Quantization effects in digitally behaving circuit implementationsof Kohonen networks

Implementing a neural network on a digital or mixed analog and digital chip yields the quantization of the synaptic weights dynamics. This paper addresses this topic in the case of Kohonen's self-organizing maps. We first study qualitatively how the quantization affects the convergence and the properties, and deduce from this analysis the way to choose the parameters of the network (adaptation gain and neighborhood). We show that a spatially decreasing neighborhood function is far more preferable than the usually rectangular neighborhood function, because of the weight quantization. Based on these results, an analog nonlinear network, integrated in a standard CMOS technology, and implementing this spatially decreasing neighborhood function is then presented. It can be used in a mixed analog and digital circuit implementation.     

Genetic evolution of the topology and weight distribution of neuralnetworks

This paper proposes a system based on a parallel genetic algorithm with enhanced encoding and operational abilities. The system, used to evolve feedforward artificial neural networks, has been applied to two widely different problem areas: Boolean function learning and robot control. It is shown that the good results obtained in both cases are due to two factors: first, the enhanced exploration abilities provided by the search-space reducing evolution of both coding granularity and network topology, and, second, the enhanced exploitational abilities due to a recently proposed cooperative local optimizing genetic operator.     

Large-deviation inequalities for parameter estimators in stochastic systems

Translated from Kibernetika i Sistemnyi Analiz, No. 2, pp. 95–112, March–April, 1994.     

Approximation of the optimal-time problem for controlled differential inclusions

Translated from Kibernetika i Sistemnyi Analiz, No. 3, pp. 172–176, May–June, 1994     

Long range peak demand forecasting under conditions of high growth

An extended logistic model with a varying asymptotic upper bound for long-range peak demand forecasting is described. The model has been applied to a typical fast growing system, the Saudi Consolidated Electric Company. The forecasts are compared with actual demands and with those obtained from classical forecasting methods. The model gave relatively accurate peak demand forecasts compared with other classical methods. The model with a single load observation is capable of producing several peak demand forecasts corresponding to different levels of maximum temperature and various levels of social activity. The forecasts produced by the model were also stable irrespective of the length of the ex-post simulation period     

Active,passive and transpassive dissolution of a nickel base super alloy in concentrated acid mixture solution

The electrochemical and corrosion behaviour of a nickel base super alloy (C-263) has been investigated in the deaerated binary and ternary solution mixture of concentrated phosphoric acid, acetic acid, sulphuric acid, nitric acid or water using potentiostatic technique at 35°C. The possibilities of electropolishing of this alloy in these solution mixtures have been also explored. The alloy showed distinct active, passive and transpassive behaviour in the experimental solutions. The alloy remained active and turned passive in the negative potential region. Transpassive dissolution of the alloy is observed and electropolishing is achieved in this region. The best electropolishing is obtained in 50% H₃PO₄ + 40% CH₃COOH + 10% H₂SO₄. Higher content of water in the electrolytic solution is not useful for electropolishing of the alloy The experimental results also suggest that a current plateau in the transpassive potential region is not a sufficient condition to achieve electropolishing.     

Characterization of allergoids from ovalbumin in vitro and in vivo

Several in vivo and in vitro methods for monitoring immunological properties of two allergoids obtained by formaldehyde treatment of ovalbumin (OA) were developed. The calculated molecular weight of allergoids was 80 kD (OA-F1) and 165 kD (OA-F2), respectively. The allergenic activity in vitro of allergoids in mast-cell histamine release assay was 1000 times lower than of OA. Both allergoids showed reduced ability to induce passive cutaneous anaphylaxis in the Sprague-Dawley rats or systemic anaphylaxis in Dunkin-Harley guinea-pigs. The ability of OA and allergoids to bind to the OA-specific IgE antibodies was measured in vivo by the inhibition of passive cutaneous anaphylaxis (PCA-inhibition). Allergoid binding to IgE was 51-66% lower than the native allergen. Moreover, the avidity of OA-specific IgG antibodies, measured by ELISA-inhibition, for allergoids and allergen was of the same order. Allergoids induced a different pattern of humoral immune response from that, induced by the native allergen. Thus, after immunization of BALB/c mouse, both allergoids induced a higher production of IgG and a lower production of IgE than OA, only OA-F2 induced a lower production of IgG1. The differences in the IgA response to the immunogens was not significant. Delayed hypersensitivity studies in the BALB/c mouse showed that allergoids were 5- to 12-times less effective in inducing a cell-mediated immune response than OA. The present study provides a battery of immunological methods for preclinical testing of modified allergens.