A Self-configuring Network

Connectionist research today is inhibited by two kinds of rigidity. First, a rigidity in architecture: the connectivity of most networks is fixed at the start by the programmer. This limits the universality of learning procedures. Secondly, a mental rigidity: the widespread assumption that a node's activation must be a real number and that activations should be combined using weighted sums. This paper explores the consequences of relaxing these rigidities. I describe a neural network for unsupervised pattern learning. Given an arbitrary environment of input patterns it grows into a configuration which allows it to represent the high-level regularities in the input. Like the Boltzmann machine, it runs in two phases: observing the environment and simulating the environment. It continually monitors its own performance and grows new nodes as the need for them is identified. Simulating the environment involves repeatedly choosing states to satisfy many constraints. The usual method is to maximize a ‘harmony’ function, which leads to a merging or blending of constraints: this lacks a clear semantics. My network uses logical inference to settle into a state consistent with as many as possible of the strongest constraints.     

Nonlinear modelling of periodic threshold autoregressions using Tsmars

We present new methods for modelling nonlinear threshold-type autoregressive behaviour in periodically correlated time series. The methods are illustrated using a series of average monthly flows of the Fraser River in British Columbia. Commonly used nonlinearity tests of the river flow data in each month indicate nonlinear behaviour in certain months. The periodic nonlinear correlation structure is modelled nonparametrically using TSMARS, a time series version of Friedman's extended multivariate adaptive regression splines (MARS) algorithm, which allows for categorical predictor variables. We discuss two methods of using the computational algorithm in TSMARS for modelling and fitting periodically correlated data. The first method applies the algorithm to data from each period separately. The second method models data from all periods simultaneously by incorporating an additional predictor variable to distinguish different behaviour in different periods, and allows for coalescing of data from periods with similar behaviour. The models obtained using TSMARS provide better short-term forecasts for the Fraser River data than a corresponding linear periodic AR model.     

FATTY ACID PROFILES OF LEAVES OF NINE EDIBLE WILD PLANTS: AN AUSTRALIAN STUDY

Nine species (Amaranthus viridis L. , Atriplexnummularia L. , Chenopodium album L. , Plantago major L. , Portulaca oleracea L. , Solanum nigrum L. , Sonchus oleraceus L. , Stellaria media L. and Taraxacum officinale W.) of edible wild plants grown in Australia were examined for their fatty acid composition by gas chromatography. The total fatty acid contents in the young leaves of these species ranged from 8.75 to 29.12 mg/g of dry matter, and were predominantly comprised of the alpha-linolenic acid (4.78 to 19.88 mg/g). These plants did not contain any of the longer-chain omega-3 fatty acids, namely eicosapentaenoic acid, docosahexaenoic acid or docosapentaenoic acid.     

Finite element analysis of heat conduction in lasagna during thermal processing

Conductive heat transfer in lasagna during heating was computed by means of the finite element method. the thermophysical properties of the pasta and tomato sauce layers were derived from their chemical composition. Simulation results obtained with an axisymmetric 2D finite element mesh without the typical layer structure differed only slightly from results obtained with a 1D mesh in which the typical layer structure was incorporated. the predicted temperatures at various locations in the lasagna agreed well with experimentally determined temperatures during heating in a commercial medium scale oven under various operating conditions. It is shown that the temperature dependency of the thermophysical parameters has only a minor influence on the temperature near the centre of the lasagna. However, when the lasagna surface is subjected to temperatures above its boiling temperature, incorporation of the water/vapour phase change in the specific heat capacity can avoid over-estimation of the volume averaged temperature and process value.     

INFUSION OF CITRUS FRUIT PEEL BY LOW-PRESSURE FLUID INJECTION USING A SELF-SEALING NOZZLE

Removal of citrus peel is facilitated by infusion of water into the peel. A low-pressure fluid injection (LPFI) technique was developed to improve infusion efficacy. Water under low pressure (340 to 1700 mbar) was forced through an opening formed near the equator of the fruit. The water spread uniformly throughout the albedo before exiting through an additional hole made on the opposite side of the fruit. Increasing pressure and flow rate reduced injection time, but excessive levels created high internal pressures that occasionally caused peels to rupture. A rubber bulb with an orifice at the tip created a nozzle that produced a water-tight interface with the fruit during injection.     

A DIGITAL POSITION INDICATOR FOR THE INSTRON CROSSHEAD1

An accessory for indicating the position of the crosshead of the Instron Universal Testing Machine relative to the machine frame is described. A linear displacement transducer allows the crosshead position to be selected within 0.35 mm or returned to a previous setting within at least 0.0083 mm. This facilitates accurate crosshead positioning and reproduction of the settings for different tests. The transducer also has the potential to indicate or record test sample deformations.     

A VALIDITY AND RELIABILITY STUDY OF THE PROCESS-REACTIVE SELF-REPORT SCALE.

EXAMINED THE RELIABILITY, INTERNAL CONSISTENCY, AND CONCURRENT VALIDITY OF ULLMANN AND GIOVANNONI'S PROCESS-REACTIVE SELF-REPORT SCALE. THE STABILITY COEFFICIENT WAS COMPARABLE TO THE PREVIOUSLY REPORTED SPLIT-HALF RELIABILITY OF THE SELF-REPORT AND THE PRESENTLY FOUND BETWEEN-RATER RELIABILITY OF THE PHILLIPS PREMORBID ADJUSTMENT SCALE. THE INTERNAL CONSISTENCY COEFFICIENTS OF EACH ITEM WERE COMPARABLE TO THOSE PREVIOUSLY REPORTED IN THE DEVELOPMENT OF THE SCALE. THE CONCURRENT VALIDITY COEFFICIENT WAS INSIGNIFICANT WHEN COMPARING THE SELF-REPORT TO PREMORBID ADJUSTMENT SCALE. THIS MAY BE DUE TO THE SAMPLE BEING CHRONIC AND THE PATIENT ATTEMPTING TO PRESENT AN IMAGE OF BEING "ILL" RATHER THAN IN REPORTING PAST SOCIAL HISTORY. RESULTS CAUTIONED AGAINST USING BOTH INSTRUMENTS INTERCHANGEABLY UNTIL FURTHER WORK CAN EXAMINE THE POSSIBLE DISTORTION IN THE SELF-REPORT OF SPECIFIC PATIENT GROUPS. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

A NEURAL NETWORK APPROACH TO ON-LINE IDENTIFICATION OF NONLINEAR SYSTEMS

Research investigating neural identification of dynamic systems has concentrated on off-line techniques. To be suitable for adaptive process control, on-line algorithms must be developed. This study investigates a modified back-propagation technique to achieve practical on-line capability. A technique denoted history-stack enhancement greatly improves the identification performance of the neural network. As a demonstration, a composite system of formidable but realistic nonlinear components was constructed and used to compare identification techniques including a recursive linear estimator and die new neural method. The results show that on-line neural identification is feasible for a wide range of formidable nonlinear characteristics individually found in industrial processes. Although performance is slower than with linear identification, the asymptotic accuracy of the neural technique is better for the nonlinear system being identified.     

ARE FUZZY SYSTEMS UNIVERSAL APPROXIMATORS?

Abstract

This paper is a critical reflection on various results in lileralure claiming that fuzzy systems are universal approximators. For this purpose the most specific features of fuzzy systems are outlined and it is discussed to which extent they are incorporated in the formal definition of a Tuzzy system in literature. It is argued that fuzzy systems can only be universal approximators in a rather reduced sense where some crucial features are neglected. The goal is to give an impulse to investigate more adequate mathematical concepts of a fuzzy system thai also take into account such features as transparency and linguistic interpretability.