We present an implementation of Kohonen Self-Organizing Feature Maps for the Spert-II vector microprocessor system. The implementation supports arbitrary neural map topologies and arbitrary neighborhood functions. For small networks, as used in real-world tasks, a single Spert-II board is measured to run Kohonen net classiication at up to 208 million connections per second (MCPS). On a speech c...

Nowadays, neural networks are largely used in signal processing and images. In particular, Kohonen networks or Self Organizing Maps are unsupervised learning models. This method performs a vector quantization (VQ) on the values obtained after processing. The vector quantization has a potential to give more data compression maintaining the same quality. In this paper we propose new scheme to ima...

This paper introduces an intelligent system for monitoring and recognition of process disturbances during short-circuiting gas–metal arc welding. It is based on the measured and statistically processed data of welding electrical parameters. A 12-dimensional array of process features is designed to describe various welding conditions and is employed as input vector of the intelligent system. Thr...

We propose a two-stage hierarchical arti®cial neural network for the segmentation of color images based on the Kohonen self-organizing map (SOM). The ®rst stage of the network employs a ®xed-size two-dimensional feature map that captures the dominant colors of an image in an unsupervised mode. The second stage combines a variable-sized one-dimensional feature map and color merging to control th...

We compared the application of different algorithms to document clustering. The algorithms studied were Fuzzy C-Means, Fuzzy ART, Fuzzy ART for Fuzzy Clusters, Fuzzy Max-Min, and the Kohonen neural network (only the first is not a neural network). We generated a testbed from LISA, using some of the descriptors corresponding to the different records for the comparison of the results. The best re...

A feed-forward neural network is proposed for monitoring operating modes of large scale processes. A Gaussian hidden layer associated with a Kohonen output layer map the principal features of measurements of state variables. Subsets of selective neurons are generated into the hidden layer by means of self adapting of centers and dispersions parameters of the Gaussian functions. The output layer...

This study presents a novel Adaptive resonance theory-Counterpropagation neural network (ART-CPN) for solving forecasting problems. The network is based on the ART concept and the CPN learning algorithm for constructing the neural network. The vigilance parameter is used to automatically generate the nodes of the cluster layer for the CPN learning process. This process improves the initial weig...

This article describes research in the application of a ( topics) in an electronic meeting setting. Electronic meetKohonen Self-Organizing Map (SOM) to the problem of ing systems (EMSs) provide support for large groups classification of electronic brainstorming output and an interactively working on a single problem or collection evaluation of the results. Electronic brainstorming is one of pro...

* R & D Center for Intelligent Systems, King Mongkut’s Institute of Technology North Bangkok. ** Department of Computing and Electrical Engineering, Heriot-Watt University, Edinburgh, United Kingdom Abstract This paper presents the use of neural networks and a genetic algorithm in time-optimal control of a closed-loop 3-dof robotic system. Extended Kohonen networks which contain an additional l...

Computer-aided systems based on artificial neural networks (ANN) are suitable for automatic chromosome classification process. However, the software based implementation of these systems running on conventional computer transforms the parallelism features of the ANN into serial operations that reduce their computation power. The hardware implementation of such system can achieve the parallelism...