Application of neural networks and machine learning in networkdesign

Communication network design is becoming increasingly complex, involving making networks more usable, affordable, and reliable. To help with this, we have proposed an expert network designer (END) for configuring, modeling, simulating, and evaluating large structured computer networks, employing artificial intelligence, knowledge representation, and network simulation tools. We present a neural network/knowledge acquisition machine-learning approach to improve the END's efficiency in solving the network design problem and to extend its scope to acquire new networking technologies, learn new network design techniques, and update the specifications of existing technologies     

Series equivalence/operation of current-controlled boost-typesingle-phase voltage source converters for bidirectional power flow

Single-phase voltage source power converters (VSCs) under consideration are AC-DC current-controlled boost-type power converters with bidirectional power-handling capability. Equivalence between two series-connected two-level power converters and a single three-level power converter is considered here. Further considered is the series operation of three-level power converters. Simulation results and experimental verification for both are provided. Economical configurations of three-level power converters leading to multilevel waveforms are presented thereafter     

Low-truncation-error finite difference equations for photonicssimulation. II. Vertical-cavity surface-emitting lasers

For part I see, ibid., p. 134, 1998. The basic approach outlined in the previous article is applied to the difficult problem of computing the optical modes of a vertical-cavity surface-emitting laser. The formulation utilizes a finite difference equation based upon the lowest order term of an infinite series solution of the scalar Helmholtz equation in a local region. This difference equation becomes exact in the one-dimensional (1-D) limit, and is thus ideally suited for nearly 1-D devices such as vertical-cavity lasers. The performance of the resulting code is tested on both a simple cylindrical cavity with known solutions and an oxide-confined vertical-cavity laser structure, and the results compared against second-order-accurate code based upon Crank-Nicolson differencing     

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.     

E-kernel: an embedding kernel on the IBM victor V256 multiprocessorfor program mapping and network reconfiguration

We present the design of E-kernel, an embedding kernel on the Victor V256 message-passing partitionable multiprocessor, developed for the support of program mapping and network reconfiguration. E-kernel supports the embedding of a new network topology onto Victor's 2D mesh and also the embedding of a task graph onto the 2D mesh network or the reconfigured network. In the current implementation, the reconfigured network can be a line or an even-size ring, and the task graphs meshes or tori of a variety of dimensions and shapes or graphs with similar topologies. For application programs having these task graph topologies and that are designed according to the communication model of E-kernel, they can be run without any change on partitions connected by the 2D mesh, line, or ring. Further, E-kernel attempts the communication optimization of these programs on the different networks automatically, thus making both the network topology and the communication optimization attempt completely transparent to the application programs. Many of the embeddings used in E-kernel are optimal or asymptotically optimal (with respect to minimum dilation cost). The implementation of E-kernel translated some of the many theoretical results in graph embeddings into practical tools for program mapping and network reconfiguration in a parallel system. E-kernel is functional on Victor V256. Measurements of E-kernel's performance on V256 are also included     

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.     

Algebra of algorithms and Kaluzhnin's graphschemas

This research was supported under project No. 6.02.02/128-93 as part of the state scientific-technical program on future information technologies and systems by the Ukrainian State Committee, of Science and Technology.     

Components of random generation by normal subjects and patients with dysexecutive syndrome

The study presents a hypothesis on how randomness could be simulated by human subjects. Three sources of deviation from randomness are predicted: (1) the preferred application of overlearned production schemata for producing sequences of digits, (2) a wrong concept of randomness, and (3) the impossibility to monitor for redundancies of higher- than those of first-order. Deviations of random generation of digits produced by healthy subjects, patients with chronic frontal lobe damage, and patients with Parkinson's disease from random sequences produced by a computer program can be explained by the differential influence of these factors. Whereas incorrect concepts of randomness and limits on monitoring capacity distinguished all sequences produced by humans from actual random sequences, persistence on a single production strategy distinguished brain-damaged patients from controls. Random generation of digits appears to be a theoretically transparent and clinically useful test of executive function.