Serial and parallel algorithms for short time horizon multi-attribute queries on stochastic multi-agent systems

We propose to perform statistical probabilistic model checking by using perfect simulation in order to verify steady-state and time unbounded until formulas over Markov chains. The model checking of probabilistic models by statistical methods has received increased attention in the last years since it provides an interesting alternative to numerical model checking which is poorly scalable with the increasing model size. In previous statistical model checking works, unbounded until formulas could not be efficiently verified, and steady-state formulas had not been considered due to the burn-in time problem to detect the steady-state. Perfect simulation is an extension of Markov Chain Monte Carlo (MCMC) methods that allows us to obtain exact steady-state samples of the underlying Markov chain, and thus it avoids the burn-in time problem to detect the steady-state. Therefore we suggest to verify time unbounded until and steady-state dependability properties for large Markov chains through statistical model checking by combining perfect simulation and statistical hypothesis testing.

Controlling stochastic multi-agent systems (MAS) is a promising branch of systems engineering owing to its ability to solve complex practical problems by taking inspiration from the social behavior of living organisms. Consensus on more than one state is a less considered problem in the framework of leader-following control of high-order uncertain nonlinear stochastic multi-agent systems with unknown dynamics. To this end, in this paper, a sliding surface integrates a tracking error polynomial with consensus error between states of the connected agents. Accordingly, a new distributed fuzzy sliding mode controller is designed for a class of stochastic multi-agent systems with partially unknown deterministic dynamics and unknown bounded stochastic dynamics. The proposed control protocol steers agents to reach agreement on states of the leader and concurrently approximates the unknown part of their deterministic dynamics by using fuzzy systems with adaptive parameters. In the sense of Lyapunov, it is proved that consensus error and all of the closed-loop signals of each agent are semi-globally uniformly ultimately bounded (SGUUB) in mean square. Under the designed controller, nonlinear closedloop experimental dynamics behave well in both tracking and reaching state consensus.

This paper considers the optimal distributed control problem for a linear stochastic multi-agent system (MAS). Due to the distributed nature of MAS network, the information available to an individual agent is limited to its vicinity. From the entire MAS aspect, this imposes the structural constraint on the control law, making the optimal control law computationally intractable. This paper attempts to relax such a structural constraint by expanding the neighboring information for each agent to the entire MAS, enabled by the distributed estimation algorithm embedded in each agent. By exploiting the estimated information, each agent is not limited to interact with its neighborhood but further establishing the `virtual interactions' with the non-neighboring agents. Then the optimal distributed MAS control problem is cast as a synthesized control-estimation problem. An iterative optimization procedure is developed to find the control-estimation law, minimizing the global objective cost of MAS.

IntroductionThe purpose of this study was to examine whether the time horizon of time trade-off (TTO) and standard gamble (SG) utility assessment influences utility scores and discrimination between health states.MethodsIn two phases, UK general population participants rated three osteoarthritis health states in TTO and SG procedures with two time horizons: (1) 10-year and (2) a time horizon derived from self-reported additional life expectancy (ALE). The two time horizons were compared in terms of mean utilities and discrimination among health states.ResultsIn Phase 1, the 10-year tasks were completed by 80 participants, 35 of whom also completed utility assessment with the ALE. In Phase 2, all 101 participants completed procedures with both time horizons. Utility scores tended to be lower with the ALE than the 10-year, a difference that was statistically significant for two health states with SG in Phase 1 (P < 0.05), two health states with TTO in Phase 2 (P < 0.01), and one health state with SG in Phase 2 (P < 0.001). In Phase 1, rates of discrimination between mild and moderate osteoarthritis health states were significantly higher with the ALE than the 10-year (TTO: P = 0.03; SG: P = 0.001). This pattern of discrimination was similar in Phase 2.DiscussionResults suggest that the time horizon could influence utility scores and discrimination among health states. When designing utility evaluations, researchers should carefully consider the time horizon so that the value of health states is accurately represented in cost-utility models.

BackgroundCardiovascular diseases are the leading cause of death in the United States, yet a significant proportion of adults at high risk remain undetected by standard screening practices. Polygenic risk score for coronary artery disease (CAD‐PRS) improves precision in determining the 10‐year risk of atherosclerotic cardiovascular disease but health benefits and health care costs associated with CAD‐PRS are unknown. We examined the cost‐effectiveness of including CAD‐PRS as a risk‐enhancing factor in the pooled cohort equation (PCE)—the standard of care for determining the risk of atherosclerotic cardiovascular disease—versus PCE alone.Methods and ResultsWe applied a Markov model on a cohort of 40‐year‐old individuals with borderline or intermediate 10‐year risk (5% to <20%) for atherosclerotic cardiovascular disease to identify those in the top quintile of the CAD‐PRS distribution who are at high risk and eligible for statin prevention therapy. Health outcomes examined included coronary artery disease (CAD; ie, myocardial infarction) and ischemic stroke. The model projected medical costs (2019 US$) of screening for CAD, statin prevention therapy, treatment, and monitoring patients living with CAD or ischemic stroke and quality‐adjusted life‐years for PCE+CAD‐PRS versus PCE alone. Deterministic and probabilistic sensitivity analyses and scenario analyses were performed to examine uncertainty in parameter inputs. PCE+CAD‐PRS was dominant compared with PCE alone in the 5‐ and 10‐year time horizons. We found that, respectively, PCE+CAD‐PRS had 0.003 and 0.011 higher mean quality‐adjusted life‐years and $40 and $181 lower mean costs per person screened, with 29 and 50 fewer events of CAD and ischemic stroke in a cohort of 10 000 individuals compared with PCE alone. The risk of developing CAD, the effectiveness of statin prevention therapy, and the cost of treating CAD had the largest impact on the cost per quality‐adjusted life‐year gained. However, this cost remained below the $50 000 willingness‐to‐pay threshold except when the annual risk of developing CAD was <0.006 in the 5‐year time horizon. Results from Monte Carlo simulation indicated that PCE+CAD‐PRS would be cost‐effective. with the probability of 94% and 99% at $50 000 willingness‐to‐pay threshold in the 5‐ and 10‐year time horizon, respectively.ConclusionsImplementing CAD‐PRS as a risk‐enhancing factor in the PCE to determine the risk of atherosclerotic cardiovascular disease reduced the mean cost per individual, improved quality‐adjusted life‐years, and averted future events of CAD and ischemic stroke when compared with PCE alone.

A parameter generation method based on model checking is proposed to tackle the parameter selection of hierarchical scheduling systems in Integrated Modular Avionics (IMA) by combining the classical symbolic model checking and the Statistical Model Checking (SMC). It builds a generic timed automata network to describe the temporal behavior of hierarchical systems. A distributed genetic algorithm is adopted to search the optimum partition parameters with respect to processor utilization while guaranteeing the schedulability of the system, which is formulated as safety properties of symbolic model checking and hypothesis testing of SMC. Comparing with the widely-used response time analysis, the formal model of this method is more expressive to cover complex features. The application of SMC alleviates the "state space explosion" of classical model checking. Finally, the parameter generation experiments show that the present method is able to find the global optimum solutions in the parameter space.

Most works of recommender system seek to provide highly accurate item prediction while having potentially great bias to popular items. Both users and items' providers will suffer if their system has strong preference for monotonous popular items. A better system should consider also item novelty. Previous works of novel recommendation focus mainly on re-ranking a top-N list generated by an accuracy-focused base model. As a result, these frameworks are 2-stage and essentially limited to the base model. In addition, when training the base model, the common BRP loss function treats all pairs in the same manner, consistently suppresses interesting negative items which should have been recommended. In this work, we propose a personalized pairwise novelty weighting for BPR loss function, which covers the limitations of BPR and effectively improves novelty with marginal loss in accuracy. Base model will be guided by the loss weights to learn user preference and to generate novel suggestion list in 1 stage. Comprehensive experiments on 3 public datasets show that our approach effectively promotes novelty with almost no decrease in accuracy.

Despite significant advances in recent years, structure-from-motion (SfM) pipelines suffer from two important drawbacks. Apart from requiring significant computational power to solve the large-scale computations involved, such pipelines sometimes fail to correctly reconstruct when the accumulated error in incremental reconstruction is large or when the number of 3D to 2D correspondences are insufficient. In this paper we present a novel approach to mitigate the above-mentioned drawbacks. Using an image match graph based on matching features we partition the image data set into smaller sets or components which are reconstructed independently. Following such reconstructions we utilise the available epipolar relationships that connect images across components to correctly align the individual reconstructions in a global frame of reference. This results in both a significant speed up of at least one order of magnitude and also mitigates the problems of reconstruction failures with a marginal loss in accuracy. The effectiveness of our approach is demonstrated on some large-scale real world data sets.

Rare event simulation for highly dependable systems with fast repairs

With the development of swarm intelligence in last decades, consensus problem of multi-agent systems has been attracting much attention. To reveal the inherent mechanism of leader-following consensus in multi-agent systems with stochastic dynamics, PI control protocols are designed in this paper. Owing to the stochastic analysis techniques, algebraic graph theory, and constructing appropriate Lyapunov function, it is proved that the leader-following consensus of nonlinear stochastic dynamical multi-agent systems can be reached in mean square. Sufficient condition are deduced to select the PI control protocol parameters. Finally, the theoretical results are demonstrated through a simulation example.

Transaction‐level modeling with SystemC has been very successful in describing the behavior of embedded systems by providing high‐level executable models, in which many of them have inherent probabilistic behaviors, eg, random data and unreliable components. It is thus crucial to have both quantitative and qualitative analysis of the probabilities of system properties.Such analysis can be conducted by constructing a formal model of the system under verification and using Probabilistic Model Checking. However, this method is infeasible for large systems, due to the state space explosion. In this article, we demonstrate the successful use of statistical model checking to conduct such analysis directly from large SystemC models and allow designers to express a wide range of useful properties.The first contribution of this work is a framework to verify properties expressed in Bounded Linear Temporal Logic for SystemC models with both timed and probabilistic characteristics.Second, the framework allows users to expose a rich set of user code primitives as atomic propositions in Bounded Linear Temporal Logic. Moreover, users can define their own fine‐grained time resolution rather than the boundary of clock cycles in the SystemC simulation.The third contribution is an implementation of a statistical model checker. It contains an automatic monitor generation for producing execution traces of the model‐under‐verification, the mechanism for automatically instrumenting the model‐under‐verification, and the interaction with statistical model checking algorithms.

Neural networks (NN) are taking over ever more decisions thus far taken by humans, even though verifiable system-level guarantees are far out of reach. Neither is the verification technology available, nor is it even understood what a formal, meaningful, extensible, and scalable testbed might look like for such a technology. The present paper is an attempt to improve on both the above aspects. We present a family of formal models that contain basic features of automated decision-making contexts and which can be extended with further orthogonal features, ultimately encompassing the scope of autonomous driving. Due to the possibility to model random noise in the decision actuation, each model instance induces a Markov decision process (MDP) as verification object. The NN in this context has the duty to actuate (near-optimal) decisions. From the verification perspective, the externally learnt NN serves as a determinizer of the MDP, the result being a Markov chain which as such is amenable to statistical model checking. The combination of an MDP and an NN encoding the action policy is central to what we call “deep statistical model checking” (DSMC). While being a straightforward extension of statistical model checking, it enables to gain deep insight into questions like “how high is the NN-induced safety risk?”, “how good is the NN compared to the optimal policy?” (obtained by model checking the MDP), or “does further training improve the NN?”. We report on an implementation of DSMC inside the ModestToolset in combination with externally learnt NNs, demonstrating the potential of DSMC on various instances of the model family, and illustrating its scalability as a function of instance size as well as other factors like the degree of NN training.

SummaryMany studies of adaptive harvest management already exist in the literature, but most (if not all) use long, sometimes infinite, time horizons. Such long‐term objectives provide an opportunity to manage experimentally, so that poorly understood dynamics are learned and any returns sacrificed for experimentation are repaid by improved management over the remaining time horizon.However, a manager is unlikely to weight outcomes in the distant future equally against outcomes in the present. Furthermore, the most appropriate model of system dynamics may not remain constant over the time‐frame required to experiment, learn and improve management. In these cases the use of discounting and/or a finite time horizon fit the manager's assumptions and goals more effectively, and the value of experimentation is likely to be diminished.In this paper we construct a simple model of a hypothetical population and compare optimal passive and active adaptive harvest strategies over a range of time horizons. This allows us to determine the optimal level of experimentation for short‐, medium‐ and long‐term goals.We discover that the optimal active adaptive harvest strategy may be precautionary over short to medium time horizons, rather than experimental. That is, an action with known moderate benefits is preferred over an action with uncertain but marginally larger expected benefits. This runs counter to the widespread assumption in the adaptive management literature that incorporating learning into an optimization of management will encourage experimentation.Synthesis and applications. The general results of this paper have potential application to any environmental management problem where adaptive management might be applied; for example, conservation, pest control, harvesting and management of water flows. We examine adaptive management over a range of finite time horizons to reflect a variety of possible management goals and assumptions. Our simple example demonstrates that in the face of model uncertainty, the management strategy that maximizes benefits does not necessarily include deliberate experimentation and learning. Optimal active adaptive management weighs experimentation against all its potential consequences, and this can yield a precautionary approach.

In this article, the author argues that the time horizon a dictator faces affects his incentives over the use of aid in three ways. First, dictators have a greater incentive to invest in public goods when they have a long time horizon. Second, dictators with short time horizons often face the threat of challengers to the regime; this leads them to forgo investment and instead consume state resources in two forms that harm growth: repression and private pay-offs to political opponents. Third, dictators with short time horizons have a strong incentive to secure personal wealth as a form of insurance in case the regime falls. Using panel data on dictatorships in 71 developing countries from 1961 to 2001, the author finds that time horizons have a positive impact on aid effectiveness: Foreign aid is associated with positive growth when dictators face long time horizons and negative growth when time horizons are short.

In this article, we provide a review of the consensus controlproblem for stochastic multi-agent systems (MASs). Recent advancesare surveyed according to the method of occurrence of the stochasticity of the MASs. First, the consensus problem is discussed for MASs, whereinindividual agents are corrupted by random noises, i.e., thedynamics of agents involve stochasticity in process and/ormeasurement equations. Both additive noises and multiplicativenoises are surveyed in detail and special attention is paid to theMASs whose dynamics are governed by Itodifferential equations.Moreover, particular effort is devoted to presenting the latestprogress on the consensus problem for a special type of stochasticMAS with Markovian jump parameters. Subsequently, the relevant research issummarized for MASs with noisy communication environments andstochastic sampling. Further, we provide a systematic review ofthe consensus problems for MASs whose communication topologyvaries randomly in the process of data propagation among agents.Finally, conclusions are drawn and several potential future researchdirections are outlined.