1,721,021 research outputs found
Bounds on the maximal number of corrupted nodes via Boolean Network Tomography
Full text linkIn this thesis we are concentrating on identifying defective items in larger sets which is a main problem with many applications in real life situations, e.g., fault diagnosis, medical screening and DNA screening. We consider the problem of localizing defective nodes in networks through an approach based on Boolean Network Tomography (BNT), which is grounded on inferring informations from the Boolean outcomes of end-to-end measurement paths. In particular, we focus on the following three:
• Studying Maximal Identifiability, which was recently introduced in BNT to measure the maximal number of corrupted nodes which can be uniquely localized in sets of end-to-end measurement paths on networks;
• Central role of Vertex-Connectivity in maximal identifiability;
• Investigating identifiability conditions on the set of paths which guarantee discovering or counting unambiguously the defective nodes and contributing this problem both from a theoretical and applied perspective.
We prove tight upper and lower bounds on the maximal identifiability for sets of end-to-end paths in network topologies obtained from trees and d-(dimensional) grids over n^d nodes. For trees (both directed and undirected) we show that the maximal identifiability is 1. For undirected d-grids we prove that, using only 2d monitors, maximal identifiability is at least d − 1 and at most d. In the directed case proving that the maximal identifiability is d and can be reached at the cost of placing 2d(n − 1) + 2 monitors on the d-grid. This monitor placement is optimal and adding more monitors will not increase the identifiability. We also study maximal identifiability for directed topologies under embeddings establishing new relations with embeddability, graph dimension and proving that under the operation of transitive closure maximal identifiability grows linearly.
Our results suggest the design of networks over n nodes reaching maximal identifiability Ω(log n) using O(log n) monitors and an heuristic to boost maximal identifiability increasing the minimal degree of the network which we test experimentally.
Moreover we prove tight bounds on the maximal identifiability first in a particular class of graphs, the Line of Sight networks and then slightly weaker bounds for arbitrary networks. Furthermore we initiate the study of maximal identifiability in random networks. We investigate two models: the classical Erdős-Rényi model, and that of Random Regular graphs. The proposed framework allows a probabilistic analysis of the identifiability in random networks giving a tradeoff between the number of monitors to place and the maximal identifiability.
Further in this thesis, we work on the precise tradeoff between number of nodes and number of paths such that at most k nodes can be identified unambiguously. The answer to this problem is known only for k = 1 and we answer it for any k, setting a problem implicitly left open in previous works. We focus on upper and lower bounds on the number of unambiguously identifiable nodes, introducing new identifiability measures (Separability and Distinguishability) which strictly imply and are strictly implied by the notion of identifiability introduced in [39]. We utilize these new measures to design algorithmic heuristics to count failure nodes in a fine-grained way and further to prove the first complexity hardness results on the problem of identifying failure nodes in networks via BNT.
At last but not least, we introduce a random model so as to achieve lower bounds on the number of unambiguously identifiable defective nodes. We use this model to approximate that number on real networks by a maximum likelihood estimate approach
Stochastic bounds on execution times of parallel computations
No full textWe obtain stochastic bounds on execution times of parallel computations assuming ideal conditions for shared resources. A parallel computation is modelled as a task system with precedence constraints expressed as a directed acyclic graph (DAG). The task execution times are assumed independent random variables. The performance measure considered is the overall execution time of the computation. To obtain upper bounds on this measure, we apply stochastic ordering and stochastic comparison technique
Estimation of distribution parameters as a tool for model-based system engineering and model identification
Full text linkThe estimation of the parameters of a probability distribution (e.g., moments) plays an important role both in the model-based system engineering (e.g., analysis and verification through Statistical Model Checking (SMC)) and in the identification of parameters of predictive models (e.g., systems biology, social networks). The contribution of this PhD thesis is both on the algorithm side and on the modeling side. On the algorithm side, we overview a set of Monte Carlo-based Statistical Model Checking tools and algorithms for the verification of Cyber-Physical Systems, and we provide selection criteria for the verification problem at hand. Furthermore, we present an efficient Monte Carlo-based algorithm to estimate the expected value of a multivariate random variable, when marginal density functions are not known. We prove the correctness of our algorithm, we give an Upper Bound and a Lower Bound to its complexity and we present experimental results confirming our evaluations. On the modeling side, we present a mechanistic and identifiable model to predict, at the node level and at a set of nodes level, the expected value of the retweeting rate of a message inside a social network, at a certain time. Our model parameters are random variables, whose distribution parameters are estimated from an available dataset. We experimentally show that our model reliably predicts both the qualitative and the quantitative time behavior of retweeting rates. This is confirmed by the high correlation between the predicted and the observed data. These results enable a simulation-based analysis of users or of a set of users' behaviors inside a network
Artificial intelligence and model checking methods for in silico clinical trials
Full text linkModel-based approaches to safety and efficacy assessment of pharmacological treatments (In Silico Clinical Trials, ISCT) hold the promise to decrease time and cost for the needed experimentations, reduce the need for animal and human testing, and enable personalised medicine, where treatments tailored for each single patient can be designed before being actually administered.
Research in Virtual Physiological Human (VPH) is harvesting such promise by developing quantitative mechanistic models of patient physiology and drugs. Depending on many parameters, such models define physiological differences among different individuals and different reactions to drug administrations. Value assignments to model parameters can be regarded as Virtual Patients (VPs). Thus, as in vivo clinical trials test relevant drugs against suitable candidate patients, ISCT simulate effect of relevant drugs against VPs covering possible behaviours that might occur in vivo.
Having a population of VPs representative of the whole spectrum of human patient behaviours is a key enabler of ISCT.
However, VPH models of practical relevance are typically too complex to be solved analytically or to be formally analysed. Thus, they are usually solved numerically within simulators. In this setting, Artificial Intelligence and Model Checking methods are typically devised. Indeed, a VP coupled together with a pharmacological treatment represents a closed-loop model where the VP plays the role of a physical subsystem and the treatment strategy plays the role of the control software. Systems with this structure are known as Cyber-Physical Systems (CPSs). Thus, simulation-based methodologies for CPSs can be employed within personalised medicine in order to compute representative VP populations and to conduct ISCT.
In this thesis, we advance the state of the art of simulation-based Artificial Intelligence and Model Checking methods for ISCT in the following directions.
First, we present a Statistical Model Checking (SMC) methodology based on hypothesis testing that, given a VPH model as input, computes a population of VPs which is representative (i.e., large enough to represent all relevant phenotypes, with a given degree of statistical confidence) and stratified (i.e., organised as a multi-layer hierarchy of homogeneous sub-groups). Stratification allows ISCT to adaptively focus on specific phenotypes, also supporting prioritisation of patient sub-groups in follow-up in vivo clinical trials.
Second, resting on a representative VP population, we design an ISCT aiming at optimising a complex treatment for a patient digital twin, that is the virtual counterpart of that patient physiology defined by means of a set of VPs. Our ISCT employs an intelligent search driving a VPH model simulator to seek the lightest but still effective treatment for the input patient digital twin.
Third, to enable interoperability among VPH models defined with different modelling and simulation environments and to increase efficiency of our ISCT, we also design an optimised simulator driver to speed-up backtracking-based search algorithms driving simulators.
Finally, we evaluate the effectiveness of our presented methodologies on state-of-the-art use cases and validate our results on retrospective clinical data
Adult Bicuspid Aortic Valve
No full textBicuspid aortic valve (BAV) is the most common congenital defect of the heart, occurring in 0.5-2% of live births. It is estimated to be responsible for a relevant burden of valvular and vascular disease in the adulthood (bicuspid valvulo-aortopathy). The present chapter focuses on the aspects of valvular morbidity (aortic valve stenosis, regurgitation, endocarditis) and complications of the thoracic aorta in the adult (aortic dilatation, aortic dissection), trying to underscore similarities and unique features of BAV-related conditions compared to the respective diseases in tricuspid aortic valve patients. Epidemiological aspects, pathogenetic theories, risk stratification strategies and treatment principles will be briefly reviewed
Extracorporeal life support and unconventional mode of respiratory management
No full textAcute respiratory failure is very common in interstitial lung diseases and optimal ventilatory management in this setting is still under investigation. The dismal prognosis of patients needing mechanical ventilation urged the search for alternative strategies, such as extracorporeal life support. Since its first use as a bridge to lung transplant, extracorporeal membrane oxygenation (ECMO) improved its technology and results. Despite available data on the topic mostly derive from small non-homogeneous case series or reports, usually with a single center setting and a retrospective design, the literature shows that the use of ECMO bridging transplants has risen. Clinical success is more related to appropriate candidacy and timely deployment rather than to device-specific complications. Anyway, further efforts need to be done to improve outcomes and to increase the bridge to transplant/bridge to no-where ratio
Use of Noninvasive Ventilation in Postoperative Patients in Cardiac Surgery
No full textWorldwide, approximately 1.5 million open-heart surgeries are performed yearly, under general anesthesia, through a median sternotomy, and with the support of cardiopulmonary bypass. An adequate tissue oxygenation in the postoperative is essential for organ function and for wound healing, allowing for uneventful recovery. The aim of this chapter is to assess whether it is achievable by postoperative noninvasive ventilation
Toward Optimal Cross-layer Solutions for Cognitive Radio Wireless Networks
Full text linkCognitive radio (CR) networks have been proposed as a viable solution to spectrum scarcity problems. In CR networks, CR nodes exploit spectrum holes in space, time and/or frequency to transmit on licensed frequency bands without affecting primary users. In such a dynamic and unpredictable environment, CR networks require the ability to gather information on the surrounding available spectrum and to exploit this information to maximize CR nodes performance. In a companion paper we deal with sensing architecture and protocols. In this paper, instead, we derive a cross-layer scheme for cognitive radio networks which jointly optimize the sources flow rates, routing and medium access control while accounting for and exploiting the available spectrum resources. The proposed scheme builds on important recent results on close to optimal fully distributed CSMA-based scheduling algorithms, which allows us to derive a fully distributed solution
A comparative study on shear modulus and damping ratio of cohesive soil from laboratory tests
No full textOn Guaranteeing Global Dependability Properties in Collaborative Business Process Management
No full text- …
