Learn More
47 descriptions of classes (patterns), providing mass parallelism at data processing and significant acceleration of processes for solution making in the process of pattern recognition has been described. Results have been obtained at support of RFBR grant № 06-08-01612-а and Program " Innovations Support " of Presidium of RAS. Abstract: The architecture of(More)
In the paper, an ontogenic artificial neural network (ANNs) is proposed. The network uses orthogonal activation functions that allow significant reducing of computational complexity. Another advantage is numerical stability, because the system of activation functions is linearly independent by definition. A learning procedure for proposed ANN with(More)
New non-conventional system of the computational intelligence is proposed. It has growing structure similar to the Cascade-Correlation Learning Architecture designed by S. E. Fahlman and C. Lebiere but differs from it in type of artificial neurons. Quadratic neurons are used as nodes in introduced architecture. These simple elements can be quickly adjusted(More)
The architecture of forecasting adaptive wavelet-neuro-fuzzy-network and its learning algorithm for the solving of nonstationary processes forecasting tasks are proposed. The learning algorithm is optimal on rate of convergence and allows to tune both the synaptic weights and dilations and translations parameters of wavelet activation functions. The(More)
In the paper the problem of on-line diagnostics and properties change detection of systems whose output signal is multidimensional non-stationary stochastic sequence is considered. The six-layer diagnostic neuro-fuzzy system is proposed. The first layer of this system consists of membership functions blocks, the second layer provides aggregation of(More)
in the paper new non-conventional growing neural network is proposed. It coincides with the Cascade-Correlation Learning Architecture structurally, but uses ortho-neurons as basic structure units, which can be adjusted using linear tuning procedures. As compared with conventional approximating neural networks proposed approach allows significantly to reduce(More)
Nowadays computational intelligence methods are widely spread in different tasks solving in Data Mining under uncertain, nonlinear, stochastic, chaotic and disturbed by different type of noises conditions. In the paper the hybrid neuro-neo-fuzzy system of computational intelligence is proposed. This system is distinguished by the computational simplicity,(More)
Architecture and learning algorithm of self-learning spiking neural network in fuzzy clustering task are outlined. Fuzzy receptive neurons for pulse-position transformation of input data are considered. It is proposed to treat a spiking neural network in terms of classical automatic control theory apparatus based on the Laplace transform. It is shown that(More)
The paper introduces a Newton-type modification of temporal Hebbian rule-based learning algorithm of a self-learning spiking neural network. Similar to conventional artificial neural networks domain, the learning algorithm modification based on second-order optimization procedures allows of improving performance of the third generation neural networks. The(More)