Incremental Evolution in ANNs: Neural Nets which Grow

This paper explains the optimisation of neuralnetwork topology using Incremental Evolution;that is, by allowing the network to expand byadding to its structure. This method allows anetwork to grow from a simple to a complexstructure until it is capable of fulfilling itsintended function. The approach is somewhatanalogous to the growth of an embryo or theevolution of a fossil line through time, it istherefore sometimes referred to as anembryology or embryological algorithm. Thepaper begins with a general introduction,comparing this method to other competingtechniques such as The Genetic Algorithm, otherEvolutionary Algorithms and SimulatedAnnealing. A literature survey of previous workis included, followed by an extensive newframework for application of the technique.Finally, examples of applications and a generaldiscussion are presented.     

Incorporating Software Visualization in the Design of Intelligent Diagnosis Systems for User Programming

Program diagnosis systems were developed to help users solve programming problems. By providing guidence on errors and misconceptions, these systems can help the users in writing programs and understanding their dynamic behavior. Features of software visualization which aim at providing visual and concrete depictions to the abstractions and operations of programs have also shown to be making programs more understandable. The main theme of this paper is to asses the usefulness of incorporating features of software visualization into the design of program diagnosis systems intended for novices. We report an empirical evaluation to assess the effectiveness of supporting visualization features during problem solving. The system used in the evaluation integrates visualzation and immediacy features and supports a model-tracing based approach to program diagnosis. Unlike other similar systems, our prototype system supports a more flexible style of interaction by increasing the grain size of diagnosis to a complete programming statement. The evaluation reported here seems to suggest that when supported with visualization features, systems for program diagnosis tend to be more effective in helping the users during problem solving.     

Cognitive Factors in Programming with Diagrams

Visual programming languages aim to broaden the use ofdiagrams within the software industry, to the extentthat they are integrated into the programming languageitself. As a result, they provide an ideal opportunityto study the benefits of diagrams as an externalrepresentation during problem solving: not only isprogramming a challenging problem-solving activity,but the effect of diagram usage can be directlyassessed by comparing performance while using a visualprogramming language to performance with a standardtextual language. There have been severalmisconceptions amongst visual language researchersregarding the role of diagrams in software design, butthese are being addressed by empirical studies and bynew theories of notation design derived from studiesof visual programming. Based on this research, theauthors are able to recommend several new directionsfor research into thinking with diagrams.     

On the Convergence of Temporal-Difference Learning with Linear Function Approximation

The asymptotic properties of temporal-difference learning algorithms with linear function approximation are analyzed in this paper. The analysis is carried out in the context of the approximation of a discounted cost-to-go function associated with an uncontrolled Markov chain with an uncountable finite-dimensional state-space. Under mild conditions, the almost sure convergence of temporal-difference learning algorithms with linear function approximation is established and an upper bound for their asymptotic approximation error is determined. The obtained results are a generalization and extension of the existing results related to the asymptotic behavior of temporal-difference learning. Moreover, they cover cases to which the existing results cannot be applied, while the adopted assumptions seem to be the weakest possible under which the almost sure convergence of temporal-difference learning algorithms is still possible to be demonstrated.     

Support for Implementation of Evolutionary Concurrent Systems

Concurrent programs that embed specifications of synchronizations in the body of their component are difficult to extend and modify. Small changes in a concurrent program, particularly changes in the interactions among components, may require re-implementation of a large number of components. Even specifications of components cannot be reused easily. This paper presents a concurrent program composition mechanism in which both specification and implementation of computations and interactions are completely separated. Separation of specifications and implementations facilitates extensions and modifications of programs by allowing one to separately change the implementations of computations and interactions. It also supports their reusability. The paper also describes the design and implementation of a concurrent object-oriented programming language based on this model, including a compiler for the language, and reports on the execution behavior of programs written in the language.     

A Two-Layer Robot Controller Design Using Evolutionary Algorithms

The results obtained by a rule-based proportional, integral, derivative (PID) precompensator controller applied to a two-joint manipulator are discussed. The end effector is made to follow a specified trajectory obtained from the inverse kinematics by an appropriate design of a fuzzy control law. The desired trajectory is determined by the values of the joint variables and the structural kinematics parameters of the manipulator. The performance of the PID controller is exploited here to build a fuzzy precompensator that will enhance the conventional PID and to obtain better performances and results. The fuzzy rule base of the precompensator designed is found by associating two evolutionary algorithms that search for the optimal solution.     

用进化RBF神经网络控制Lorenz混沌

对镇定一嵌入在Lorenz混沌吸引子内的不稳定平衡点上的混沌轨道提出了一种利用进化RBF网控制混沌系统的新方法,采用了基于两层编码改进进化规划（IEP）的RBF网学习算法,要以同时确定网络的拓扑结构和参数,仿真结果表明本文控制Lorenz混沌响应速度快,控制精度高。     

动态系统的常微分方程组建模—基于不同搜索技术的实验研究

以人口模型和化学反应模型为例,通过大量实验研究比较了分别采用基于两种传统的搜索方法即局部搜索算法和模拟退火算法、遗传算法（简称GA）四者相结合的14种不同算法建立动态系统的常微分方程组模型的实验结果,得到了有关各算法性能比较的一些新的结论。两个实例的实验结果表明：在14种算法中,GP＋GA＋LS－MU算法（即在采用GP的模型结构的优化过程中嵌入采用GA的模型参数的优化过程,并且在每一演化代对种群中的部分个体进行基于GP的标准变异算子产生邻域解的局域搜索过程）是目前解决常微分方程组建模问题的最好算法。     

开发支持IPv6的应用程序

随着IPv6(Internet Protocol Version 6）网络的部署,程序员必须面对多种地址协议族（Address Families,AF)的接口（socket）连接问题,使其程序同时支持IPv6和IPv4。结合作者的实验经验,研究网络编程时处理多个地址协议族的问题,以期实现与地址协议族无关的编程。     

支票挂失查询语音服务系统的设计与实现

自动声讯服务日益成为信息查询或传递的一条重要途径,文中介绍了支票挂失查询语音服务系统的功能,结构以及系统设计中 音卡编程的状态轮询方法和数据库的操作方法。