Search

Wavelet image decomposition generates a hierarchical data structure to represent an image. Recently, a new class of image compression algorithms has been developed for exploiting dependencies between the hierarchical wavelet coefficients using zerotrees. This paper deals with a fuzzy inference filter for image entropy coding by choosing significant coefficients and zerotree roots in the higher frequency wavelet subbands. Moreover, an adaptive quantization is proposed to improve the coding performance. Evaluating with the standard images, the proposed approaches are comparable or superior to most state-of-the-art coders. Based on the fuzzy energy judgment, the proposed approaches can achieve an excellent performance on the combination applications of image compression and watermarking.

This paper proposes an intelligent control system called self-exploring-based intelligent control system (SEICS). The SEICS is comprised of three basic mechanisms, namely, controller, performance evaluator (PE), and adaptor. The controller is constructed by a fuzzy neural network (FNN) to carry out the control tasks. The PE is used to determine whether or not the controller's performance is satisfactory. The adaptor, comprised of two elements, action explorer (AE) and rule generator (RG), plays the main role in the system for generating new control behaviors in order to enhance the control performance. AE operates through a three-stage self-exploration process to explore new actions, which is realized by the multiobjective genetic algorithm (GA). The RG transforms control actions to fuzzy rules based on a numerical method. The application of the adaptor can make a control system more adaptive in various environments. A simulation of robotic path-planning is used to demonstrate the proposed model. The results show that the robot reaches the target point from the start point successfully in the lack-of-information and changeable environments.

In this research, a novel vehicle-borne system of measuring three-dimensional (3-D) urban data using single-row laser range scanners is proposed. Two single-row laser range scanners are mounted on the roof of a vehicle, doing horizontal and vertical profiling respectively. As the vehicle moves ahead, a horizontal and a vertical range profile of the surroundings are captured at each odometer trigger. The freedom of vehicle motion is reduced from six to three by assuming that the ground surface is flat and smooth so resulting in the vehicle moving on almost the same horizontal plane. Horizontal range profiles, which have an overwhelming overlay between successive ones, are registered to trace vehicle location and attitude. Vertical range profiles are aligned to the coordinate system of the horizontal one according to the physical geometry between the pair of laser range scanners, and subsequently to a global coordinate system to make up 3-D data. An experiment is conducted where 3-D data of a real urban scene is obtained by registering and integrating 2412 horizontal and vertical range profiles. Two ground truths are used in examination. They are the outputs of a GPS/INS/Odometer based positioning system and a 1:500 digital map of the testing site. Accuracy and efficiency of the method in measuring 3-D urban scene is demonstrated.

Many existing independent component analysis (ICA) approaches result in deteriorated performance in temporal source separation because they have not taken into consideration of the underlying temporal structure of sources. In this paper, we model temporal sources as a general multivariate auto-regressive (AR) process whereby an underlying multivariate AR process in observation space is obtained. In this dual AR modeling, the mixing process from temporal sources to observations is the same as the mixture from the nontemporal residuals of the source AR (SAR) process to that of the observation AR (OAR) process. We can therefore avoid the source temporal effects in performing ICA by learning the demixing system on the independently distributed OAR residuals rather than the time-correlated observations. Particularly, we implement this approach by modeling each source signal as a finite mixture of generalized autoregressive conditional heteroskedastic (GARCH) process. The adaptive algorithms are proposed to extract the OAR residuals appropriately online, together with learning the demixing system via a nontemporal ICA algorithm. The experiments have shown its superior performance on temporal source separation.

This paper explores our solution aiming to provide efficient retrieval of medical imaging. Depending on the user, the same image can be described through different views. In essence, an image can be described on the basis of either low-level properties, such as texture or color; contextual data, such as date of acquisition or author; or semantic content, such as real-world objects and relations. Our approach consists of providing a multispaced description model capable of integrating different facets (or views) of the medical image. Visual retrieval solutions are recommended and are the most appropriated for noncomputer-science users. However, current visual languages suffer from several problems, especially ambiguities generated by the user and/or the system, and imprecision at different levels of image description. In this paper, we expose our solution and demonstrate how spatial precision of medical image content and ambiguities can be resolved. An implementation called Medical Image Management System (MIMS) has been realized to prove our proposition. A set of tests has been deployed to validate our prototype.

This paper addresses the problem of generic object classification from three-dimensional depth or meshed data. First, surface patches are segmented on the basis of differential geometry and quadratic surface fitting. These are represented by a modified Gaussian image that includes the well-known shape index. Learning is an interactive process in which a human teacher indicates corresponding patches, but the formation of generic classes is unaided. Classification of unknown objects is based on the measurement of similarities between feature sets of the objects and the generic classes. The process is demonstrated on a group of three-dimensional (3-D) objects built from both CAD and laser-scanned depth data.

This paper proposes a kind of evolutionary parallel local search technique (EPLS) that integrates the reproduction mechanisms from evolutionary algorithms and simplex method. The major aim is to explore the tradeoff between exploration and exploitation for optimizing multimodal functions. It has been cost-efficiently reached by means of parallel local search using simplex method. In each generation, EPLS partitions the population into a group of subpopulations, each of which consists of several individuals with adjacent space locations. EPLS independently locates multiple local optima in these disjoint neighborhoods, thus to reduce the probability of losing the global optimum. The local search in a neighborhood speeds up the convergence rate of simplex method. Recombination, adaptive Gaussian mutation and selection are incorporated into EPLS to further enhance the ability of global exploration and exploitation. The experimental observations and the extensive comparisons show that EPLS remarkably outperforms the standard evolutionary algorithms (EA) and some hybrid ones for almost all the problems tested, thus justifying the rationality and the competitive potential of EPLS for optimizing multimodal functions, especially for those with very rugged and deceptive topological structures.

In many applications of C-means clustering, the given data set often contains noisy points. These noisy points will affect the resulting clusters, especially if they are far away from the data points. In this paper, we develop a pruning approach for robust C-means clustering. This approach identifies and prunes the outliers based on the sizes and shapes of the clusters so that the resulting clusters are least affected by the outliers. The pruning approach is general, and it can improve the robustness of many existing C-means clustering methods. In particular, we apply the pruning approach to improve the robustness of hard C-means clustering, fuzzy C-means clustering, and deterministic-annealing C-means clustering. As a result, we obtain three clustering algorithms that are the robust versions of the existing ones. In addition, we integrate the pruning approach with the fuzzy approach and the possibilistic approach to design two new algorithms for robust C-means clustering. The numerical results demonstrate that the pruning approach can achieve good robustness.

Manipulation of perceptions is a remarkable human capability in a wide variety of physical and mental tasks under fuzzy or uncertain surroundings. Possibilistic reasoning can be treated as a mechanism that mimics human inference mechanisms with uncertain information. Petri nets are a graphical and mathematical modeling tool with powerful modeling and analytical ability. The focus of this paper is on the integration of Petri nets with possibilistic reasoning to reap the benefits of both formalisms. This integration leads to a possibilistic Petri nets model (PPN) with the following features. A possibilistic token carries information to describe an object and its corresponding possibility and necessity measures. Possibilistic transitions are classified into four types: inference transitions, duplication transitions, aggregation transitions, and aggregation-duplication transitions. A reasoning algorithm, based on possibilistic Petri nets, is also presented to improve the efficiency of possibilistic reasoning and an example related to diagnosis of cracks in reinforced concrete structures is used to illustrate the proposed approach.

In this paper, a novel methodology called a reference model approach to stability analysis of neural networks is proposed. The core of the new approach is to study a neural network model with reference to other related models, so that different modeling approaches can be combinatively used and powerfully cross-fertilized. Focused on two representative neural network modeling approaches (the neuron state modeling approach and the local field modeling approach), we establish a rigorous theoretical basis on the feasibility and efficiency of the reference model approach. The new approach has been used to develop a series of new, generic stability theories for various neural network models. These results have been applied to several typical neural network systems including the Hopfield-type neural networks, the recurrent back-propagation neural networks, the BSB-type neural networks, the bound-constraints optimization neural networks, and the cellular neural networks. The results obtained unify, sharpen or generalize most of the existing stability assertions, and illustrate the feasibility and power of the new method.

In this paper, a novel learning scheme is proposed to speed up the learning process in cerebellar model articulation controllers (CMAC). In the conventional CMAC learning scheme, the correct numbers of errors are equally distributed into all addressed hypercubes, regardless of the credibility of the hypercubes. The proposed learning approach uses the inverse of learned times of the addressed hypercubes as the credibility (confidence) of the learned values, resulting in learning speed becoming very fast. To further demonstrate online learning capability of the proposed credit assigned CMAC learning scheme, this paper also presents a learning robust controller that can actually learn online. Based on robust controllers presented in the literature, the proposed online learning robust controller uses previous control input, current output acceleration, and current desired output as the state to define the nominal effective moment of the system from the CMAC table. An initial trial mechanism for the early learning stage is also proposed. With our proposed credit-assigned CMAC, the robust learning controller can accurately trace various trajectories online.

Benchmarking is an important phase in developing any new software technique because it helps to validate the underlying theory in the specific problem domain. But benchmarking of new software strategies is a very complex problem, because it is difficult (if not impossible) to test, validate and verify the results of the various schemes in completely different settings. This is even more true in the case of database systems because the benchmarking also depends on the types of queries presented to the databases used in the benchmarking experiments. Query optimization strategies in relational database systems rely on approximately estimating the query result sizes to minimize the response time for user-queries. Among the many query result size estimation techniques, the histogram-based techniques are by far the most commonly used ones in modern-day database systems. These techniques estimate the query result sizes by approximating the underlying data distributions, and, thus, are prone to estimation errors. In two recent works , we proposed (and thoroughly analyzed) two new forms of histogram-like techniques called the rectangular and trapezoidal attribute cardinality maps (ACM), respectively, that give much smaller estimation errors than the traditional equi-width and equi-depth histograms currently being used by many commercial database systems. This paper reports how the benchmarking of the Rectangular-ACM (R-ACM) and the Trapezoidal-ACM (T-ACM) for query optimization can be achieved. By conducting an extensive set of experiments using the acclaimed TPC-D benchmark queries and database , we demonstrate that these new ACM schemes are much more accurate than the traditional histograms for query result size estimation. Apart from demonstrating the power of the ACMs, this paper also shows how the TPC-D benchmarking can be achieved using a large synthetic database with many different patterns of synthetic queries, which are representative of a real-world business environment.

A key element of bioinformatics research is the extraction of meaningful information from large experimental data sets. Various approaches, including statistical and graph theoretical methods, data mining, and computational pattern recognition, have been applied to this task with varying degrees of success. Using a novel classifier based on the Bayes discriminant function, we present a hybrid algorithm that employs feature selection and extraction to isolate salient features from large medical and other biological data sets. We have previously shown that a genetic algorithm coupled with a k-nearest-neighbors classifier performs well in extracting information about protein-water binding from X-ray crystallographic protein structure data. The effectiveness of the hybrid EC-Bayes classifier is demonstrated to distinguish the features of this data set that are the most statistically relevant and to weight these features appropriately to aid in the prediction of solvation sites.

Bipolar logic, bipolar sets, and equilibrium relations are proposed for bipolar cognitive mapping and visualization in online analytical processing (OLAP) and online analytical mining (OLAM). As cognitive models, cognitive maps (CMs) hold great potential for clustering and visualization. Due to the lack of a formal mathematical basis, however, CM-based OLAP and OLAM have not gained popularity. Compared with existing approaches, bipolar cognitive mapping has a number of advantages. First, bipolar CMs are formal logical models as well as cognitive models. Second, equilibrium relations (with polarized reflexivity, symmetry, and transitivity), as bipolar generalizations and fusions of equivalence relations, provide a theoretical basis for bipolar visualization and coordination. Third, an equilibrium relation or CM induces bipolar partitions that distinguish disjoint coalition subsets not involved in any conflict, disjoint coalition subsets involved in a conflict, disjoint conflict subsets, and disjoint harmony subsets. Finally, equilibrium energy analysis leads to harmony and stability measures for strategic decision and multiagent coordination. Thus, this work bridges a gap for CM-based clustering and visualization in OLAP and OLAM. Basic ideas are illustrated with example CMs in international relations.

This paper is an improvement on the characterization of edges. Using a novel wavelet function, it is proven that the maximum moduli of the wavelet transform (MMWT) of a curve produces two new symmetrical curves on both sides of the original with the same direction. The distance between the two curves is shown to be independent of the width d of the original curve if the scale s of the wavelet transform satisfies s/spl ges/d. This property provides a novel method of obtaining the skeletons of the curves in an image.

Recently, many virtual reality and robotics applications have been called on to create virtual environments from real scenes. A catadioptric omnidirectional image sensor composed of a convex mirror can simultaneously observe a 360-degree field of view making it useful for modeling man-made environments such as rooms, corridors, and buildings, because any landmarks around the sensor can be taken in and tracked in its large field of view. However, the angular resolution of the omnidirectional image is low because of the large field of view captured. Hence, the resolution of surface texture patterns on the three-dimensional (3-D) scene model generated is not sufficient for monitoring details. To overcome this, we propose a high resolution scene texture generation method that combines an omnidirectional image sequence using image mosaic and superresolution techniques.

Under the framework of the Kullback-Leibler (KL) distance, we show that a particular case of Gaussian probability function for feedforward neural networks (NNs) reduces into the first-order Tikhonov regularizer. The smooth parameter in kernel density estimation plays the role of regularization parameter. Under some approximations, an estimation formula is derived for estimating regularization parameters based on training data sets. The similarity and difference of the obtained results are compared with other work. Experimental results show that the estimation formula works well in sparse and small training sample cases.

As a typical combinatorial optimization problem, the traveling salesman problem (TSP) has attracted extensive research interest. In this paper, we develop a self-organizing map (SOM) with a novel learning rule. It is called the integrated SOM (ISOM) since its learning rule integrates the three learning mechanisms in the SOM literature. Within a single learning step, the excited neuron is first dragged toward the input city, then pushed to the convex hull of the TSP, and finally drawn toward the middle point of its two neighboring neurons. A genetic algorithm is successfully specified to determine the elaborate coordination among the three learning mechanisms as well as the suitable parameter setting. The evolved ISOM (eISOM) is examined on three sets of TSP to demonstrate its power and efficiency. The computation complexity of the eISOM is quadratic, which is comparable to other SOM-like neural networks. Moreover, the eISOM can generate more accurate solutions than several typical approaches for TSP including the SOM developed by Budinich, the expanding SOM, the convex elastic net, and the FLEXMAP algorithm. Though its solution accuracy is not yet comparable to some sophisticated heuristics, the eISOM is one of the most accurate neural networks for the TSP.

The granularity and interpretability of a fuzzy model are influenced by the method used to construct the rule base. Models obtained by a heuristic assessment of the underlying system are generally highly granular with interpretable rules, while models algorithmically generated from an analysis of training data consist of a large number of rules with small granularity. This paper presents a method for increasing the granularity of rules while satisfying a prescribed precision bound on the training data. The model is generated by a two-stage process. The first step iteratively refines the partitions of the input domains until a rule base is generated that satisfies the precision bound. In this step, the antecedents of the rules are obtained from decomposable partitions of the input domains and the consequents are generated using proximity techniques. A greedy merging algorithm is then applied to increase the granularity of the rules while preserving the precision bound. To enhance the representational capabilities of a rule and reduce the number of rules required, the rules constructed by the merging procedure have multi-dimensional antecedents. A model defined with rules of this form incorporates advantageous features of both clustering and proximity methods for rule generation. Experimental results demonstrate the ability of the algorithm to reduce the number of rules in a fuzzy model with both precise and imprecise training information.

Two novel systems computing dense three-dimensional (3-D) scene flow and structure from multiview image sequences are described in this paper. We do not assume rigidity of the scene motion, thus allowing for nonrigid motion in the scene. The first system, integrated model-based system (IMS), assumes that each small local image region is undergoing 3-D affine motion. Non-linear motion model fitting based on both optical flow constraints and stereo constraints is then carried out on each local region in order to simultaneously estimate 3-D motion correspondences and structure. The second system is based on extended gradient-based system (EGS), a natural extension of two-dimensional (2-D) optical flow computation. In this method, a new hierarchical rule-based stereo matching algorithm is first developed to estimate the initial disparity map. Different available constraints under a multiview camera setup are further investigated and utilized in the proposed motion estimation. We use image segmentation information to adopt and maintain the motion and depth discontinuities. Within the framework for EGS, we present two different formulations for 3-D scene flow and structure computation. One formulation assumes that initial disparity map is accurate, while the other does not. Experimental results on both synthetic and real imagery demonstrate the effectiveness of our 3-D motion and structure recovery schemes. Empirical comparison between IMS and EGS is also reported.

Nowadays volume images are frequently used in many applications. Volume images of paper can be analyzed to increase the understanding of the complexity of the fiber network in paper and its effect on the optical and mechanical properties of paper. We show how curve and surface representations of the fiber, the fiber wall, and the fiber lumen can be computed using distance transform based algorithms. These representations are used for easy computation of wall thickness, degree of collapse, fiber length, slenderness ratio, fiber curl, and torsional resistance for the fibers. Free-fiber segments are identified, again using a distance transform based algorithm. Finally, we show tools for qualitative visual inspection of the fibers. The methods are evaluated and illustrated using both sets of synthetic and of real data. In addition to the analysis of the fiber network in paper, they can be used in many other applications where shape analysis of elongated objects is to be performed.

This paper describes an approach based on evolutionary algorithms, hierarchical decision rules (HIDER), for learning rules in continuous and discrete domains. The algorithm produces a hierarchical set of rules, that is, the rules are sequentially obtained and must therefore be tried until one is found whose conditions are satisfied. Thus, the number of rules may be reduced because the rules could be inside of one another. The evolutionary algorithm uses both real and binary coding for the individuals of the population. We tested our system on real data from the UCI repository, and the results of a ten-fold cross-validation are compared to C4.5s, C4.5Rules, See5s, and See5Rules. The experiments show that HIDER works well in practice.

Zou and Yan have recently developed a skeletonization algorithm of digital shapes based on a regularity/singularity analysis; they use the polygon whose vertices are the boundary pixels of the image to compute a constrained Delaunay triangulation (CDT) in order to find local symmetries and stable regions. Their method has produced good results but it is slow since its complexity depends on the number of contour pixels. This paper presents an extension of their technique to handle arbitrary polygons, not only polygons of short edges. Consequently, not only can we achieve results as good as theirs for digital images, but we can also compute skeletons of polygons of any number of edges. Since we can handle polygonal approximations of figures, the skeletons are more resilient to noise and faster to process.

A divide and conquer deformable contour method is presented with an initial inside closed contour being divided into arbitrary segments, and these segments are allowed to deform separately preserving the segments' connectivity. A maximum area threshold, A/sub max/, is used to stop these outward contour segments' marching. Clear and blur contour points are then identified to partition the whole contour into clear and blur segments. A bi-directional searching method is then recursively applied to each blur segment including a search for contour-within-contour segment to reach a final close contour. Further improvements are provided by a model based searching algorithm. It is a two-step process with step 1 being a linked contour model matching operation where landmarks are extracted, and step 2 being a posteriori probability model matching and correction operation where large error segments are fine tuned to obtain the final results. The experiments include ultrasound images of pig heart, MRI brain images, MRI knee images having complex shapes with or without gaps, and inhomogeneous interior and contour region brightness distributions. These experiments have shown that the method has the capability of moving a contour into the neighboring region of the desired boundary by overcoming inhomogeneous interior, and by adapting each contour segment searching operation to different local difficulties, through a contour partition and repartition scheme in searching for a final solution.

The paper proposes a network routing method based on a computational ecology model. The computational ecology model is a mathematical model proposed by B.A. Huberman and T. Hogg (1988), which represents a macro action of multi-agent systems. We formulate routing on a computer network as a resource allocation problem, where packets and links are regarded as agents and resources, respectively. Then, we apply an extended computational ecology model for this problem. Agents conflict so as to get more payoffs from links. As a result, they get the same payoffs, and a good resource allocation is achieved. In each node, each packet selects a link according to the selection rate decided through conflicts, and routing is accomplished autonomously with adaptability on the computer network. Moreover, we improve fault-tolerance of the system by local information exchanges. Finally, we examine the efficiency of the proposed method by computer simulation.

Segmentation of an image into regions and the labeling of the regions is a challenging problem. In this paper, an approach that is applicable to any set of multifeature images of the same location is derived. Our approach applies to, for example, medical images of a region of the body; repeated camera images of the same area; and satellite images of a region. The segmentation and labeling approach described here uses a set of training images and domain knowledge to produce an image segmentation system that can be used without change on images of the same region collected over time. How to obtain training images, integrate domain knowledge, and utilize learning to segment and label images of the same region taken under any condition for which a training image exists is detailed. It is shown that clustering in conjunction with image processing techniques utilizing an iterative approach can effectively identify objects of interest in images. The segmentation and labeling approach described here is applied to color camera images and two other image domains are used to illustrate the applicability of the approach.

The problem of automated defect detection in textured materials is investigated. A new approach for defect detection using linear FIR filters with optimized energy separation is proposed. The performance of different feature separation criteria with reference to fabric defects has been evaluated. The issues relating to the design of optimal filters for supervised and unsupervised web inspection are addressed. A general web inspection system based on the optimal filters is proposed. The experiments on this new approach have yielded excellent results. The low computational requirement confirms the usefulness of the approach for industrial inspection.

We describe a complete methodology for the validation of rule-based expert systems. This methodology is presented as a five-step process that has two central themes: 1) to create a minimal set of test inputs that adequately cover the domain represented in the knowledge base; and 2) a Turing Test-like methodology that evaluates the system's responses to the test inputs and compares them to the responses of human experts. The development of minimal set of test inputs takes into consideration various criteria, both user-defined, and domain-specific. These criteria are used to reduce the potentially very large set of test inputs to one that is practical, keeping in mind the nature and purpose of the developed system. The Turing Test-like evaluation methodology makes use of only one panel of experts to both evaluate each set of test cases and compare the results with those of the expert system, as well as with those of the other experts. The hypothesis being presented is that much can be learned about the experts themselves by having them anonymously evaluate each other's responses to the same test inputs. Thus, we are better able to determine the validity of an expert system. Depending on its purpose, we introduce various ways to express validity as well as a technique to use the validity assessment for the refinement of the rule base. Lastly, we describe a partial implementation of the test input minimalization process on a small but nontrivial expert system. The effectiveness of the technique was evaluated by seeding errors into the expert system, generating the appropriate set of test inputs and determining whether the errors could be detected by the suggested methodology.

Page-based linear genetic programming (GP) is proposed in which individuals are described in terms of a number of pages. Pages are expressed in terms of a fixed number of instructions, which is constant for all individuals in the population. Pairwise crossover results in the swapping of single pages, and thus, individuals are of a fixed number of instructions. Head-to-head comparison with Tree-structured GP and block-based linear GP indicates that the page-based approach evolves succinct solutions without penalizing generalization ability.

An extended algorithm of the relative reward strength algorithm is proposed. It is shown that the proposed algorithm ensures the convergence with probability I to the optimal path under the certain type of nonstationary environment. Several computer simulation results confirm the effectiveness of the proposed algorithm.

Although new high-bandwidth network technologies are being introduced and widely deployed, asynchronous transfer mode (ATM) is still considered one of the most important network technologies currently in use. A number of ATM switch architectures have been proposed in the literature. However, industry has shown that is better to use the well-known shared-medium technique in the design of these ATM switches. In this paper, four variations of a new distributed scheme are proposed for the arbitration of a shared bus of an ATM switch. These schemes are based on learning automata. By taking advantage of the bursty nature of ATM traffic, the new arbitration scheme demonstrates superb performance compared to the time division multiple access (TDMA) scheme.

This paper presents a new methodology for efficiently representing the content of images and comparing images by detecting and recording their visual differences. In particular, the methodology presented here is based on a stochastic Petri-net (SPN) graph approach able to extract and record local and global features from both images, compare them, and define the percentage of similarity. One of the features of the human visual perception is the detection of similarities between two images. The visual similarity is based on color, size, shape, and local and global topological changes of the image regions. Several methods dealing with image or object similarities have been proposed. The new feature of the method here is the partial emulation of the human observer's visual perception by recording differences extracted from different images. Results of the method described here are presented for a variety of images by using local and global noisy conditions.

We present a discretized learning automaton (LA) solution to the capacity assignment (CA) problem which focuses on finding the best possible set of capacities for the links that satisfy the traffic requirements in a prioritized network while minimizing the cost. Most approaches consider a single class of packets flowing through the network, but in reality, different classes of packets with different average packet lengths and different priorities are transmitted over the networks. This generalized model is the focus of this paper. Although the problem is inherently NP-hard, a few approximate solutions have been proposed in the literature. Marayuma and Tang (1977) proposed a single algorithm composed of several elementary heuristic procedures. Other solutions tackle the problem by using modern-day artificial intelligence (AI) paradigms such as simulated annealing and genetic algorithms (GAs). In 2000, we introduced a new method, superior to these, that uses continuous LA. In this paper, we present a discretized LA solution to the problem. This solution uses a meta-action philosophy new to the field of LA, and is probably the best available solution to this extremely complex problem.

The authors present a method for decomposition of Bayesian networks into their maximal prime subgraphs. The correctness of the method is proven and results relating the maximal prime subgraph decomposition (MPD) to the maximal complete subgraphs of the moral graph of the original Bayesian network are presented. The maximal prime subgraphs of a Bayesian network can be organized as a tree which can be used as the computational structure for LAZY propagation. We also identify a number of tasks performed on Bayesian networks that can benefit from MPD. These tasks are: divide and conquer triangulation, hybrid propagation algorithms combining exact and approximative inference techniques, and incremental construction of junction trees. We compare the proposed algorithm with standard algorithms for decomposition of undirected graphs into their maximal prime subgraphs. The discussion shows that the proposed algorithm is simpler, more easy to comprehend, and it has the same complexity as the standard algorithms.

Scene analysis is a major aspect of perception and continues to challenge machine perception. This paper addresses the scene-analysis problem by integrating a primitive segmentation stage with a model of associative memory. The model is a multistage system that consists of an initial primitive segmentation stage, a multimodule associative memory, and a short-term memory (STM) layer. Primitive segmentation is performed by a locally excitatory globally inhibitory oscillator network (LEGION), which segments the input scene into multiple parts that correspond to groups of synchronous oscillations. Each segment triggers memory recall and multiple recalled patterns then interact with one another in the STM layer. The STM layer projects to the LEGION network, giving rise to memory-based grouping and segmentation. The system achieves scene analysis entirely in phase space, which provides a unifying mechanism for both bottom-up analysis and top-down analysis. The model is evaluated with a systematic set of three-dimensional (3-D) line drawing objects, which are arranged in an arbitrary fashion to compose input scenes that allow object occlusion. Memory-based organization is responsible for a significant improvement in performance. A number of issues are discussed, including input-anchored alignment, top-down organization, and the role of STM in producing context sensitivity of memory recall.

The fastest learning automata (LA) algorithms currently available fall in the family of estimator algorithms introduced by Thathachar and Sastry (1986). The pioneering work of these authors was the pursuit algorithm, which pursues only the current estimated optimal action. If this action is not the one with the minimum penalty probability, this algorithm pursues a wrong action. In this paper, we argue that a pursuit scheme that generalizes the traditional pursuit algorithm by pursuing all the actions with higher reward estimates than the chosen action, minimizes the probability of pursuing a wrong action, and is a faster converging scheme. To attest this, we present two new generalized pursuit algorithms (GPAs) and also present a quantitative comparison of their performance against the existing pursuit algorithms. Empirically, the algorithms proposed here are among the fastest reported LA to date.

The paper describes a novel design of a robot tactile sensor, called hybrid tactile sensor, based on analog resistive technology. The design employs a set of parallel analog resistive sensing strips, rather than a two-dimensional (2D) grid of sensing elements, to sample the sensor surface. It is therefore analog in one dimension and discrete in the other. As a result, the number of samples that need to be obtained is significantly reduced, as is the number of connectors in the sensor circuitry. In addition, the sensor design scales easily to a large sensing surface area, and offers flexibility in sensor geometry. However, only the contact location and shape can be sensed, but not the contact force. The paper provides the theoretical model of the sensor, and presents simulation and experimental results.

This paper introduces a new learning methodology to quickly generate accurate and simple linguistic fuzzy models: the cooperative rules (COR) methodology. It acts on the consequents of the fuzzy rules to find those that are best cooperating. Instead of selecting the consequent with the highest performance in each fuzzy input subspace, as ad-hoc data-driven methods usually do, the COR methodology considers the possibility of using another consequent, different from the best one, when it allows the fuzzy model to be more accurate thanks to having a rule set with the best cooperation. Our proposal has shown good results in solving three different applications when compared to other methods.

In this paper, we report our study on embedding fuzzy mechanisms and knowledge into box-type reinforcement learning controllers. One previous approach for incorporating fuzzy mechanisms can only achieve one successful run out of nine tests compared to eight successful runs in a nonfuzzy learning control scheme. After analysis, the credit assignment problem and the weighting domination problem are identified. Furthermore, the use of fuzzy mechanisms in temporal difference seems to play a negative factor. Modifications to overcome those problems are proposed. Furthermore, several remedies are employed in that approach. The effects of those remedies applied to our learning scheme are presented and possible variations are also studied. Finally, the issue of incorporating knowledge into reinforcement learning systems is studied. From our simulations, it is concluded that the use of knowledge for the control network can provide good learning results, but the use of knowledge for the evaluation network alone seems unable to provide any significant advantages. Furthermore, we also employ Makarovic's (1988) rules as the knowledge for the initial setting of the control network. In our study, the rules are separated into four groups to avoid the ordering problem.

In this paper, a neural tree-based approach for classifying range images into a set of nonoverlapping regions is presented. An innovative procedure is applied to extract invariant surface features from each pixel of the range image. These features are: 1) robust to noise, and 2) invariant to scale, shift, rotations, curvature variations, and direction of the normal. Then, a generalized neural tree is used to classify each image point as belonging to one of the six surface models of differential geometry, i.e., peak, ridge, valley, saddle, pit, and flat. Comparisons with other methods and experiments on both synthetic and real three-dimensional range images are proposed.

This paper describes a novel knowledge discovery and data mining framework dealing with nonlinear interactions among domain attributes. Our network-based model provides an effective and efficient reasoning procedure to perform prediction and decision making. Unlike many existing paradigms based on linear models, the attribute relationship in our framework is represented by nonlinear nonnegative multiregressions based on the Choquet integral. This kind of multiregression is able to model a rich set of nonlinear interactions directly. Our framework involves two layers. The outer layer is a network structure consisting of network elements as its components, while the inner layer is concerned with a particular network element modeled by Choquet integrals. We develop a fast double optimization algorithm (FDOA) for learning the multiregression coefficients of a single network element. Using this local learning component and multiregression-residual-cost evolutionary programming (MRCEP), we propose a global learning algorithm, called MRCEP-FDOA, for discovering the network structures and their elements from databases. We have conducted a series of experiments to assess the effectiveness of our algorithm and investigate the performance under different parameter combinations, as well as sizes of the training data sets. The empirical results demonstrate that our framework can successfully discover the target network structure and the regression coefficients.

By exploiting the Fourier series expansion, we have developed a new constructive method of automatically generating a multivariable fuzzy inference system from any given sample set with the resulting multivariable function being constructed within any specified precision to the original sample set. The given sample sets are first decomposed into a cluster of simpler sample sets such that a single input fuzzy system is constructed readily for a sample set extracted directly from the cluster independent of the other variables. Once the relevant fuzzy rules and membership functions are constructed for each of the variables completely independent of the other variables, the resulting decomposed fuzzy rules and membership functions are integrated back into the fuzzy system appropriate for the original sample set requiring only a moderate cost of computation in the required decomposition and composition processes. After proving two basic theorems which we need to ensure the validity of the decomposition and composition processes of the system construction, we have demonstrated a constructive algorithm of a multivariable fuzzy system. Exploiting an implicit error bound analysis available at each of the construction steps, the present Fourier method is capable of implementing a more stable fuzzy system than the power series expansion method of ParNeuFuz and PolyNeuFuz, covering and implementing a wider range of more robust applications.

Fuzzy rule bases have proven to be an effective tool for modeling complex systems and approximating functions. The generation of a fuzzy rule base has generally been accomplished by a heuristic analysis of the relationships of the underlying system or by algorithmic rule generation from training data. Automatic rule generation has utilized clustering algorithms, proximity analysis, and evolutionary techniques to identify approximate relationships between the input and the output. In this research, two general approaches for the evolutionary generation of fuzzy rules are identified and compared: global and local rule generation. Global rule production, which is the standard method of employing evolutionary techniques in fuzzy rule base generation, considers an entire rule base as an element of population. The fitness evaluation of a rule base aggregates the performance of the model over the entire space into a single value. The local approach utilizes the limited scope of a fuzzy rule to evaluate performance in regions of the input space. The local generation of rule bases employs an independent evolutionary search in each region and combines the local results to produce a global model. An experimental suite has been developed to compare the effectiveness of the two strategies for the evolutionary generation of fuzzy models.

The paper presents a new genetic algorithm (GA)-based discrete dynamic programming (DDP) approach for generating static schedules in a flexible manufacturing system (FMS) environment. This GA-DDP approach adopts a sequence-dependent schedule generation strategy, where a GA is employed to generate feasible job sequences and a series of discrete dynamic programs are constructed to generate legal schedules for a given sequence of jobs. In formulating the GA, different performance criteria could be easily included. The developed DDF algorithm is capable of identifying locally optimized partial schedules and shares the computation efficiency of dynamic programming. The algorithm is designed In such a way that it does not suffer from the state explosion problem inherent in pure dynamic programming approaches in FMS scheduling. Numerical examples are reported to illustrate the approach.

A learning automaton (LA) is an automaton that interacts with a random environment, having as its goal the task of learning the optimal action based on its acquired experience. Many learning automata (LAs) have been proposed, with the class of estimator algorithms being among the fastest ones, Thathachar and Sastry, through the pursuit algorithm, introduced the concept of learning algorithms that pursue the current optimal action, following a reward-penalty learning philosophy. Later, Oommen and Lanctot extended the pursuit algorithm into the discretized world by presenting the discretized pursuit algorithm, based on a reward-inaction learning philosophy. In this paper we argue that the reward-penalty and reward-inaction learning paradigms in conjunction with the continuous and discrete models of computation, lead to four versions of pursuit learning automata. We contend that a scheme that merges the pursuit concept with the most recent response of the environment, permits the algorithm to utilize the LAs long-term and short-term perspectives of the environment. In this paper, we present all four resultant pursuit algorithms, prove the E-optimality of the newly introduced algorithms, and present a quantitative comparison between them.

This paper presents an evaluation of edge detector performance. We use the task of structure from motion (SFM) as a "black box" through which to evaluate the performance of edge detection algorithms. Edge detector goodness is measured by how accurately the SFM could recover the known structure and motion from the edge detection of the image sequences. We use a variety of real image sequences with ground truth to evaluate eight different edge detectors from the literature. Our results suggest that ratings of edge detector performance based on pixel-level metrics and on the SFM are well correlated and that detectors such as the Canny detector and Heitger detector offer the best performance.

The decomposition of high-dimensional multivariate time series (MTS) into a number of low-dimensional MTS is a useful but challenging task because the number of possible dependencies between variables is likely to be huge. This paper is about a systematic study of the "variable groupings" problem in MTS. In particular, we investigate different methods of utilizing the information regarding correlations among MTS variables. This type of method does not appear to have been studied before. In all, 15 methods are suggested and applied to six datasets where there are identifiable mixed groupings of MTS variables. This paper describes the general methodology, reports extensive experimental results, and concludes with useful insights on the strength and weakness of this type of grouping method.

An attributed graph (AG) is a useful data structure for representing complex patterns in a wide range of applications such as computer vision, image database retrieval, and other knowledge representation tasks where similar or exact corresponding structural patterns must be found. Existing methods for attributed graph matching (AGM) often suffer from the combinatorial problem whereby the execution cost for finding an exact or similar match is exponentially related to the number of nodes the AG contains. The square matching error of two AGs subject to permutations is approximately relaxed to a square matching error of two AGs subject to orthogonal transformations. Hence, the principal component analysis (PCA) algorithm can be used for the fast computation of the approximate matching error, with a considerably reduced execution complexity. Experiments demonstrate that this method works well and is robust against noise and other simple types of transformations.

In this paper, we present a class of test feature classifiers (TFCs). We discuss the properties and performance of the proposed classifiers and describe cases when a 100% recognition rate on test data can be achieved. When the number of features increases, classes having no more than a polynomial number of instances (in the number of features) are the only cases possible to process. We prove that for almost all pairs of classes with a polynomial number of instances, a 100% recognition rate on any test data can be achieved. To test the performance of the classifiers, we apply them to both artificial and real data. For the real data, we use the well known breast cancer, phoneme, and satimage databases, which are recognized to be difficult classification problems. Our experimental results show that the proposed classifiers not only have a high recognition ability but, also, the ability to achieve a 100% recognition rate in difficult classification problems.

In this paper, the advantages of a fuzzy representation in problem solving and search is investigated using the framework of Cultural algorithms (CAs). Since all natural languages contain a fuzzy component, the natural question is "Does this fuzzy representation facilitate the problem-solving process, within these systems". In order to investigate this question we use the CA framework of Reynolds (1996), CAs are a computational model of cultural evolution derived from and used to express basic anthropological models of culture and its development. A mathematical model of a full fuzzy CA is developed there. In it, the problem solving knowledge is represented using a fuzzy framework. Several theoretical results concerning its properties are presented. The model is then applied to the solution of a set of 12 difficult, benchmark problems in nonlinear real-valued function optimization. The performance of the full fuzzy model is compared with 8 other fuzzy and crisp architectures. The results suggest that a fuzzy approach can produce a statistically significant improvement in search efficiency over nonfuzzy versions for the entire set of functions, the then investigate the class of performance functions for which the full fuzzy system exhibits the greatest improvements over nonfuzzy systems. In general, these are functions which require some preliminary investigation in order to embark on an effective search.