User-expertise modeling with empirically derived probabilistic implication networks

The application of user-expertise modeling for adaptive interfaces is confronted with a number of difficult challenges, namely, efficiency and reliability, the cost-benefit ratio, and the practical usability of user modeling techniques. We argue that many of these obstacles can be overcome by standard, automatic means of performing knowledge assessment. Within this perspective, we present the basis of a probabilistic user modeling approach, the POKS technique, which could serve as a standard user-expertise modeling tool.The POKS technique is based on the cognitive theory of knowledge structures: a formalism for the representation of the order in which we learn knowledge units (KU). The technique permits the induction of knowledge structures from a small number of empirical data cases. It uses an evidence propagation scheme within these structures to infer an individual's knowledge state from a sample of KU. The empirical induction technique is based, in part, on statistical hypothesis testing over conditional probabilities that are determined by the KUs' learning order.Experiments with this approach show that the technique is successful in partially inferring an individual's knowledge state, either through the monitoring of a user's behavior, or through a selective questioning process. However, the selective process, based on entropy minimization, is shown to be much more effective in reducing the standard error score of knowledge assessment than random sampling.     

Causal probabilistic networks with both discrete and continuousvariables

An extension of the expert system shell known as handling uncertainty by general influence networks (HUGIN) to include continuous variables, in the form of linear additive normally distributed variables, is presented. The theoretical foundation of the method was developed by S.L. Lauritzen, whereas this report primarily focus on implementation aspects. The approach has several advantages over purely discrete systems. It enables a more natural model of of the domain in question, knowledge acquisition is eased, and the complexity of belief revision is most often reduced considerably     

Hybridization of the probabilistic neural networks with feed-forward neural networks for forecasting

Feed-forward neural networks (FFNNs) are among the most important neural networks that can be applied to a wide range of forecasting problems with a high degree of accuracy. Several large-scale forecasting competitions with a large number of commonly used time series forecasting models conclude that combining forecasts from more than one model often leads to improved performance, especially when the models in the ensemble are quite different. In the literature, several hybrid models have been proposed by combining different time series models together. In this paper, in contrast of the traditional hybrid models, a novel hybridization of the feed-forward neural networks (FFNNs) is proposed using the probabilistic neural networks (PNNs) in order to yield more accurate results than traditional feed-forward neural networks. In the proposed model, the estimated values of the FFNN models are modified based on the distinguished trend of their residuals and optimum step length, which are respectively yield from a probabilistic neural network and a mathematical programming model. Empirical results with three well-known real data sets indicate that the proposed model can be an effective way in order to construct a more accurate hybrid model than FFNN models. Therefore, it can be applied as an appropriate alternative model for forecasting tasks, especially when higher forecasting accuracy is needed.     

Mixed deterministic and probabilistic networks

The paper introduces mixed networks, a new graphical model framework for expressing and reasoning with probabilistic and deterministic information. The motivation to develop mixed networks stems from the desire to fully exploit the deterministic information (constraints) that is often present in graphical models. Several concepts and algorithms specific to belief networks and constraint networks are combined, achieving computational efficiency, semantic coherence and user-interface convenience. We define the semantics and graphical representation of mixed networks, and discuss the two main types of algorithms for processing them: inference-based and search-based. A preliminary experimental evaluation shows the benefits of the new model.     

Labeling evaluation in probabilistic networks

In many pattern recognition problems a probabilistic labeling of a network is given, and it is desired to obtain a unique unambiguous labeling for the network. This labeling should be influenced by the given probabilistic labeling, and by the joint distributions of the labels at subsets of nodes of the network. Relaxation algorithms have frequently been used to find such a labeling, but no method has been available to evaluate the results or to compare two different labelings. A measure is proposed here for evaluating labelings based on the given probabilistic labeling and joint distributions. This measure can also be used to define a termination criterion for relaxation by evaluating the labeling at each iteration. In addition, it could be used to evaluate labelings derived by any other process, and to guide heuristic search.     

Context-specific sign-propagation in qualitative probabilistic networks

Qualitative probabilistic networks are qualitative abstractions of probabilistic networks, summarising probabilistic influences by qualitative signs. As qualitative networks model influences at the level of variables, knowledge about probabilistic influences that hold only for specific values cannot be expressed. The results computed from a qualitative network, as a consequence, can be weaker than strictly necessary and may in fact be rather uninformative. We extend the basic formalism of qualitative probabilistic networks by providing for the inclusion of context-specific information about influences and show that exploiting this information upon reasoning has the ability to forestall unnecessarily weak results.     

Controllability of probabilistic Boolean control networks

This paper deals with the controllability of probabilistic Boolean control networks. First, a survey on the semi-tensor product approach to probabilistic Boolean networks is given. Second, the controllability of probabilistic Boolean control networks via two kinds inputs is studied. Finally, examples are given to show the efficiency of the obtained results.     

Modeling preventive maintenance of manufacturing processes with probabilistic Boolean networks with interventions

Recent developments in intelligent manufacturing have validated the use of probabilistic Boolean networks (PBN) to model failures in manufacturing processes and as part of a methodology for Design Failure Mode and Effects Analysis (DFMEA). This paper expands the application of PBNs in manufacturing processes by proposing the use of interventions in PBNs to model an ultrasound welding process in a preventive maintenance (PM) schedule, guiding the process to avoid failure and extend its useful work life. This bio-inspired, stochastic methodology uses PBNs with interventions to model manufacturing processes under a PM schedule and guides the evolution of the network, providing a new mechanism for the study and prediction of the future behavior of the system at the design phase, assessing future performance, and identifying areas to improve design reliability and system resilience. A process engineer designing manufacturing processes may use this methodology to create new or improve existing manufacturing processes, assessing risk associated with them, and providing insight into the possible states, operating modes, and failure modes that can occur. The engineer can also guide the process and avoid states that can result in failure, and design an appropriate PM schedule. The proposed method is applied to an ultrasound welding process. A PBN with interventions model was simulated and verified using model checking in PRISM, generating data required to conduct inferential statistical tests to compare the effects of probability of failures between the PBN and PBN with Interventions models. The obtained results demonstrate the validity of the proposed methodology.     

Enhanced qualitative probabilistic networks for resolving trade-offs

Qualitative probabilistic networks were designed to overcome, to at least some extent, the quantification problem known to probabilistic networks. Qualitative networks abstract from the numerical probabilities of their quantitative counterparts by using signs to summarise the probabilistic influences between their variables. One of the major drawbacks of these qualitative abstractions, however, is the coarse level of representation detail that does not provide for indicating strengths of influences. As a result, the trade-offs modelled in a network remain unresolved upon inference. We present an enhanced formalism of qualitative probabilistic networks to provide for a finer level of representation detail. An enhanced qualitative probabilistic network differs from a basic qualitative network in that it distinguishes between strong and weak influences. Now, if a strong influence is combined, upon inference, with a conflicting weak influence, the sign of the net influence may be readily determined. Enhanced qualitative networks are purely qualitative in nature, as basic qualitative networks are, yet allow for resolving some trade-offs upon inference.     

Maximum likelihood training of probabilistic neural networks

A maximum likelihood method is presented for training probabilistic neural networks (PNN's) using a Gaussian kernel, or Parzen window. The proposed training algorithm enables general nonlinear discrimination and is a generalization of Fisher's method for linear discrimination. Important features of maximum likelihood training for PNN's are: 1) it economizes the well known Parzen window estimator while preserving feedforward NN architecture, 2) it utilizes class pooling to generalize classes represented by small training sets, 3) it gives smooth discriminant boundaries that often are "piece-wise flat" for statistical robustness, 4) it is very fast computationally compared to backpropagation, and 5) it is numerically stable. The effectiveness of the proposed maximum likelihood training algorithm is assessed using nonparametric statistical methods to define tolerance intervals on PNN classification performance.     

State feedback stabilization for probabilistic Boolean networks

A probabilistic Boolean network (PBN) is a discrete-time system composed of a family of Boolean networks (BNs) between which the PBN switches in a stochastic fashion. Studying control-related problems in PBNs may provide new insights into the intrinsic control in biological systems and enable us to develop strategies for manipulating complex biological systems using exogenous inputs. This paper investigates the problem of state feedback stabilization for PBNs. Based on the algebraic representation of logic functions, a necessary and sufficient condition is derived for the existence of a globally stabilizing state feedback controller, and a control design method is proposed when the presented condition holds. It is shown that the controller designed via the proposed procedure can simultaneously stabilize a collection of PBNs that are composed of the same constituent BNs.     

Shortest distance and reliability of probabilistic networks

When the “length” of a link is not deterministic and is governed by a stochastic process, the “shortest” path between two points in the network is not necessarily always composed of the same links and depends on the state of the network. For example, in communication and transportation networks, the travel time on a link is not deterministic and the fastest path between two points is not fixed. This paper presents an algorithm to compute the expected shortest travel time between two nodes in the network when the travel time on each link has a given independent discrete probability distribution. The algorithm assumes the knowledge of all the paths between two nodes and methods to determine the paths are referenced.In reliability (i.e. the probability that two given points are connected by a path) computations, associated with each link is a probability of “failure” and a probability of “success”. Since “failure” implies infinite travel time, the algorithm simultaneously computes reliability. The paper also discusses the algorithm's capability to simultaneously compute some other performance measures which are useful in the analysis of emergency services operating on a network.     

Optimal paths in probabilistic networks: A case with temporary preferences

The classical shortest route problem in networks assumes deterministic arc weights and a utility (or cost) function that is linear over path weights for route evaluation. When the environment is stochastic and the “traveler's” utility function for travel attributes is nonlinear, we define “optimal paths” that maximize the expected utility.We review the concepts of temporary and permanent preferences for comparing a traveler's preference for available subpaths. It has been shown before that when the utility function is linear or exponential, permanent preferences prevail and an efficient Dijkstra-type algorithm [3] is available that determines the optimal path.In this paper an exact procedure is developed for determining an optimal path when the utility function is quadratic—a case where permanent preferences do not always prevail. The algorithm uses subpath comparison rules to establish permanent preferences, when possible, among subpaths of the given network. Although in the worst case the algorithm implicitly enumerates all paths (the number of operations increasing exponentially with the size of the network), we find, from the computational experience reported, that the number of potentially optimal paths to evaluate is generally manageable.     

Biologically inspired probabilistic coverage for mobile sensor networks

Coupling sensors in a sensor network with mobility mechanism can boost the performance of wireless sensor networks (WSNs). In this paper, we address the problem of self-deploying mobile sensors to reach high coverage. The problem is modeled as a multi-objective optimization that simultaneously minimizes two contradictory parameters; the total sensor moving distance and the total uncovered area. In order to resolve the aforementioned deployment problem, this study investigates the use of biologically inspired mechanisms, including evolutionary algorithms and swarm intelligence, with their state-of-the-art algorithms. Unlike most of the existing works, the coverage parameter is expressed as a probabilistic inference model due to uncertainty in sensor readings. To the best of our knowledge, probabilistic coverage of mobile sensor networks has not been addressed in the context of multi-objective bio-inspired algorithms. Performance evaluations on deployment quality and deployment cost are measured and analyzed through extensive simulations, showing the effectiveness of each algorithm under the developed objective functions. Simulations reveal that only one multi-objective evolutionary algorithm; the so-called multi-objective evolutionary algorithm with decomposition survives to effectively tackle the probabilistic coverage deployment problem. It gathers more than 78 % signals from all of the targets (and in some cases reaches 100 % certainty). On the other hand, non-dominated sorting genetic algorithm II, multi-objective particle swarm optimization, and non-dominated sorting particle swarm optimization show inferior performance down to 16–32 %, necessitating further modifications in their internal mechanisms.     

Equivalence between RAM-based neural networks and probabilistic automata

In this letter, the computational power of a class of random access memory (RAM)-based neural networks, called general single-layer sequential weightless neural networks (GSSWNNs), is analyzed. The theoretical results presented, besides helping the understanding of the temporal behavior of these networks, could also provide useful insights for the developing of new learning algorithms.     

基于概率神经网络的入侵检测技术

提出一种基于概率神经网络的高效入侵检测技术。对网络数据处理、概率神经网络的训练与检测及其算法进行分析。在网络训练中,提出一种基于实验数据选择概率神经网络关键参数的方法,分析该方法的可行性。实验表明通过此方法能使入侵检测系统具有更高的检测精度和效率。     

Convergence analysis of max-consensus algorithm in probabilistic communication networks with Bernoulli dropouts

In the presence of probabilistic communication networks between agents, the convergence analysis of max-consensus algorithm (MCA) is addressed in this paper. It is considered that at each iteration of MCA, all agents share their measurements with adjacent agents via local communication networks which is applicable in many multi-agent systems (MASs). It is assumed that the communication networks have Bernoulli dropouts, i.e. the information exchanged between agents may be lost with Bernoulli distribution. In the proposed method, the information topology of MAS is modelled as a dynamic graph with the Bernoulli adjacency matrix. It is proved that in the presence of Bernoulli dropouts and under non-restrictive assumptions concerning the MAS features and communication topology, the MCA converges with a probability one in the finite time. Furthermore, the upper bounds are provided by means of deterministic and probabilistic expressions for the expectation and dispersion of convergence time, respectively. It is shown that the proposed upper bounds are asymptotic, i.e. there are specific conditions of MAS in which the convergence time of MCA tends to the proposed upper bounds. The convergence accuracy of MCA is discussed in terms of probabilistic equations. The validity of the proposed theorems is illustrated by means of simulation results.     

Joint probabilistic data association in distributed sensor networks

A distributed multitarget tracking problem is considered. The joint probabilistic data association (JPDA) algorithm, which has been successfully used for tracking multiple targets in a cluttered environment, assumes a centralized processing architecture. It assumes that measurements are transmitted to a central site and processed. In some applications, however, it may be desirable for the sensor measurements to be processed at or near the sensors instead of transmitting them to the central processor. The local processed results are then sent over a communication network to be used by other processors. This paper presents a distributed version of the JPDA algorithm which is applicable under such a situation.     

Adjusted probabilistic route discovery in mobile ad hoc networks

Conventional on-demand route discovery methods in mobile ad hoc networks (MANET) employ simple flooding method, where a mobile node blindly rebroadcasts received route request (RREQ) packets until a route to a particular destination is established. This can potentially lead to high channel contention, causing redundant retransmissions and thus excessive packet collisions in the network. This paper proposed two new probabilistic methods that can significantly reduce the number of RREQ packets transmitted during route discovery operation. Our simulation analysis reveals that equipping AODV with an appropriate probabilistic route discovery method can result in significant performance improvements in terms of routing overhead, MAC collisions and end-to-end delay while still achieving a good throughput when compared with the traditional AODV.