A new evolutionary algorithm for the constrained multiple destination routing problem is presented. The constrained multicast problem is characterized by a minimum cost multicast tree and a bounded end-to-end delay. I...
详细信息
A new evolutionary algorithm for the constrained multiple destination routing problem is presented. The constrained multicast problem is characterized by a minimum cost multicast tree and a bounded end-to-end delay. It has been proven that this problem is NP-complete. The proposed algorithm is based on a soft computing technique that integrates in an efficient manner the merits of genetic algorithms and concepts of the competitive learning in the artificial neural networks literature. A population based learning algorithm is utilized, among other techniques, to construct a delay bounded multicast tree. A salient feature of the algorithm is the adaptive learning concept that achieves an efficient trade-off between the exploration and exploitation of the search space.
We propose an evolutionary neural network-training algorithm for beta basis function neural networks (BBFNN). Classic training algorithms for neural networks start with a predetermined network structure. Generally the...
详细信息
We propose an evolutionary neural network-training algorithm for beta basis function neural networks (BBFNN). Classic training algorithms for neural networks start with a predetermined network structure. Generally the network resulting from learning applied to a predetermined architecture is either insufficient or over-complicated. This paper describes a hierarchical genetic learning model of the BBFNN. In order to examine the performance of the proposed algorithm, they were used for the approximation problems. The results obtained are very satisfactory with respect to the relative error.
In this paper we present a new architecture for combining classifiers. This approach integrates learning into the voting scheme used to aggregate individual classifiers decisions. This overcomes the drawbacks of havin...
详细信息
We present a comparison between different combining techniques in neural network ensembles. The main focus of this paper is on a new architecture that can be used in combining neural network ensembles. This architectu...
详细信息
ISBN:
(纸本)0780370449
We present a comparison between different combining techniques in neural network ensembles. The main focus of this paper is on a new architecture that can be used in combining neural network ensembles. This architecture is based on training two neural networks to perform the aggregation. One network is trained to establish a confidence factor for each member of the ensemble for every training entry. The other network performs the aggregation of the ensemble to present the final decision. Both these networks evolve together during training. This approach is compared with standard fixed and trained combining schemes.
An ensemble of neural networks offers several advantages over classical single classifier systems when applied to complex pattern classification problems. However, the performance of the ensemble as a unit depends not...
详细信息
ISBN:
(纸本)0780370449
An ensemble of neural networks offers several advantages over classical single classifier systems when applied to complex pattern classification problems. However, the performance of the ensemble as a unit depends not only on the effective aggregation of the modules decisions, but also on the accuracy of the individual classification decisions of each module. The accuracy at the modular level is a result of the quality of training received by each module. This paper presents an adaptive training algorithm that can be used to direct the training of the individual modules so as to improve the classification accuracy and training efficiency of the ensemble.
Artificial neural networks (ANNs) have demonstrated their success in many applications due to their ability to solve some problems with relative ease of use and the model-free property they enjoy. ANNs can solve probl...
详细信息
Artificial neural networks (ANNs) have demonstrated their success in many applications due to their ability to solve some problems with relative ease of use and the model-free property they enjoy. ANNs can solve problems without the need to understand or learn the analytical and statistical properties of the problem nor the solution steps. Research in ANNs has resulted in a variety of models and learning algorithms. In this paper, a brief review of recent advances in the field is presented. The paper then focuses on the recent work conducted by the author's group on modular neural networks. In particular, the paper discusses the different modular structures, modes of interactions, capabilities, co-operation among modules and fusion of their decisions. Performance of these models has proven to be superior to nonmodular neural networks.
Combining decisions from several classifiers can be used to improve on the results of handwritten characters recognition. There are different methods to combine these decisions, most of which are static. We present an...
详细信息
Combining decisions from several classifiers can be used to improve on the results of handwritten characters recognition. There are different methods to combine these decisions, most of which are static. We present an architecture that integrates learning into the voting scheme used to aggregate individual decisions. The focus of the work is to make the decision fusion a more adaptive process. This approach makes use of feature detectors responsible for gathering information about the input to perform adaptive decision aggregation. The approach is tested on handwritten Arabic character recognition. The results showed an improvement over any individual classifier, as well as different static classifier combining schemes.
Cooperation by voting is one of the popular modular neural network decision-making strategies. Ensemble classifiers are multiple identical modules which use voting for post-learning classification. This paper suggests...
详细信息
Cooperation by voting is one of the popular modular neural network decision-making strategies. Ensemble classifiers are multiple identical modules which use voting for post-learning classification. This paper suggests a new cooperation scheme for ensembles which utilizes voting in the learning process itself. According to the suggested scheme, different modules would, automatically, focus on different regions in the input space. Hence, temporal crosstalk decreases and decision boundaries are drawn accurately in complex overlapping regions of the input space.
Lossy image compression techniques aim at encoding images with a minimal representation. During this process, some visually useful information may be lost. Assessing the information loss in decompressed images is not ...
详细信息
Lossy image compression techniques aim at encoding images with a minimal representation. During this process, some visually useful information may be lost. Assessing the information loss in decompressed images is not an easy task. In this paper, a new quantitative image-quality measure is introduced. This new measure incorporates information theory into the most commonly used objective criterion (the mean square error). The new measure has been tested by experiments performed on a wide variety of images. The results show an increase in the correlation between subjective rating by human observers and the normalized mean square error after applying the new measure.
暂无评论