This article proposes the optimal window-symbolic time series analysis (OW-STSA) methodology to optimize parameters of feature extraction and pattern classification in industrial processes. The underlying theory is built upon minimization of an empirical risk function to discriminate between nominal and anomalous operations of the physical process under consideration. In particular, the proposed methodology produces: optimized windows of the time series used for pattern classification and anomaly detection, and optimized identification of feature extractors and classifier parameters. The algorithm is realized by segmenting a given time series into windows of equal size. Then, the stationary state probability vector is computed for each window in the sense of OW-STSA for anomaly prediction with locally optimal accuracy of detection performance. The proposed methodology has been experimentally validated in laboratory environment with different classifiers for two distinct industrial processes. The first experiment addresses detection of fatigue failure in polycrystalline alloy structures using time series of ultrasonic signals. The second experiment investigates detection of thermoacoustic instability in an emulated combustion system using time series of pressure-wave signals. In both experiments, the proposed OW-STSA methodology yielded excellent detection performance of anomalous behavior with multiple classification techniques.

This letter deals with neural networks as dynamical systems governed by finite difference equations. It shows that the introduction of [Formula: see text]-many skip connections into network architectures, such as residual networks and additive dense networks, defines [Formula: see text]th order dynamical equations on the layer-wise transformations. Closed-form solutions for the state-space representations of general [Formula: see text]th order additive dense networks, where the concatenation operation is replaced by addition, as well as [Formula: see text]th order smooth networks, are found. The developed provision endows deep neural networks with an algebraic structure. Furthermore, it is shown that imposing [Formula: see text]th order smoothness on network architectures with [Formula: see text]-many nodes per layer increases the state-space dimension by a multiple of [Formula: see text], and so the effective embedding dimension of the data manifold by the neural network is [Formula: see text]-many dimensions. It follows that network architectures of these types reduce the number of parameters needed to maintain the same embedding dimension by a factor of [Formula: see text] when compared to an equivalent first-order, residual network. Numerical simulations and experiments on CIFAR10, SVHN, and MNIST have been conducted to help understand the developed theory and efficacy of the proposed concepts.

Estimation of a generating partition is critical for symbolization of measurements from discrete-time dynamical systems, where a sequence of symbols from a (finite-cardinality) alphabet may uniquely specify the underlying time series. Such symbolization is useful for computing measures (e.g., Kolmogorov-Sinai entropy) to identify or characterize the (possibly unknown) dynamical system. It is also useful for time series classification and anomaly detection. The seminal work of Hirata, Judd, and Kilminster ( 2004 ) derives a novel objective function, akin to a clustering objective, that measures the discrepancy between a set of reconstruction values and the points from the time series. They cast estimation of a generating partition via the minimization of their objective function. Unfortunately, their proposed algorithm is nonconvergent, with no guarantee of finding even locally optimal solutions with respect to their objective. The difficulty is a heuristic nearest neighbor symbol assignment step. Alternatively, we develop a novel, locally optimal algorithm for their objective. We apply iterative nearest-neighbor symbol assignments with guaranteed discrepancy descent, by which joint, locally optimal symbolization of the entire time series is achieved. While most previous approaches frame generating partition estimation as a state-space partitioning problem, we recognize that minimizing the Hirata et al. ( 2004 ) objective function does not induce an explicit partitioning of the state space, but rather the space consisting of the entire time series (effectively, clustering in a (countably) infinite-dimensional space). Our approach also amounts to a novel type of sliding block lossy source coding. Improvement, with respect to several measures, is demonstrated over popular methods for symbolizing chaotic maps. We also apply our approach to time-series anomaly detection, considering both chaotic maps and failure application in a polycrystalline alloy material.

This paper presents information-theoretic performance analysis of passive sensor networks for detection of moving targets. The proposed method falls largely under the category of data-level information fusion in sensor networks. To this end, a measure of information contribution for sensors is formulated in a symbolic dynamics framework. The network information state is approximately represented as the largest principal component of the time series collected across the network. To quantify each sensor's contribution for generation of the information content, Markov machine models as well as x-Markov (pronounced as cross-Markov) machine models, conditioned on the network information state, are constructed; the difference between the conditional entropies of these machines is then treated as an approximate measure of information contribution by the respective sensors. The x-Markov models represent the conditional temporal statistics given the network information state. The proposed method has been validated on experimental data collected from a local area network of passive sensors for target detection, where the statistical characteristics of environmental disturbances are similar to those of the target signal in the sense of time scale and texture. A distinctive feature of the proposed algorithm is that the network decisions are independent of the behavior and identity of the individual sensors, which is desirable from computational perspectives. Results are presented to demonstrate the proposed method's efficacy to correctly identify the presence of a target with very low false-alarm rates. The performance of the underlying algorithm is compared with that of a recent data-driven, feature-level information fusion algorithm. It is shown that the proposed algorithm outperforms the other algorithm.

This paper presents a sequential learning framework for sensors in a network, where a few sensors assume the role of an instructor to train other sensors in the network. The instructors provide estimated labels for measurements of new sensors. These labels are possibly noisy, because a classifier of the instructor may not be perfect. A recursive density estimator is proposed to obtain the true measurement model (i.e., the observation density conditioned on the label) in spite of the training with noisy labels. Specifically, this paper answers the question "Can a sensor train other sensors?", provides necessary conditions for sensors to act as instructors, presents a sequential learning framework using recursive nonparametric kernel density estimation, and provides a convergence rate for the expected error in an observation density. The underlying concepts are illustrated and validated with simulation results.

This paper proposes a Bayesian nonparametric regression model of panel data for sequential pattern classification. The proposed method provides a flexible and parsimonious model that allows both time-independent spatial variables and time-dependent exogenous variables to be predictors. Not only this method improves the accuracy of parameter estimation for limited data, but also it facilitates model interpretation by identifying statistically significant predictors with hypothesis testing. Moreover, as the data length approaches infinity, posterior consistency of the model is guaranteed for general data-generating processes under regular conditions. The resulting model of panel data can also be used for sequential classification. The proposed method has been tested by numerical simulation, then validated on an econometric public data set, and subsequently validated for detection of combustion instabilities with experimental data that have been generated in a laboratory environment.

This study deals with neural networks in the sense of geometric transformations acting on the coordinate representation of the underlying data manifold which the data is sampled from. It forms part of an attempt to construct a formalized general theory of neural networks in the setting of Riemannian geometry. From this perspective, the following theoretical results are developed and proven for feedforward networks. First it is shown that residual neural networks are finite difference approximations to dynamical systems of first order differential equations, as opposed to ordinary networks that are static. This implies that the network is learning systems of differential equations governing the coordinate transformations that represent the data. Second it is shown that a closed form solution of the metric tensor on the underlying data manifold can be found by backpropagating the coordinate representations learned by the neural network itself. This is formulated in a formal abstract sense as a sequence of Lie group actions on the metric fibre space in the principal and associated bundles on the data manifold. Toy experiments were run to confirm parts of the proposed theory, as well as to provide intuitions as to how neural networks operate on data.

This paper addresses the problem of target detection in dynamic environments in a semi-supervised data-driven setting with low-cost passive sensors. A key challenge here is to simultaneously achieve high probabilities of correct detection with low probabilities of false alarm under the constraints of limited computation and communication resources. In general, the changes in a dynamic environment may significantly affect the performance of target detection due to limited training scenarios and the assumptions made on signal behavior under a static environment. To this end, an algorithm of binary hypothesis testing is proposed based on clustering of features extracted from multiple sensors that may observe the target. First, the features are extracted individually from time-series signals of different sensors by using a recently reported feature extraction tool, called symbolic dynamic filtering. Then, these features are grouped as clusters in the feature space to evaluate homogeneity of the sensor responses. Finally, a decision for target detection is made based on the distance measurements between pairs of sensor clusters. The proposed procedure has been experimentally validated in a laboratory setting for mobile target detection. In the experiments, multiple homogeneous infrared sensors have been used with different orientations in the presence of changing ambient illumination intensities. The experimental results show that the proposed target detection procedure with feature-level sensor fusion is robust and that it outperforms those with decision-level and data-level sensor fusion.

Dictionary learning algorithms have been successfully used for both reconstructive and discriminative tasks, where an input signal is represented with a sparse linear combination of dictionary atoms. While these methods are mostly developed for single-modality scenarios, recent studies have demonstrated the advantages of feature-level fusion based on the joint sparse representation of the multimodal inputs. In this paper, we propose a multimodal task-driven dictionary learning algorithm under the joint sparsity constraint (prior) to enforce collaborations among multiple homogeneous/heterogeneous sources of information. In this task-driven formulation, the multimodal dictionaries are learned simultaneously with their corresponding classifiers. The resulting multimodal dictionaries can generate discriminative latent features (sparse codes) from the data that are optimized for a given task such as binary or multiclass classification. Moreover, we present an extension of the proposed formulation using a mixed joint and independent sparsity prior, which facilitates more flexible fusion of the modalities at feature level. The efficacy of the proposed algorithms for multimodal classification is illustrated on four different applications--multimodal face recognition, multi-view face recognition, multi-view action recognition, and multimodal biometric recognition. It is also shown that, compared with the counterpart reconstructive-based dictionary learning algorithms, the task-driven formulations are more computationally efficient in the sense that they can be equipped with more compact dictionaries and still achieve superior performance.

Dictionary learning algorithms have been successfully used in both reconstructive and discriminative tasks, where the input signal is represented by a linear combination of a few dictionary atoms. While these methods are usually developed under ℓ1sparsity constrain (prior) in the input domain, recent studies have demonstrated the advantages of sparse representation using structured sparsity priors in the kernel domain. In this paper, we propose a supervised dictionary learning algorithm in the kernel domain for hyperspectral image classification. In the proposed formulation, the dictionary and classifier are obtained jointly for optimal classification performance. The supervised formulation is task-driven and provides learned features from the hyperspectral data that are well suited for the classification task. Moreover, the proposed algorithm uses a joint (ℓ12) sparsity prior to enforce collaboration among the neighboring pixels. The simulation results illustrate the efficiency of the proposed dictionary learning algorithm.

The advent of intelligent vehicle technologies holds significant potential to alter the dynamics of traffic flow. Prior work on the effects of such technologies on the formation of self-organized traffic jams has led to analytical solutions and numerical simulations at the mesoscopic scale, which may not yield significant information about the distribution of vehicle cluster size. Since the absence of large clusters could be offset by the presence of several smaller clusters, the distribution of cluster sizes can be as important as the presence or absence of clusters. To obtain a prediction of vehicle cluster distribution, the included work presents a statistical mechanics-inspired method of simulating traffic flow at a microscopic scale via the generalized Ising model. The results of the microscopic simulations indicate that traffic systems dominated by adaptive cruise control ( acc)-enabled vehicles exhibit a higher probability of formation of moderately sized clusters, as compared with the traffic systems dominated by human-driven vehicles; however, the trend is reversed for the formation of large-sized clusters. These qualitative results hold significance for algorithm design and traffic control because it is easier to predict and take countermeasures for fewer large localized clusters as opposed to several smaller clusters spread across different locations on a highway.

Recent studies have demonstrated advantages of information fusion based on sparsity models for multimodal classification. Among several sparsity models, tree-structured sparsity provides a flexible framework for extraction of cross-correlated information from different sources and for enforcing group sparsity at multiple granularities. However, the existing algorithm only solves an approximated version of the cost functional and the resulting solution is not necessarily sparse at group levels. This paper reformulates the tree-structured sparse model for multimodal classification task. An accelerated proximal algorithm is proposed to solve the optimization problem, which is an efficient tool for feature-level fusion among either homogeneous or heterogeneous sources of information. In addition, a (fuzzy-set-theoretic) possibilistic scheme is proposed to weight the available modalities, based on their respective reliability, in a joint optimization problem for finding the sparsity codes. This approach provides a general framework for quality-based fusion that offers added robustness to several sparsity-based multimodal classification algorithms. To demonstrate their efficacy, the proposed methods are evaluated on three different applications - multiview face recognition, multimodal face recognition, and target classification.

This paper presents a procedure for behavior identification of mobile robots, which requires limited or no domain knowledge of the underlying process. While the features of robot behavior are extracted by symbolic dynamic filtering of the observed time series, the behavior patterns are classified based on language measure theory. The behavior identification procedure has been experimentally validated on a networked robotic test bed by comparison with commonly used tools, namely, principal component analysis for feature extraction and Bayesian risk analysis for pattern classification.

This paper presents the qualitative nature of communication network operations as abstraction of typical thermodynamic parameters (e.g., order parameter, temperature, and pressure). Specifically, statistical mechanics-inspired models of critical phenomena (e.g., phase transitions and size scaling) for heterogeneous packet transmission are developed in terms of multiple intensive parameters, namely, the external packet load on the network system and the packet transmission probabilities of heterogeneous packet types. Network phase diagrams are constructed based on these traffic parameters, and decision and control strategies are formulated for heterogeneous packet transmission in the network system. In this context, decision functions and control objectives are derived in closed forms, and the pertinent results of test and validation on a simulated network system are presented.

This paper presents a statistical-mechanics-inspired procedure for optimization of the sensor field configuration to detect mobile targets. The key idea is to capture the low-dimensional behavior of the sensor field configurations across the Pareto front in a multiobjective scenario for optimal sensor deployment, where the nondominated points are concentrated within a small region of the large-dimensional decision space. The sensor distribution is constructed using location-dependent energy-like functions and intensive temperature-like parameters in the sense of statistical mechanics. This low-dimensional representation is shown to permit rapid optimization of the sensor field distribution on a high-fidelity simulation test bed of distributed sensor networks.

Some of the critical and practical issues regarding the problem of health monitoring of multi-component human-engineered systems have been discussed, and a syntactic method has been proposed. The method involves abstraction of a qualitative description from a general dynamical system structure, using state space embedding of the output data-stream and discretization of the resultant pseudo state and input spaces. The system identification is achieved through grammatical inference techniques, and the deviation of the plant output from the nominal estimated language gives a measure of anomaly in the system. The technique is validated on an experimental test-bed of a permanent magnet synchronous motor undergoing a gradual degradation of the encoder orientation feedback.

This paper formulates a self-organization algorithm to address the problem of global behavior supervision in engineered swarms of arbitrarily large population sizes. The swarms considered in this paper are assumed to be homogeneous collections of independent identical finite-state agents, each of which is modeled by an irreducible finite Markov chain. The proposed algorithm computes the necessary perturbations in the local agents' behavior, which guarantees convergence to the desired observed state of the swarm. The ergodicity property of the swarm, which is induced as a result of the irreducibility of the agent models, implies that while the local behavior of the agents converges to the desired behavior only in the time average, the overall swarm behavior converges to the specification and stays there at all times. A simulation example illustrates the underlying concept.

We present a new paradigm to control software systems based on the supervisory control theory (SCT). Our method uses the SCT to model the execution of a software application by restricting the actions of the OS with little or no modifications in the underlying OS. Our approach can be generalized to any software application as the interactions of the application with the OS are modeled at a process level as a deterministic finite state automaton (DFSA) termed as a "plant." A "supervisor" that controls the plant is a DFSA synthesized from a set of control specifications. The supervisor operates synchronously with the plant to restrict the language accepted by the plant to satisfy the control specifications. Using the above method of control to mitigate faults, as a proof-of-concept, we implement two supervisors under the Redhat Linux 7.2 OS to mitigate overflow and segmentation faults in five different programs. We quantify the performance of the unsupervised and supervised plant by using a language measure and give methods to compute the measure using state transition cost matrix and characteristic vector.

This paper presents a system architecture for behavior-based collaborative multi-agent systems in the discrete event setting following the Ramadge and Wonham framework. It addresses the issues of robustness to component failures, reliability of wireless communications, scalability to increase in the number of agents, and quantitative analysis of system performance at different levels of control hierarchy. The mission objectives are achieved by hierarchically structured supervisory control. The standard supervisory control theory is extended in the sense that the event alphabet is made a function of the plant parameters. The supervisory control system allows interactions with external agents including human operators. A proof-of-the-concept control architecture is experimentally validated by a wireless mobile robotic system consisting of three pioneer 2 AT mobile robots. This concept could be extended to other distributed systems provided that the continuous-varying dynamics of the underlying physical process is decoupled with the discrete state space of the supervisory control system.

This paper focuses on conceptual design, development, and implementation of a performance management tool for computer communication networks to serve large-scale integrated systems. The objective is to improve the network performance in handling various types of messages by on-line adjustment of protocol parameters. The techniques of perturbation analysis of Discrete Event Dynamic Systems (DEDS), stochastic approximation (SA), and learning automata have been used in formulating the algorithm of performance management. The efficacy of the performance management tool has been demonstrated on a network testbed. The conceptual design presented offers a step forward to bridging the gap between management standards and users' demands for efficient network operations since most standards such as ISO (International Standards Organization) and IEEE address only the architecture, services, and interfaces for network management. The proposed concept for performance management can also be used as a general framework to assist design, operation, and management of various DEDS such as computer integrated manufacturing and battlefield C/sup 3/ (Command, Control, and Communications).>

This article presents the concept and development of a prototype diagnostic decision support system for real-time control and monitoring of dynamical processes. This decision support system, known as Diagnostic Evaluation and Corrective Action (DECA), employs qualitative reasoning, in conjunction with quantitative models, for monitoring and diagnosis of malfunctions in dynamical processes under routine operations and emergency situations. DECA is especially suited for application to time-constrained environments where an immediate action is needed to avoid catastrophic failure(s). DECA is written in common Lisp and has been implemented on a Symbolics 3670 machine; its efficacy has been verified using the data from the Three Mile Island No. 2 Nuclear Reactor Accident.

A failure detection and isolation (FDI) method for validation of ultrasonic ranging sensor (URS) signals in robot position control systems is presented. The technique builds on the concepts of parity space and analytic redundancy, where integration of analytic and sensor redundancy provides a direct, reliable method for measuring the end-effector position of a robot relative to the world coordinates. These measurements are not influenced by deflections caused by the payload, accumulated joint measurement errors in a serial mechanism, or computational errors in executing kinematic relationships. The position control system's insensitivity to structural deflections allows the robot to handle larger payloads. Simulation results are presented to demonstrate how the FDI technique can be applied.>

The concept, theory, and the issues that are pertinent to the design and implementation of an adaptive real-time, intelligent seam tracker (ARTIST) are discussed. ARTIST is designed to operate in the unstructured environment of 3-D seams without preprogramming. A salient feature of the ARTIST design is that the tracking errors in any cycle do not affect the system performance during subsequent cycles. The conceptual and analytical development of a two-tier hierarchical control scheme for ARTIST is presented. The high-level controller models and interacts with the global seam environment while the low-level controller positions the torch over the seam with the correct orientation.>

For pt.I see ibid., p.1904-9 (1990). The design and implementation issues for ARTIST (adaptive real-time intelligent seam tracker), which is essentially a microcomputer-based single-pass system characterized by the absence of an additional teaching phase, are addressed. ARTIST uses a welding torch and a laser-ranging sensor mounted on the end-effector of a six-axis robot. Sources of error that could potentially contribute to inaccuracy in the seam tracking process are discussed. Suggestions are also made for enhancing the performance and applicability of ARTIST for welding automation.>

A brief overview is given of the special problems to be solved by the integrated communication network for computer-integrated manufacturing (CIM) and of fiber-optics-based networks. Network topologies considered include star, ring, unidirectional bus, and their combinations.>

Results of experimentation at the network testbed which emulates the heterogeneous traffic in an integrated manufacturing system are reported. As the network traffic increases, the dynamic performance of the delayed control system which regulates the servomotor in the experimental facility degrades. The experimental device is representative of the drive mechanisms of machine tools which are affected by network-induced delays in a real-time manufacturing environment. The concept of transmitting real-time unacknowledged data and delayable acknowledged data packets is applicable to both manufacturing systems and to integrated voice and data communication networks.>

The paper presents the concept of a real-time seam tracking system for welding automation and the initial design for building a prototype. The ARTIST - Adaptive Real-Time Intelligent Seam Tracker - uses a robot-held, laser based vision system for automation of arc welding processes, and is currently under development at the Applied Research Laboratory of The Pennsylvania State University. The design of ARTIST builds upon the concept of a zero-pass technique where 3D information of the seam geometry is collected and processed for real time guidance and control of the welding torch trailing behind the laser-based vision sensor. This zero-pass concept eliminates the need for pre-programming of the weld path and thus potentially enhances the welding cycle time for small batches. The ARTIST is designed to support multipass arc welding and to handle any tack welds which are encountered during the seam welding operation.