Artificial Intelligence and Hybrid Systems

The use of artificial intelligence and hybrid systems has increased dramatically due to their ability in handling real world problems involving uncertainty, vagueness and high complexity. The development of these systems has attracted the interest of the Artificial Intelligence community and established as a promising field of research. In order to present the ideas and practices of the hybridization process of intelligent systems, in this book, we included some recent and interesting studies on this topic. Chapter 1 thoroughly analyzes the behaviors of perceptron, including solution space for every possible patterns and convergence trajectories toward solutions, as well as briefly discusses the behaviors of multiple perceptrons. In Chapter 2, accurate as well as efficient non-invasive computational intelligence approaches (different artificial neural network models and a Sugeno-type fuzzy logic inference system) are implemented for the classification of two-phase flow in boiling water reactors. Chapter 3 contains a nice compilation of indifference-zone selection procedures as they apply to reliability analysis; many of those results come from the authors' own research. Chapter 4 discusses the software project design with the implementation of product line concepts to meet the customers demands for the production of high quality software applications in shortest possible time, within low budget and using less number of resources. Chapter 5 proposes integration of deterministic and probabilistic for improvement of human reliability analysis. The operator action success criteria time windows needed for human reliability analysis were determined through deterministic safety analysis. Chapter 6 presents a scheme to produce Network Anomaly Detection models based on Evolutionary Computation. The models are Hidden Markov Models, produced automatically, with no human intervention. Chapter 7 proposes an integrated online system which ranks the relational data. A compressed data structure is introduced to encode the dominant relationship of the data. Efficient querying strategies and updating scheme are devised to facilitate the ranking process. Chapter 8 proposes a method to combine the output of dependency parsers trained with the same parser generator but with specifically selected corpora. Chapter 9 proposes a new optimization model for the unit commitment problem using a stochastic hybrid algorithm combining simulated annealing and evolutionary algorithm. Chapter 10 develops neural network (NN) models to predict 85th percentile speed for two-lane rural highways in Oklahoma. Several input parameters, namely physical characteristics of road, traffic parameters, pavement condition indices, and accident data were considered in developing the models. In Chapter 11, a comparative study between micro-controller and computer for various network types is given, and an implementation of the a Radial Basis Function (RBF) network on the low end and inexpensive micro-controller is proposed. Chapter 12 provides an improved and updated survey of the recent researches and patents which concern about statistical background modeling and subtraction. Chapter 13 offers an outline of the hybrid approach, the design methodology of the INVENTS system, as well as its structure and user interface. In Chapter 14, it presents and discusses applications of the artificial neural network approach for improving equivalent circuit based transistor noise models in terms of the accuracy and the range of validity. Chapter 15 presents a simple yet efficient approach to improve IR on the web by reusing the information contained in the past user queries as the expansion for the initial query.