International Journal of Computational Intelligence and Applications

Recently the study on the theory of wavelets shows that the wavelets have not only the multi-resolution property both in frequency and time domain, but also the good approximation ability. SVMs based on the statistical learning theory are a kind of general and effective learning machines, and have described for us the nice application blueprint in machine learning domain. There exists a bottleneck problem, or the pre-selection of kernel parameter for SVMs. In this paper, the orthogonal projection kernels of father wavelet (OPFW kernels) are introduced into SVMs. In doing so SVMs based on the OPFW kernels can have good performance in both approximation and generalisation. Simultaneously the parameter pre-selection of wavelet kernels can be implemented by discrete wavelet transform. Experiments on regression estimation illustrate the approximation and generalisation ability of our method.

We present a system for the automatic synthesis of classifiers. The CellNet system for generating binary pattern classifiers is used as a base for further experimentation. As in the original CellNet software, we evolve pattern recognisers (hunters). However, in this version called CellNet Co-Ev, we also evolve the patterns (prey) in a competitive co-evolution. Patterns evolve through the application of camouflage functions, which are used to obscure the data naturally found in the database. The addition of this competitive co-evolution yields a larger and more varied database, artificially increasing the difficulty of the classification task. Application to the CEDAR database of handwritten characters shows an increase in the reliability of the evolution of recognisers, as well as in the elimination of over-fitting, relative to the original CellNet software.

A robust method, fuzzy kNNModel, for toxicity prediction of chemical compounds is proposed. The method is based on a supervised clustering method, called kNNModel, which employs fuzzy partitioning instead of crisp partitioning to group clusters. The merits of fuzzy kNNModel are two-fold: (1) it overcomes the problems of choosing the parameter ε — allowed error rate in a cluster and the parameter N — minimal number of instances covered by a cluster, for each data set; (2) it better captures the characteristics of boundary data by assigning them with different degrees of membership between 0 and 1 to different clusters. The experimental results of fuzzy kNNModel conducted on thirteen public data sets from UCI machine learning repository and seven toxicity data sets from real-world applications, are compared with the results of fuzzy c-means clustering, k-means clustering, kNN, fuzzy kNN, and kNNModel in terms of classification performance. This application shows that fuzzy kNNModel is a promising method for the toxicity prediction of chemical compounds.

How to detect the topological degree (TD) of a function is of vital importance in investigating the existence and the number of zero values in the function, which is a topic of major significance in the theory of nonlinear scientific fields. Usually a sufficient refinement of the boundary of the polyhedron decided by Boult and Sikorski algorithm (BS) is needed as prerequisite when the well known method of Stenger and Kearfott is chosen for computing TD. However two linchpins are indispensable to BS, the parameter δ on the boundary of the polyhedron and an estimation of the Lipschitz constant K of the function, whose computations are analytically difficult. In this paper, through an appropriate scheme that transforms the problems of computing δ and K into searching optimums of two non-differentiable functions, a novel differential evolution algorithm (DE) combined with established techniques is proposed as an alternative method to computing δ and K. Firstly it uses uniform design method to generate the initial population in feasible field so as to have the property of large scale convergence, without better approximation of the unknown parameter as iterative initial point. Secondly, it restrains the normal DE's local convergence limitation virtually through deflection and stretching of objective function. The main advantages of the put algorithm are its simplicity and its ability to work by using function values solely. Finally, details of applying the proposed method into computing δ and K are given, and experimental results on two benchmark problems in contrast to the results reported have demonstrated the promising performance of the proposed algorithm in different scenarios.

Consensus clustering refers to combining multiple clusterings over a common set of objects into a single consolidated partition. After introducing the distribution-based view of partitions, we propose a series of entropy-based distance functions for comparing various partitions. Given a candidate partition set, consensus clustering is then formalized as an optimization problem of searching for a centroid partition with the smallest distance to that set. In addition to directly selecting the local centroid candidate, we also present two combining methods for the global centroid based on the new similarity determined by the whole candidate set. The centroid partition is likely to be top/middle-ranked in terms of closeness to the true partition. Finally we evaluate its effectiveness on both artificial and real datasets, with candidates from either the full space or the subspace.

The paper presents an electric load balancing system for domestic use. An electric load forecasting system, which is realized by a genetic algorithm-based modified neural network, is employed. On forecasting the home power consumption profile, the load balancing system can adjust the amount of energy stored in battery accordingly, preventing it from reaching certain practical limits. A steady consumption from the AC mains can then be obtained which will benefit both the users and the utility company. An example will be given to illustrate the merits of the forecaster, and its performance on achieving the load balancing.

The article presents the development of a Classifiers Combination based on machine learning techniques (Artificial Neural Networks, Support Vector Machines and C4.5 algorithm) which were able to increase the performance achieved by the Refractive Errors Measurement System (REMS) that analyzes Hartmann-Shack (HS) images from human eyes. The HS images are analyzed in order to extract relevant data for identification of refractive errors (myopia, hypermetropia and astigmatism). Those data are extracted using Gabor wavelets transform and afterwards, machine learning techniques are employed to carry out the image analysis.