Marta Kolasa

A programmable triangular neighborhood function for a Kohonen self-organizing map implemented on chip

A new hardware implementation of the triangular neighborhood function (TF) for ultra-low power, self-organizing maps (SOM) is presented. Simulations carried out in the software model of this network show that even for low signal resolutions (3–6 bits) performance of the network is not affected. Resolution of the signal at the output of this block has a dominant influence on the circuit complexity as well as the energy consumption. The proposed mechanism is very fast. For a neighborhood range of 15 a delay in the circuit equals 20 ns that allows for data rates of 20–40 MHz, even for large maps with several hundreds neurons.

Parallel Programmable Asynchronous Neighborhood Mechanism for Kohonen SOM Implemented in CMOS Technology

We present a new programmable neighborhood mechanism for hardware implemented Kohonen self-organizing maps (SOMs) with three different map topologies realized on a single chip. The proposed circuit comes as a fully parallel and asynchronous architecture. The mechanism is very fast. In a medium sized map with several hundreds neurons implemented in the complementary metal-oxide semiconductor 0.18 μm technology, all neurons start adapting the weights after no more than 11 ns. The adaptation is then carried out in parallel. This is an evident advantage in comparison with the commonly used software-realized SOMs. The circuit is robust against the process, supply voltage and environment temperature variations. Due to a simple structure, it features low energy consumption of a few pJ per neuron per a single learning pattern. In this paper, we discuss different aspects of hardware realization, such as a suitable selection of the map topology and the initial neighborhood range, as the optimization of these parameters is essential when looking from the circuit complexity point of view. For the optimal values of these parameters, the chip area and the power dissipation can be reduced even by 60% and 80%, respectively, without affecting the quality of learning.

Application of Artificial Neural Network to Predict Survival Time for Patients with Bladder Cancer

This paper presents an application of an artificial neural network to determine survival time of patients with a bladder cancer. Different learning methods have been investigated to find a solution, which is most optimal from a computational complexity point of view. In our study, a model of a multilayer perceptron with a training algorithm based on an error back-propagation method with a momentum component was applied. Data analysis was performed using the perceptron with one hidden layer and training methods with incremental and cumulative neuron weight updating. We have examined an influence of the order in the training data file on the final prediction results. The efficiency of the proposed methodology in the bladder urothelial cancer prediction after cystectomy is on the level of 90%, which is the best result ever reported. Best outcomes one achieves for 5 neurons in the hidden layer.

Low Power, Low Chip Area, Digital Distance Calculation Circuit for Self-Organizing Neural Networks Realized in the CMOS Technology

This paper presents a new distance calculation circuit (DCC) that in artificial neural networks is used to calculate distances between vectors of signals. The proposed circuit is a digital, fully parallel and asynchronous solution. The complexity of the circuit strongly depends on the type of the distance measure. Considering two popular measures i.e. the Euclidean (L2) and the Manhattan (L1) one, it is shown that in the L2 case the number of transistors is even ten times larger than in the L1 case. Investigations carried out on the system level show that the L1 measure is a good estimate of the L2 one. For the L1 measure, for an example case of 4 inputs, for 10 bits of resolution of the signals, the number of transistors is equal to c. 2500. As transistors of minimum sizes can be used, the chip area of a single DCC, if realized in the CMOS 180 nm technology, is less than 0.015 mm2.

A Comparative Study of Different Neighborhood Topologies in WTM Kohonen Self-Organizing Maps

In this paper we present a software model of the Winner Takes Most (WTM) Kohonen neural network (KNN) with different types of the neighborhood grid. The proposed network model allows for analysis of the convergence properties such as the quantization error and the convergence time for different grids, which is essential looking from the hardware implementation point of view of such networks. Particular grids differ in complexity, which in hardware implementation has a direct influence on power dissipation as well as on chip area and the final production cost. The presented results show that even the simplest rectangular grid with four neighbors allows for good convergence properties for different training data files.

Analysis of Significant Prognostic Factors of Patients with Bladder Cancer Using Self-Organizing Maps

This study presents a new approach to determine significant prognostic factors for patients suffering from the bladder cancer. The analysis of medical data has been performed by the use of the Kohonen self-organizing map (SOM). The SOM allows visualizing and identifying the prognostic factors indicating which of them are significant. A database comprised of ninety patients has been used in this study. Seven predictors were investigated. The cluster analysis indicates that the significant prognostic factors for the bladder cancer are: histological grade (cG) and stage (cT). The obtained results also showed that the sex and the cG variables are highly correlated and that the number of non-classic differentiation (NDNc) features in bladder cancer is somewhat correlated to surgically removed lymphnode number (LN) and metastatic positive lymphnode number (PLN).

A New Model of the Neuron for Biological Spiking Neural Network Suitable for Parallel Data Processing Realized in Hardware

The paper presents a modification of the structure of a biological neural network (BNN) based on spiking neuron models. The proposed modification allows to influence the level of the stimulus response of particular neurons in the BNN. We consider an extended, three-dimensional Hodgkin-Huxley model of the neural cell. A typical BNN composed of such neural cells have been expanded by addition of resistors in each branch point. The resistors can be treated as the weights in such BNN. We demonstrate that adding these elements to the BNN significantly affects the waveform of the potential on the membrane of the neuron, causing an uncontrolled excitation. This provides a better description of processes that take place in nervous cell. Such BNN enables an easy adaptation of the learning rules used in artificial or spiking neural networks. The modified BNN has been implemented on Graphics Processing Unit (GPU) in the CUDA C language. This platform enables a parallel data processing, which is an important feature in such applications.

Human Gait Analysis and Classification Based on Neural Networks and Fuzzy Logic

Human gait analysis and classification is the process of identifying individuals by their walking manners. Computerized gait analysis using neural networks and fuzzy logic has become an integral part of the treatment decision-making process. Authors proposed the integration of kinetic data, more specifically power joints in combination with neural networks and fuzzy logic. It is a relatively new addition to other types of data including temporal and stride parameters. The performance of our approach was verified in laboratory for motion analysis. The obtained results are satisfying.