Learning in neural networks and stochastic approximation methods with averaging

The problem of adjusting the weights (learning) in multilayer feedforward neural networks (NN) is known to be of a high importance when utilizing NN techniques in various practical applications. The learning procedure is to be performed as fast as possible and in a simple computational fashion, the two requirements which are usually not satisfied practically by the methods developed so far. Moreover, the presence of random inaccuracies are usually not taken into account. In view of these three issues, an alternative stochastic approximation approach discussed in the paper, seems to be very promising.

Turning on the alarm: the neural mechanisms of the transition from innocuous to painful sensation

The experience of pain occurs when the level of a stimulus is sufficient to elicit a marked affective response, putatively to warn the organism of potential danger and motivate appropriate behavioral responses. Understanding the biological mechanisms of the transition from innocuous to painful levels of sensation is essential to understanding pain perception as well as clinical conditions characterized by abnormal relationships between stimulation and pain response. Thus, the primary objective of this study was to characterize the neural response associated with this transition and the correspondence between that response and subjective reports of pain. Towards this goal, this study examined BOLD response profiles across a range of temperatures spanning the pain threshold. 14 healthy adults underwent functional magnetic resonance imaging (fMRI) while a range of thermal stimuli (44-49oC) were applied. BOLD responses showed a sigmoidal profile along the range of temperatures in a network of brain regions including insula and mid- cingulate, as well as a number of regions associated with motor responses including ventral lateral nuclei of the thalamus, globus pallidus and premotor cortex. A sigmoid function fit to the BOLD responses in these regions explained up to 85% of the variance in individual pain ratings, and yielded an estimate of the temperature of steepest transition from non-painful to painful heat that was nearly identical to that generated by subjective ratings. These results demonstrate a precise characterization of the relationship between objective levels of stimulation, resulting neural activation, and subjective experience of pain and provide direct evidence for a neural mechanism supporting the nonlinear transition from innocuous to painful levels along the sensory continuum.

The system identification and control of Hammerstein system using non-uniform rational B-spline neural network and particle swarm optimization

In this paper a new system identification algorithm is introduced for Hammerstein systems based on observational input/output data. The nonlinear static function in the Hammerstein system is modelled using a non-uniform rational B-spline (NURB) neural network. The proposed system identification algorithm for this NURB network based Hammerstein system consists of two successive stages. First the shaping parameters in NURB network are estimated using a particle swarm optimization (PSO) procedure. Then the remaining parameters are estimated by the method of the singular value decomposition (SVD). Numerical examples including a model based controller are utilized to demonstrate the efficacy of the proposed approach. The controller consists of computing the inverse of the nonlinear static function approximated by NURB network, followed by a linear pole assignment controller.

A model of UK domestic appliance-use using Artificial Neural Networks

The development of an Artificial Neural Network model of UK domestic appliance energy consumption is presented. The model uses diary-style appliance use data and a survey questionnaire collected from 51 households during the summer of 2010. It also incorporates measured energy data and is sensitive to socioeconomic, physical dwelling and temperature variables. A prototype model is constructed in MATLAB using a two layer feed forward network with backpropagation training and has a12:10:24architecture.Model outputs include appliance load profiles which can be applied to the fields of energy planning (micro renewables and smart grids), building simulation tools and energy policy.

Stable Nonlinear Receding Horizon Regulator Using RBF Neural Network Models

The general stability theory of nonlinear receding horizon controllers has attracted much attention over the last fifteen years, and many algorithms have been proposed to ensure closed-loop stability. On the other hand many reports exist regarding the use of artificial neural network models in nonlinear receding horizon control. However, little attention has been given to the stability issue of these specific controllers. This paper addresses this problem and proposes to cast the nonlinear receding horizon control based on neural network models within the framework of an existing stabilising algorithm.

Real-time application of a constrained predictive controller based on dynamic neural networks with feedback linearization

This paper describes an experimental application of constrained predictive control and feedback linearisation based on dynamic neural networks. It also verifies experimentally a method for handling input constraints, which are transformed by the feedback linearisation mappings. A performance comparison with a PID controller is also provided. The experimental system consists of a laboratory based single link manipulator arm, which is controlled in real time using MATLAB/SIMULINK together with data acquisition equipment.