Abstract: Support vector machines (SVMs) have shown superior performance compared to other machine learning techniques, especially in classification problems. Yet one limitation of SVMs is the lack of an explanation capability which is crucial in some applications, e.g. in the medical and security domains. In this paper, a novel approach for eclectic rule- extraction from support vector machines is presented. This approach utilizes the knowledge acquired by the SVM and represented in its support vectors as well as the parameters associated with them. The approach includes three stages; training, propositional rule- extraction and rule quality evaluation. Results from four different experiments have demonstrated the value of the approach for extracting comprehensible rules of high accuracy and fidelity.

Abstract: The first 24 hours after a natural or man-made disaster are the most critical for the survival of victims. Unfortunately, this is also the time period when the fewest resources are available to rescuers. This paper describes the potential for using swarms of autonomous mobile robots to help the first responders to a disaster site focus their search for victims on those areas with the highest probability of finding survivors. Specifically, the paper starts with an overview of the current state-of-the-art in rescue robots, both autonomous and teleoperated; proposes a scenario for deploying autonomous resale robot swarms at a disaster site; summarizes work that has been done at Utah State University in developing autonomous rescue robot swarms; identifies some challenges for moving these rescue swarms out of artificial environments, like the RoboCup competition, into the real world; and, finally, makes some suggestions for future research in this area

Abstract: Automatic detection and classification of cardiac arrhythmias from a limited number of ECG signals is of considerable importance in critical care or operating room patient monitoring. We propose a method to accurately classify the heartbeat of ECG signals through the artificial neural networks (ANN). Feature sets are based on Heartbeat intervals, RR intervals and Spectral entropy of the ECG signal. The ability of properly trained artificial neural networks to correctly classify and recognize patterns makes them particularly suitable for use in an expert system that aids in the interpretation of ECG signals. In the present work the ECG data is taken from standard MIT-BIH arrhythmia database. The proposed method is capable of distinguishing the normal beat and 9 different arrhythmias. The overall accuracy of classification of the proposed approach is 99.02%. The results of the analysis are found to be more accurate than the other existing methods. Detection and classification of cardiac signals is important for diagnosis of cardiac abnormalities and hence any automated processing of the ECG that assists this process would be of assistance and is the focus of this paper.

Abstract: E-science is about global collaboration in key areas of science such as cognitive science and brain science, and the next generation of infrastructure such as the Wisdom Web and Knowledge Grids As a case study, we investigate human multiperception mechanism by cooperatively using various psychological experiments, physiological measurements, and data mining techniques for developing artificial systems which match human ability in specific aspects In particular, we observe fMRI (functional magnetic resonance imaging) and EEG (electroencephalogram) brain activations from the viewpoint of peculiarity oriented mining and propose a way of peculiarity oriented mining for knowledge discovery in multiple human brain data Based on such experience and needs, we concentrate on the architectural aspect of a brain-informatics portal from the perspective of the Wisdom Web and Knowledge Grids We describe how to build a data-mining grid on the Wisdom Web for multiaspect human brain data analysis The proposed methodology attempts to change the perspective of cognitive scientists from a single type of experimental data analysis toward a holistic view at a long-term, global field of vision

Abstract: Wireless sensor networks (WSN) are designed to monitor physical phenomena. The main task of WSN is to perform event detection, tracking, and classification. So, compared with traditional ad-hoc networks, WSN is event-centric. Therefore, an important question in WSN is to detect events. In this paper, we present two methods to do event detection, one is double sliding window detection, and the other one is fuzzy logic approach. The accuracy of the results is established via sensor network testbed and simulations

Abstract: This paper presents a few new competitive approaches to particle swarm optimization (PSO) algorithm in terms of the global and local best values (GLbest-PSO) and the standard PSO along with three set of variants namely, inertia weight (IW), acceleration co-efficient (AC) and mutation operators in this paper. Standard PSO is designed with time varying inertia weight (TVIW) and either time varying AC (TVAC) or fixed AC (FAC) while GLbest-PSO comprises of global-average local best IW (GaLbestIW) with either global-local best AC (GLbestAC) or FAC. The performances of these two algorithms are improved considerably in solving an optimal control problem, by introducing the concept of mutation variants between particles in each generation. The presence of mutation operator sharpens the convergence and tunes to the best solution. In order to compare and verify the validity and effectiveness of the new approaches for PSO, several statistical analyses are carried out. The results clearly demonstrate the improved performances of the proposed PSOs over the standard PSOs.

Abstract: This paper describes the Adaptive Constraint Engine (ACE), an ambitious ongoing research project to support constraint programmers, both human and machine. The program begins with substantial knowledge about constraint satisfaction. The program harnesses a cognitively oriented architecture—FOr the Right Reasons (FORR) to manage search heuristics and to learn new ones. ACE can transfer what it learns on simple problems to solve more difficult ones, and can readily export its knowledge to ordinary constraint solvers. It currently serves both as a learner and as a test bed for the constraint community. Many large-scale, real-world problems are readily represented, solved, and understood as constraint satisfaction problems (CSPs). Constraint programming offers a wealth of good, general-purpose methods to solve problems in such fields as telecommunications, Internet commerce, electronics, bioinformatics, transportation, network management, supply chain management, and finance (Nudel 1983; Freuder and Mackworth 1992). As a result, organizations throughout the world already exploit CSP technology to solve difficult problems in design and configuration, planning and scheduling, and diagnosis and testing. Yet each new, large-scale CSP faces the same bottleneck: difficult constraint programming problems need people to “tune” a solver efficiently. Armed with hard-to-extract domain expertise, scarce human CSP experts must now select, combine, and refine the various techniques currently available for constraint satisfaction and optimization. CSP solution remains more art form than automated process, in part because the interactions among existing CSP methods are not well understood. There is increasing evidence to suggest that different classes of CSPs respond best to different heuristics (Borrett, Tsang, and Walsh 1996), but arriving at appropriate methods in practice is not a trivial cookbook exercise (Beck, Prosser, and Selensky 2003). At present, for each new, large-scale CSP, a constraint programmer must seek an effective program based upon the right method combination. The thesis of our work is that a program can learn to synthesize, from generic components, effective programs adapted to specific CSP problem classes. This paper reports on initial results with the Adaptive Constraint Engine (ACE) . We do not propose ACE as a substitute for any particular constraint-solving program, but as a colleague in research. ACE can support constraint programmers in their quest for method combination appropriate to a particular class of problems specified by the user. (Throughout this paper, we distinguish carefully between the programmer, who writes code, and the user, who merely submits experiments to ACE for execution using that code.) ACE can support a novice constraint programmer in the selection of heuristics. It can learn new, efficient heuristics, those that were previously unidentified by experts and can be readily used by them in other programming environments. ACE can learn heuristics for problem classes that do not succumb to the ordinary, off-the-shelf CSP approaches. Thus we do not pit the program against others, but show results that improve problem solving, provide insight, and/or export to other solvers.

Abstract: We consider the problem of generating dominating sets for applications in information communication and sensor network. Known algorithms for solving this problem consider number of nodes in the dominating set as the sole criteria. We report algorithms for generating dominating sets by considering diameter and interference as the additional factors. Experimental investigation shows that the proposed algorithms can generate diameter reduced and interference aware dominating sets without increasing the size of the solution.

Abstract: This paper describes Eureka, a problem-solving architecture that operates under strong constraints on its memory and processes. Most significantly, Eureka does not assume free access to its entire long-term memory. That is, failures in problem solving may arise not only from missing knowledge, but from the (possibly temporary) inability to retrieve appropriate existing knowledge from memory. Additionally, the architecture does not include systematic backtracking to recover from fruitless search paths. These constraints significantly impact Eureka's design. Humans are also subject to such constraints, but are able to overcome them to solve problems effectively. In Eureka's design, we have attempted to minimize the number of additional architectural commitments, while staying faithful to the memory constraints. Even under such minimal commitments, Eureka provides a qualitative account of the primary types of learning reported in the literature on human problem solving. Further commitments to the architecture would refine the details in the model, but the approach we have taken de-emphasizes highly detailed modeling to get at general root causes of the observed regularities. Making minimal additional commitments to Eureka's design strengthens the case that many regularities in human learning and problem solving are entailments of the need to handle imperfect memory.

Abstract: Over the years we have developed the Disciple theory, methodology, and family of tools for building knowledge-based agents. This approach consists of developing an agent shell that can be taught directly by a subject matter expert in a way that resembles how the expert would teach a human apprentice when solving problems in cooperation. This paper presents the most recent version of the Disciple approach and its implementation in the Disciple–RKF (rapid knowledge formation) system. Disciple–RKF is based on mixed-initiative problem solving, where the expert solves the more creative parts of the problem and the agent solves the more routine ones, integrated teaching and learning, where the agent helps the expert to teach it, by asking relevant questions, and the expert helps the agent to learn, by providing examples, hints, and explanations, and multistrategy learning, where the agent integrates multiple learning strategies, such as learning from examples, learning from explanations, and learning by analogy, to learn from the expert how to solve problems. Disciple–RKF has been applied to build learning and reasoning agents for military center of gravity analysis, which are used in several courses at the US Army War College.

Abstract: A great challenge in developing planning systems for practical applications is the difficulty of acquiring the domain information needed to guide such systems. This paper describes a way to learn some of that knowledge. More specifically, the following points are discussed. (1) We introduce a theoretical basis for formally defining algorithms that learn preconditions for Hierarchical Task Network (HTN) methods. (2) We describe Candidate Elimination Method Learner (CaMeL), a supervised, eager, and incremental learning process for preconditions of HTN methods. We state and prove theorems about CaMeL's soundness, completeness, and convergence properties. (3) We present empirical results about CaMeL's convergence under various conditions. Among other things, CaMeL converges the fastest on the preconditions of the HTN methods that are needed the most often. Thus CaMeL's output can be useful even before it has fully converged.

Abstract: Security for wireless sensor networks (WSNs) is becoming an increasingly important issue in recent years. However, current works on media access control (MAC) protocols in WSNs mainly concentrate on balancing the efficiency and fairness of common channel access. Without proper security mechanisms, WSNs will be confined to limited and controlled environments, as well as negate numerous promises they hold. In this paper, we propose a novel secure MAC protocol for WSNs. Our algorithm-FSMAC firstly design multiple indicators for intrusion detection according to the classification of popular DoS attacks on MAC layer of WSNs, and innovatively utilizes fuzzy logic theory to implement making decision on intrusion. Moreover, appropriate countermeasures are adopted to reduce the destruction of attacks basing on intrusion detection results. Simulation results are presented to verify the effectiveness of our approach in terms of possibility of successful detection, possibility of false detection, data packet successful transmission rate, average energy consumption and time of first node dead

Abstract: Analysis of the Fractal Dimension of the EEG appears to be a good approach for the computation of biomarkers for dementia. Several Fractal Dimension algorithms have been used in the EEG analysis of cognitive and sleep disorders. The aim of this paper is to find an accurate Fractal Dimension algorithm that can be applied to the EEG for computing reliable biomarkers, specifically, for the assessment of dementia. To achieve this, some of the common methods for estimating the Fractal Dimension of the EEG are reviewed and compared using serial EEG recordings of normal and subjects with dementia. Biomarkers computed from the Fractal Dimensions are assessed according to their ability to perform early detection, differential diagnosis of dementia and in identifying effects of channel variations in subjects with dementia. The initial findings have shown that not all Fractal Dimension algorithms are suitable for computation of EEG biomarkers for dementia. Typically, biomarkers obtained from the Zero Set and the Adapted Box algorithms have shown good discriminating power in the early detection and differential diagnosis of dementia. Two channels, namely P3 and PZ have also been singled out as the most affected channels in dementing subjects. This bodes well with recent neuroimaging findings which indicate that the posterior cortex is one of the main sites of cortical atrophy in early Alzheimer's Disease.

Abstract: The hierarchical fuzzy signatures structure is a novel concept that can be used to find the degree of similarity or dissimilarity of objects which contain complex structured data, for classification or decision making. Fuzzy signatures are vector valued fuzzy sets, where each vector component can be a further vector valued fuzzy set. Thus, it differs from sparse hierarchical fuzzy rule based systems. Medical and economic diagnoses are the obvious applications of the fuzzy signatures. In this report we present results of three experiments, which were carried out to find the applicability of different aggregation functions, the relationship between the fuzzy signature structure and aggregation functions, and applicability of the fuzzy signatures method for different real world problems. Also, a new method of aggregating fuzzy signatures using weights called the weighted aggregation method has been proposed. Experiments show that the weighted aggregation method provides better results for fuzzy signatures.

Abstract: Texture classification has long been an important research topic in image processing. Classification based on the wavelet transform has become very popular. Wavelets are very effective in representing objects with isolated point singularities, but failed to represent line singularities. Recently, a ridgelet transform which deals effectively with line singularities in 2-D is introduced. It allows representing edges and other singularities along lines in a more efficient way. In this paper, the issue of texture classification based on a ridgelet transform has been analyzed. Features are derived from sub-bands of the ridgelet decomposition and are used for classification for a data set containing 20 texture images. Experimental results show that this approach allows to obtain a high degree of success in classification.

Abstract: Texture and color are two very important attributes in image analysis. While the color information describes the first order image properties, the texture generally describes second order property of surfaces and scenes, measured over image intensities. The need to include color aspect in texture analysis is being felt increasingly. The important aspect is, the way in which the chromatic information is involved in the formation and description of a texture. This paper describes an algorithm for classifying color textures using wavelet transform. Wavelet transform is useful for extracting texture features of images. A set of features are derived and color texture classification is done for different combination of the features and for different color models. The results obtained are found to be satisfactory.

Abstract: In this paper we propose a distributed knowledge management framework for semantics and knowledge creation, population and reuse on the Grid. Its objective is to evolve the Grid towards the Semantic Grid with the ultimate purpose of facilitating problem solving in e-Science. The framework uses ontology as the conceptual backbone and adopts the service-oriented computing paradigm for information-level and knowledge-level computation. We further present a semantics-based approach to problem solving, which exploits the rich semantic information of grid resource descriptions for resource discovery, instantiation and composition. The framework and approach has been applied to a UK e-Science project - Grid Enabled Engineering Design Search and Optimisation in Engineering (GEODISE). An ontology-enabled Problem Solving Environment (PSE) has been developed in GEODISE to leverage the semantic content of GEODISE resources and the Semantic Grid infrastructure for engineering design. Implementation and initial experimental results are reported.

Abstract: This paper presents an FSM implementation method based on symbolic functional decomposition. This novel approach in multilevel logic synthesis of finite state machines targets FPGA architectures. Traditional methods are based on two steps: internal state encoding and then mapping the encoded state transition table into target architecture. In the case of FPGAs, functional decomposition is recognized as the most efficient method of implementing digital circuits. However none of the known state encoding algorithms can be considered as a good method to be used with functional decomposition. In this paper the concept of symbolic functional decomposition is applied to obtain a multilevel structure that is suitable for implementing in FPGA logic cells. The symbolic decomposition does not require separate encoding step. It accepts FSM description with symbolic states and performs decomposition introducing such a state encoding that guarantees the best solution known.

Abstract: Currently, researchers have proposed several sequential association-rule models for predicting the next HTTP request. These researches focus on using sequence and temporal constrains for pruning to improve prediction precision. In this paper, we provide a comparative study on different kinds of sequential association rules for Web document prediction. Firstly, we give algorithms on mining sequential association rules, which is based on different sequence and temporal constrains combination. Then, the performance of all such algorithms has been compared on a real Web log dataset. Based on the comparison, by the method of variance analysis, we explore the effect of sequence and temporal information on influencing the precision of prediction. We show that the sequence constrains, the temporal constrains and the interaction between these two constrains can affect the precision of prediction. Furthermore, temporal constrains can affect more than sequence constrains. These results show light on the future research on improving the precisions of prediction.

Abstract: In modern space physics research, digital imagers are widely utilised in studies of the near-Earth space environment. The physical process being directly observed is the aurora, and millions of auroral images are acquired annually. These data sets provide a wealth of opportunities for developing and testing content-based image retrieval (CBIR) techniques with the irregular natural shapes occurring in auroral displays. Our CBIR implementation with relevance feedback was used in searching for one rare auroral form (”North-South structure”) that is a manifestation of an important physical process of general interest to space physics researchers today. We finish with a brief discussion of important benefits of anticipated application of this technique to multi-terabyte multi-million auroral im age data sets.

Abstract: In medical image visualization and analysis, segmentation is an indispensable step in the processing of images. MR has become a particularly useful medical diagnostic tool for cases involving soft tissues, such as in brain imaging. The aim of our research is to develop an effective algorithm for the segmentation of the MRI images. This paper discusses the use and implementation of fuzzy C means clustering and genetic algorithm (GA) for an automatic segmentation of white matter (WM), gray matter (GM), cerebro spinal fluid (CSF), the extra cranial regions and the presence of tumor regions. The results were analyzed and compared with the reference "gold standard" obtained from radiologists.

Abstract: This paper presents data fusion of visual and thermal infrared (IR) images in discrete wavelet transform (DWT) domain for robust face recognition. A combined use of face images in visible and thermal IR spectra has demonstrated robustness in face recognition under illumination variations. In the proposed approach, different fusion rules are applied separately to the approximation and the details components of the level-2 DWT decomposition to produce illumination-invariant face images. In case eyeglasses are present in the face, thermal images fail to provide useful information around the eyes since glass blocks a large portion of thermal energy. The DWT-based fusion method preserves visual details of eyeglass regions useful for face recognition in the fused images. Experiment results demonstrate that the fusion method is effective in terms of both visual quality and the entropy compared to conventional fusion approaches

Abstract: This paper investigates a feature selection and classification technique for face recognition using genetic algorithms and artificial neural networks. The experiments using separate facial features and combined facial features have been conducted on a face image dataset which is extracted from FERET benchmark database and was used in our previous study. The experiments using just combined features have also been conducted on an extended version of this dataset. The new experiments have achieved much better recognition rate than some of the existing face recognition techniques and significantly improved our previously published results. A detailed comparative analysis of experimental results is included in this paper.

Abstract: There is a need to determine the modulation type of an incoming signal. This paper proposes the use of the wavelet transform (WT) for modulation identification of digital signals without requiring any priori knowledge. The identifier consists of inter-class and intra-class identification. Different features have been used for different signals to increase the number of identification types and improve the processing rate. The performance of correct identification is increased too.

Abstract: The next generation Web Intelligence (WI) aims at enabling users to go beyond the existing online information search and knowledge queries functionalities and to gain, from the Web,1 practical wisdom for problem solving. To support such a Wisdom Web, we envision that a grid-like computing infrastructure with intelligent service agencies is needed, where these agencies can interact, self-organize, learn, and evolve their course of actions, identities, and interrelationships for new knowledge creation, as well as scientific and social evolution. In this paper, we first provide an overview of recent development in WI and Semantic/Knowledge Grid. Then, the fundamental capabilities of the Wisdom Web as well as the conceptual architecture of an intelligent Grid for supporting it are described. Technical challenges for realizing Grid Intelligence are highlighted and the recent advancements in related research areas are reviewed. 1.1. Web Intelligence and Wisdom Web The Web has irrevocably revolutionized the world we live in. This impact is inevitable due to the facts that the Web connectivity rapidly increases and that the online information astronomically explodes. In order not only to live with such a change but also to benefit from the information infrastructure that the Web has empowered, we have witnessed the fast development as well as applications of many Web Intelligence (WI) techniques and technologies (Zhong, Liu, and Yao 2003), which cover: 1. Internet-level communication, infrastructure, and security protocols. The Web is regarded as a computer-networked system. WI techniques for this level include, for instance, Web data-prefetching systems built upon Web-surfing patterns to resolve the issue of Web latency. The intelligence of the Web prefetching comes from adaptive learning based on observations of user-surfing behavior. 2. Interface-level multimedia presentation standards. The Web is regarded as an interface for human‐Internet interaction. WI techniques for this level are used to develop the intelligent Web interfaces in which the capabilities of adaptive cross-language processing, personalized multimedia representation, and multimodel data processing are required. 3. Knowledge-level information processing and management tools. The Web is regarded as a distributed data/knowledge base. We need to develop semantic markup languages to represent the semantic contents of the Web available in machine-understandable formats for agent-based computing, such as searching, aggregation, classification, filtering, managing, mining, and discovery on the Web (Berners-Lee, Hendler, and Lassila 2001). 4. Application-level ubiquitous computing and social intelligence environments. The Web is regarded as a basis for establishing social networks that contain communities for establishing social networks that contain communities of people (or organizations or other social entities) connected by social relationships, such as friendship, coworking, or information exchange with common interests. They are Web-supported social networks or virtual communities. The study of WI concerns the important issues central to social

Abstract: Image enhancement is very important for increasing the sensitivity of screening luggage performance at airports. On the basis of 11 statistical measures of image viewability we propose a novel approach to optimizing the choice of image enhancement tools. We propose a neural network predictor that can be used for predicting, on a given test image, the best image enhancement algorithm for it. The network is trained using a number of image examples. The input to the neural network is a set of viewability measures and its output is the choice of enhancement algorithm for that image. On a number of test images we show that such a predictive system is highly capable in forecasting the correct choice of enhancement algorithms (as judged by human experts). We compare our predictive system against a baseline approach that uses a fixed enhancement algorithm for all batch test images, and find the proposed model to be substantially superior

Abstract: This paper evaluates the use of an artificial neural network within a stockmarket trading strategy. The neural network was previously developed by the same authors, and has been trained using fundamental, company specific data. This study sites the neural network within a trading context, and demonstrates it is capable of producing economically significant results after accounting for costs.

Abstract: The complexity of computing systems steadily increases and their future administration will soon exceed any human capabilities. A resort to this scenario are self-managing systems that administrate themselves according to highlevel policies established by an administrator. Thus systems configure, optimize, protect and heal themselves autonomously making administrative interferences unnecessary. Organic Computing (OC) keeps this track by drawing analogies from biological systems such as ant colonies or dissipative structures, both resulting in emergent behavior. Due to its characteristics agent technology is particularly suitable for an implementation of Organic Computing Systems (OCS). However for a widespread industrial application acceptable software standards are required for both system architecture and software engineering. Therefore we propose a multi-agent system architecture and an appropriate development process as a first step towards a software engineering methodology for OCS. The process is based on the Model Driven Architecture (MDA) by the OMG and the UML 2 standard as well.

Abstract: The concept of mobile computing and its associated content is a topic receiving growing attention. The computing paradigm has evolved since the days of the mainframe system of previous decades to the laptop and handheld devices of today. These new systems require purpose-built applications, as users have ever increasing expectations of computational capabilities while on the move. Today, it is expected that the utilities that are accessible on the desktop are available when mobile. The intention of this paper is to describe the implementation of a Web-based, multi-agent system specifically designed for users on the move, to stream multimedia content to a hand-held device.