Foreign exchange trading has emerged in recent times as a significant activity in many countries. As with most forms of trading, the activity is influenced by many random parameters so that the creation of a system that effectively emulates the trading process is very helpful. In this paper, we try to create such a system with a genetic algorithm engine to emulate trader behaviour on the foreign exchange market and to find the most profitable trading strategy.

Support vector machines (SVMs) have recently emerged as a powerful technique for solving problems in pattern classification and regression. Best performance is obtained from the SVM its parameters have their values optimally set. In practice, good parameter settings are usually obtained by a lengthy process of trial and error. This paper describes the use of genetic algorithm to evolve these parameter settings for an application in mobile robotics.

This paper is concerned with the use of scientific visualization methods for the analysis of feedforward neural networks (NNs). Inevitably, the kinds of data associated with the design and implementation of neural networks are of very high dimensionality, presenting a major challenge for visualization. A method is described using the well-known statistical technique of principal component analysis (PCA). This is found to be an effective and useful method of visualizing the learning trajectories of many learning algorithms such as backpropagation and can also be used to provide insight into the learning process and the nature of the error surface.

A neural network model called the hierarchical overlapped neural gas (HONG) network is introduced and its performance on several datasets is described. In order to obtain improved classification accuracy, the HONG network partitions the input space by projecting the input data onto several different second layer neural gas networks. This duplication enables the HONG network to generate multiple classifications for every sample presented in the form of confidence values, and these confidence values are combined to obtain the final classification. Excellent recognition rates for several benchmark datasets are presented.

Our primary aim is to develop a classifier system that is capable of predicting the success or failure of hip prostheses on the basis of data from early radiological observations. The data set we employ (collected at The Royal London Hospital) records observations taken in the early years following fixation of the prosthesis and failure or otherwise after ten years. Many of the records contained in this data set have missing values. Recent work on the well-known Pima Indian data set has demonstrated the effectiveness of the Naive-Bayes (NB) method, coupled with boosting, on data with missing values. In this paper we investigate the performance of the NB method and boosting on the hip prosthesis data which contains a much greater proportion of missing values than the Pima Indian data. Our data set is additionally challenging in that it contains many more examples of one class (success) than the other.

A constructive neural-network algorithm is presented. For any consistent classification task on real-valued training vectors, the algorithm constructs a feedforward network with a single hidden layer of threshold units which implements the task. The algorithm, which we call CARVE, extends the "sequential learning" algorithm of Marchand et al. from Boolean inputs to the real-valued input case, and uses convex hull methods for the determination of the network weights. The algorithm is an efficient training scheme for producing near-minimal network solutions for arbitrary classification tasks. The algorithm is applied to a number of benchmark problems including Gorman and Sejnowski's sonar data, the Monks problems and Fisher's iris data. A significant application of the constructive algorithm is in providing an initial network topology and initial weights for other neural-network training schemes and this is demonstrated by application to backpropagation.

We describe a sequence of experiments investigating the strengths and limitations of Fukushima's neocognitron as a handwritten digit classifier. Using the results of these experiments as a foundation, we propose and evaluate improvements to Fukushima's original network in an effort to obtain higher recognition performance. The neocognitron performance is shown to be strongly dependent on the choice of selectivity parameters and we present two methods to adjust these variables. Performance of the network under the more effective of the two new selectivity adjustment techniques suggests that the network fails to exploit the features that distinguish different classes of input data. To avoid this shortcoming, the network's final layer cells were replaced by a nonlinear classifier (a multilayer perceptron) to create a hybrid architecture. Tests of Fukushima's original system and the novel systems proposed in this paper suggest that it may be difficult for the neocognitron to achieve the performance of existing digit classifiers due to its reliance upon the supervisor's choice of selectivity parameters and training data.

A difficulty with the application of the closed-form training algorithm for the neocognitron proposed by T.H. Hildebrandt (ibid., vol.2, p.557-88, Nov. 1991) is reported. In applying this algorithm the commenters have observed that S-cells frequently fail to respond to features that they have been trained to extract. Results which indicate that this training vector rejection in an important factor in the overall classification performance of the neocognitron trained using Hildebrandt's procedure are presented. In reply, Hildebrandt explains that the negative results obtained by the commenter are not specific to the proposed algorithm and are easily explained in terms of set theory.

A gradient descent algorithm suitable for training multilayer feedforward networks of processing units with hard-limiting output functions is presented. The conventional backpropagation algorithm cannot be applied in this case because the required derivatives are not available. However, if the network weights are random variables with smooth distribution functions, the probability of a hard-limiting unit taking one of its two possible values is a continuously differentiable function. In the paper, this is used to develop an algorithm similar to backpropagation, but for the hard-limiting case. It is shown that the computational framework of this algorithm is similar to standard backpropagation, but there is an additional computational expense involved in the estimation of gradients. Upper bounds on this estimation penalty are given. Two examples which indicate that, when this algorithm is used to train networks of hard-limiting units, its performance is similar to that of conventional backpropagation applied to networks of units with sigmoidal characteristics are presented.

It is shown how a major (and questionable) assumption underlying a previously reported approach to the modeling of software testing can be relaxed in order to provide a more realistic model. Under the assumption of uniform execution the new model is found to perform only marginally better than the previous model, indicating that the uniform execution assumption is a poor one. A nonuniform execution model is then shown to give very good performance on application to three sets of software reliability data. Attention is also devoted to the problem of comparing the performance of different models, and some difficulties in this area are noted.

In this paper, an approach to the modeling of software testing is described. A major aim of this approach is to allow the assessment of the effects of different testing (and debugging) strategies in different situations. It is shown how the techniques developed can be used to estimate, prior to the commencement of testing, the optimum allocation of test effort for software which is to be nonuniformly executed in its operational phase. In addition, the question of application of statistical models in cases where the data environment undergoes changes is discussed. Finally, two models are presented for the assessment of the effects of imperfections in the debugging process.

A method is proposed for the analysis of a teletraffic network with alternative routing. The approach used is to model a "general link" as a separate entity, being a first step towards analyzing the network in a link-by-link fashion. To this end the general link is considered to be a group of fully available trunks offered a number of separate streams of traffic. By considering the multivariate point process formed by the total stream of arriving calls, the individual streams of traffic overflowing from, and the individual streams of traffic carried on the common trunk, are distinguished. All streams of traffic in the network are considered to be well described by simple renewal processes. The validity of the method is illustrated with exact numerical results and by simulation.

In this paper is a formulation of the one-moment method of analysis for a telephone traffic network in terms of the link offered traffics. By using a slightly strengthened version of the assumptions hitherto used in one-moment analysis, and by developing a type of limited availability link model, it is shown how the offered traffic to each link may be found, and thus how the network analysis may be reduced to a valid set of link analyses.