new opportunities--have emerged. In this survey, we provide a thorough analysis of issues related to test procedures for AxICs and review the state-of-the-art techniques to deal with them. We resort to an illustrative example having the twofold aim of: 1) guiding the reader through the AxIC testing challenges and 2) illustrating the existing solutions to correctly overcome them, while suitably taking advantage of opportunities coming from approximation. We analyze experimentally the most recent testing techniques for AxICs and highlight their mature aspects, as well as their shortcomings. Experimental outcomes show that the testing process for AxIC is not completely mature. Indeed, only under specific conditions existing testing procedures achieve good results.

In the study of muscle synergies during the maintenance of single-leg stance (SLS) there are methodological issues that must be taken into account before performing the synergy extraction. In particular, it is important to distinguish between epochs of surface electromyographic (sEMG) signals corresponding to a 'good' balance control during the SLS test, from those characterized by an 'excessive' body sway. The aim of this work is to assess the robustness in the segmentation and selection of sEMG signal epochs to be chosen as input for the synergy extraction algorithm. The robustness is evaluated in terms of: 1) consistency of the number of muscle synergies, and 2) weight vector correlation. Our results show that the same number of muscle synergies and similar weight vectors are obtained, independently from the threshold chosen to build the segmentation mask. The methodology proposed may help the interpretation of muscle synergies in SLS test.

In this work, we consider the problem of scheduling a set of trains (i.e., determining their departure and arrival times at the visited stations) and simultaneously deciding their stopping patterns (i.e., determining at which stations the trains should stop) with constraints on passenger demand, given as the number of passengers that travel between an origin station and a destination station. In particular, we face the setting in which demand can be uncertain, and propose Mixed Integer Linear Programming (MILP) models to derive robust solutions in planning, i.e., several months before operations. These models are based on the technique of Light Robustness, in which uncertainty is handled by inserting a desired protection level, and solution efficiency is guaranteed by limiting the worsening of the nominal objective value (i.e., the objective value of the problem in which uncertainty is neglected). In our case, the protection is against a potential increased passenger demand, and the solution efficiency is obtained by limiting the train travel time and the number of train stops. The goal is to determine robust solutions in planning so as to reduce the passenger inconvenience that may occur in real-time due to additional passenger demand. The proposed models differ in the way of inserting the protection, and show different levels of detail on the required information about passenger demand. They are tested on real-life data of the Wuhan–Guangzhou high-speed railway line under different demand scenarios, and the obtained results are compared with those found by solving the nominal problem. The comparison shows that robust solutions can handle uncertain passenger demand in a considerably more effective way.

Most of the existing buildings in seismic prone regions have been built before the publication of modern design provisions against earthquakes, resulting in the need for structural retrofitting. Furthermore, some of those buildings are also subjected to additional hazards that may be either triggered by earthquakes (e.g., landslides, soil liquefaction, tsunamis) or associated with other natural or anthropogenic events, such as floods, vehicle collision, blast, and fire. A multi-hazard performance assessment of building structures is thus of paramount importance to implement integrated retrofit strategies, which otherwise would not be economically sustainable if oriented to structural risk mitigation against a single hazard. While retrofit strategies to improve the seismic performance of reinforced concrete (RC) structures have been widely investigated, structural retrofitting against progressive collapse has received very little attention. Within this context, the present paper illustrates a numerical investigation on the influence of seismic retrofitting on structural robustness of a four-storey, five-bay, RC frame building designed only to gravity loads. Seismic performance and structural robustness were respectively evaluated in OpenSees through pushover and pushdown analyses of a fibre-based finite element model. Structural robustness was evaluated under two relevant column-removal scenarios, i.e., the sudden loss of a central and a corner column, whereas earthquake resistance was assessed according to the N2 method, evidencing the need for seismic retrofitting. A retrofit measure based on carbon fibre reinforced polymers was then considered to avoid premature brittle failures. Analysis results show that this retrofit strategy was able to increase both seismic safety and structural robustness. Subsequently, a parametric analysis was carried out in order to evaluate the impact of beam span length and shear strength of the retrofitting system.

Gully erosion has been identified as an important soil degradation process and sediment source, especially in arid and semiarid areas. Thus, it is useful to identify the spatial occurrence of this form of water erosion in the landscape and the most vulnerable areas. In this study, we explored the effects of different pixel sizes on some controlling factors extracted from a digital elevation model and remote sensing data when producing a gully erosion susceptibility map (GESM) of Ekbatan Dam Basin, Hamadan, Iran. An inventory map of the gully landforms was prepared based on global positioning system routes of the gullies, extensive field surveys, and visual interpretations of satellite images obtained from Google Earth. Five data sets with pixel sizes ranging from 2 to 30 m were obtained using topographic attributes and remote sensing data comprising the elevation, slope degree, slope aspect, catchment area, plan curvature, profile curvature, stream power index, topographic position index, topographic wetness index, land use, and normalized difference vegetation index, which can affect the distribution of gully erosion. For each data set, 70% and 30% of the data were selected randomly for calibrating and validating the models, respectively. The statistical relationships between the occurrence of gully erosion and controlling factors were calculated using four machine-learning models, i.e., generalized linear model, boosted regression tree (BRT), multivariate adaptive regression spline, and artificial neural network (ANN). Statistical tests comprising the kappa coefficient and the area under the receiver operating characteristic curve (AUC) were calculated for both the calibration and validation data sets to estimate the optimal pixel size. The results showed that among the data sets with different pixel sizes, the optimal pixel size was 10 m for each model. In addition, the capacity of the four techniques for modeling gully erosion occurrence was quite stable when the calibration and validation samples were changed in the data set. Finally, based on three changes of the calibration and validation data sets with a pixel size of 10 m, the BRT and ANN models obtained outstanding performance (AUC > 0.9), where they had the highest goodness-of-fit and predictive power, and thus the greatest robustness to changes in the calibration/validation data (i.e., lowest sensitivity to altering calibration/validation data). Our results demonstrate the importance of selecting a suitable pixel size when producing a GESM for soil and water management practices.

© 2018 Elsevier Ltd The aim of this paper is to investigate the impact of electro-thermal device parameter spread on the avalanche ruggedness of parallel silicon carbide (SiC) power MOSFETs representative of multi-chip layout within an integrated power module. The tests were conducted on second generation 1200 V, 36 A–80 mΩ rated devices. Different temperature-dependent electrical parameters were identified and measured for a number of devices. The influence of spread in measured parameters was investigated experimentally during avalanche breakdown transient switching events and important findings have been highlighted.

In this paper we provide an axiomatic foundation to Orlicz risk measures in terms of properties of their acceptance sets, by exploiting their natural correspondence with shortfall risk Follmer and Schied (Stochastic finance. De Gruyter, Berlin, 2011), thus paralleling the characterization in Weber (Math Financ 16:419–442, 2006). From a financial point of view, Orlicz risk measures assess the stochastic nature of returns, in contrast to the common use of risk measures to assess the stochastic nature of a position’s monetary value. The correspondence with shortfall risk leads to several robustified versions of Orlicz risk measures, and of their optimized translation invariant extensions (Rockafellar and Uryasev in J Risk 2:21–42, 2000, Goovaerts et al. in Insur Math Econ 34:505–516, 2004), arising from an ambiguity averse approach as in Gilboa and Schmeidler (J Math Econ 18:141–153, 1989), Maccheroni et al. (Econometrica 74:1447–1498, 2006), Chateauneuf and Faro (J Math Econ 45:535–558, 2010), or from a multiplicity of Young functions. We study the properties of these robust Orlicz risk measures, derive their dual representations, and provide some examples and applications.

Le misure di rischio e l’asset allocation sono questioni di fondamentale importanza per i mercati finanziari. Lo studio è suddiviso in due parti. Nella prima parte vengono dimostrate proprietà molto importanti per il Lambda Value at Risk, mentre nella seconda parte vengono utilizzati modelli Markov Switching per modellizzare i rendimenti di serie finanziarie e viene introdotta una regola di trade basata sui regimi. L’ultima crisi finanziaria ha evidenziato le debolezze del Value at Risk. Per questo l’interesse verso misure di rischio alternative è aumentato notevolmente negli ultimi anni. In questo studio viene dimostrato che il Lambda Value at Risk è robuto ed elicitabile in particolari famiglie di distribuzioni ed è anche consistente. Il comportamento dei mercati finanziari è generalmente sottoposto a radicali cambiamenti quando si verificano guerre, crisi politiche o economiche ed altri eventi. Tali cambiamenti non sono permanenti ma possono persistere per periodi più o meno lunghi. Questo si rispecchia in alcune caratteristiche proprie delle serie storiche finanziare come ad esempio l’eccesso di kurtosis, gli effetti di skewness e la volatilità non costante. I modelli Markov Switching riescono a catturare questi cambiamenti che si verificano in maniera non deterministica e possono persistere per alcuni periodi dopo il cambiamento. Più precisamente, utilizziamo misture Markov Switching di distributioni normali fissando il numero di stati a N=2 corrispondenti a Normale Volatilità e Alta Volatilità. Gli obiettivi dello studio sono fondamentalmente due. Prima di tutto i parametri del modello vengono stimati attraverso Gibbs Sampling. Successivamente viene presentata una regola di trade basata sui regimi e confrontata con una semplice strategia buy-and-hold. Il database consiste in rendimenti giornalieri da Gennaio 1997 a Giugno 2018. Vengono analizzate diverse asset class appartenenti a diverse aeree geografiche. Il modello Markov-Switching è stato stimato tanto a livello univariato per le singole serie storiche quanto a livello multivariato trattando i processi correlativi tra le asset class. Per quanto riguarda l’univariato la maggior parte degli indici mostrano due stati nettamente separati con lo stato di Volatilità Normale come stato predominante. In generale le volatilità in Alta volatilità sono doppie rispetto a quella in Volatilità Normale. Il caso multivariato invece mostra che lo stato di Alta Volatilità è caratterizzato da un aumento delle correlazioni. L’esistenza di due regimi con caratteristiche diverse tra loro evidenzia la necessità di utilizzare strategie di trading differenti. Nell’ultima parte dello studio viene analizzata una strategia di trading basata sui regimi. Risk measures and Asset allocation are a matter of primary concern for the financial market. My study is divided into two parts. In the first part important properties of the Lambda Value at Risk are showed. In the second part Markov Switching models are used to handle the stocks returns and regime-based trade rule is introduced. The last global financial crisis has highlighted the lacks of the Value at Risk. Thus, the interest on alternative risk measures has considerably increased in the last years. In this study we showed that Lambda Value at Risk is robust and elicitable within particular classes of distributions. In addition, it also satisfies the consistency property without any condition on the mechanism generating data. The behavior of financial markets may be changed radically when wars, economical or political crises and other events occur. This changes are generally not permanent but persist for longer or shorter periods of time. This is reflected in certain specific features of financial time series such as the leptokurtosis, the skewness and the heteroskedasticity. Markov Switching models can handle these behavioral changes that occur randomly and persist for several periods after the change. Specifically, we model the returns by a Markov Switching mixture of gaussian distributions and we fix the number of regimes to N=2 corresponding to Normal Volatility and High Volatility. The purpose of the study is twofold. First of all, the model is estimated using Markov Chain Monte Carlo methods. Specifically Gibbs Sampling algorithm is used. Secondly, regime-based trade rule is presented and compared with a buy-and-hold strategy. The data consists of daily returns from Jan 1997 to June 2018. We analyzed different Asset classes across different geographic areas. We estimate both univariate and multivariate Markov Switching models to take into account the correlations among asset classes. In the univariate case, most indices exhibit two states clearly separated and Normal Volatility state is the predominant State. In general, the volatilities in High Volatility are twice those in Normal Volatility. The multivariate case showed that High Volatility state is characterize by an increase of correlations. Thus, the diversification could be only apparent. The existence of two regimes with different features leads to the necessity of different strategies. In the last part of the study a trade rule regime-based is analyzed.

We study the properties of an M-estimator arising from the minimization of an integrated version of the quantile loss function. The estimator depends on a tuning parameter which controls the degree of robustness. We show that the sample median and the sample mean are obtained as limit cases. Consistency and asymptotic normality are established and a link with the Hodges–Lehmann estimator and the Wilcoxon test is discussed. Asymptotic results indicate that high levels of efficiency can be reached by specific choices of the tuning parameter. A Monte Carlo analysis investigates the finite sample properties of the estimator. Results indicate that efficiency can be preserved in finite samples by setting the tuning parameter to a low fraction of a (robust) estimate of the scale.

The key contribution of this paper is to embed the analysis of the network in a framework based on a mapping from the input space whose elements are nodes of a graph or the entire graph into an information space whose elements are probability distributions associated to objects in the input space. Specifically, a node is associated to the probability distribution of its node-to-node distances and the whole graph to the aggregation of these node distributions. In this space two distances are proposed for this analysis: Jensen-Shannon and Wasserstein, based respectively on information theory and optimal transport theory. This representation allows to compute the distance between the original network and the one obtained by the removal of nodes or edges and use this distance as an index of the increase in vulnerability induced by the removal. In this way a new characterization of vulnerability is obtained. This new index has been tested in two real-world water distribution networks. The results obtained are discussed along those which relate vulnerability to the loss of efficiency and those given by the analysis of the spectra of the adjacency and Laplacian matrices of the network. The models and algorithms considered in this paper have been integrated into an analytics framework which can also support the analysis of other networked infrastructures among which power grids, gas distribution, and transit networks are included.

Seismic design criteria allow enhancing the structural ductility and controlling the damage distribution. Therefore, detailing rules and design requirements given by current seismic codes might be also beneficial to improve the structural robustness. In this paper a comprehensive parametric study devoted to quantifying the effectiveness of seismic detailing for steel Moment Resisting Frames (MRF) in limiting the progressive collapse under column loss scenarios is presented and discussed. The overall structural performance was analysed through nonlinear static and dynamic analyses. With this regard the following cases were examined: (i) MRF structures designed for wind actions according to Eurocode 1; (ii) MRF structures designed for seismic actions according to Eurocode 8. The investigated parameters were (i) the number of storeys; (ii) the interstorey height; (iii) the span length; (iv) the building plan layout; and (v) the column loss scenario. Results show that structures designed according to capacity design principles are less robust than wind designed ones, provided that the connections have the same capacity threshold in both cases. In addition, the numerical outcomes show that both the number of elements above the removed column and stiffness of beams are the key parameters in arresting progressive collapse.

The present paper deals with a parametric class of models implemented for ordered categorical data, denoted as cub model, which is defined as a discrete mixture of a shifted binomial and a uniform random variable. For these models, robustness issues are considered. In particular, the influence function is introduced and subsequently used to define the robustness measures for categorical data. By exploiting the peculiar parametrization of the cub models, diagnostic plots are proposed which allow to display the effect of a contamination in the data, simultaneously for all categories. The breakdown point is also considered and a computational procedure is suggested to determine an upper bound. The paper provides evidence that, despite the limited range of the support, contaminations in the data can heavily affect the inferential procedures and hence robustness topics are indeed relevant for ordinal data.

In the present contribution, we characterise law determined convex risk measures that have convex level sets at the level of distributions. By relaxing the assumptions in Weber (Math. Finance 16:419–441, 2006), we show that these risk measures can be identified with a class of generalised shortfall risk measures. As a direct consequence, we are able to extend the results in Ziegel (Math. Finance, 2014, http://onlinelibrary.wiley.com/doi/10.1111/mafi.12080/abstract) and Bellini and Bignozzi (Quant. Finance 15:725–733, 2014) on convex elicitable risk measures and confirm that expectiles are the only elicitable coherent risk measures. Further, we provide a simple characterisation of robustness for convex risk measures in terms of a weak notion of mixture continuity.

Graph robustness—the ability of a graph to preserve its connectivity after the loss of nodes and edges—has been extensively studied to quantify how social, biological, physical, and technical systems withstand to external damages. In this paper, we prove that graph robustness can be quickly estimated through the Randic index, a parameter introduced in chemistry to study organic compounds. We prove that Erdos-Renyj (ER) graphs are a good specimen of robust graphs because they lack of a clear modular structure; we derive an analytical expression for the Randic index of ER graphs and use ER graphs as an effective term of comparison to decide about graph robustness. Experiments on real datasets from different domains (scientific collaboration networks, content-sharing systems, co-purchase networks from an e-commerce platform, and a road network) show that real-life large graphs are more robust than ER ones with the same number of nodes and edges. We also observe that if node degree distribution closely follows a power law, then few edges contribute for more than half of the Randic index, thus indicating that the selective removal of those edges has devastating impact on graph robustness. Finally, we describe sampling-based algorithms to efficiently but accurately approximate the Randic index.

In this chapter, we will discuss the principles and criteria underpinning robust and agile policy designs. We begin with a conceptual delimitation of robustness and agility, focusing in particular on their relevance for policy design. We then provide a brief overview of policy change and stability, before focusing on agility and robustness as criterions for policy design and on the conditions that allow their fruitful adoption.

One of the key issues in the acquisition of sparse data by means of compressed sensing (CS) is the design of the measurement matrix. Gaussian matrices have been proven to be information-theoretically optimal in terms of minimizing the required number of measurements for sparse recovery. In this paper we provide a new approach for the analysis of the restricted isometry constant (RIC) of finite dimensional Gaussian measurement matrices. The proposed method relies on the exact distributions of the extreme eigenvalues for Wishart matrices. First, we derive the probability that the restricted isometry property is satisfied for a given sufficient recovery condition on the RIC, and propose a probabilistic framework to study both the symmetric and asymmetric RICs. Then, we analyze the recovery of compressible signals in noise through the statistical characterization of stability and robustness. The presented framework determines limits on various sparse recovery algorithms for finite size problems. In particular, it provides a tight lower bound on the maximum sparsity order of the acquired data allowing signal recovery with a given target probability. Also, we derive simple approximations for the RICs based on the Tracy-Widom distribution., 11 pages, 6 figures, accepted for publication in IEEE transactions on information theory