Previous research that has evaluated the effectiveness of personality variables for predicting work performance has predominantly relied on methods designed to detect simple relationships. The research reported in this thesis employed artificial neural networks ??? a method that is capable of capturing complex nonlinear and configural relationships among variables ??? and the findings were compared to those obtained by the more traditional method of linear regression. Six datasets that comprise a range of occupations, personality inventories, and work performance measures were used as the basis of the analyses. A series of studies were conducted to compare the predictive performance of prediction equations that a) were developed using either artificial neural networks or linear regression, and b) differed with respect to the type and number of personality variables that were used as predictors of work performance. Studies 1 and 2 compared the two methods using individual personality variables that assess the broad constructs of the five-factor model of personality. Studies 3 and 4 used combinations of these broad variables as the predictors. Study 5 employed narrow personality variables that assess specific facets of the broad constructs. Additional methodological contributions include the use of a resampling procedure, the use of multiple measures of predictive performance, and the comparison of two procedures for developing neural networks. Across the studies, it was generally found that the neural networks were rarely able to outperform the simpler linear regression equations, and this was attributed to the lack of reliable nonlinearity and configurality in personality-work performance relationships. However, the neural networks were able to outperform linear regression in the few instances where there was some independent evidence of nonlinear or configural relationships. Consequently, although the findings do not support the usefulness of neural networks for specifically improving the effectiveness of personality variables as predictors of work performance, in a broader sense they provide some grounds for optimism for organisational researchers interested in applying this method to investigate and exploit complex relationships among variables.

The aim of the thesis is to develop a model-based control strategy, namely, the Model Predictive Control (MPC) method, for robot position control using artificial neural networks. MPC is primarily developed for process control. Therefore its application in robot control has been less reported. In addition, conventional MPC uses linear model of the system for prediction which leads to inaccuracy for highly non-linear systems, such as robot. In this thesis a simulation model of a modified PUMA robot is constructed. This model is built using both MATLAB/SIMULINK and FORTRAN languages. In this model, the full robot dynamics is used together with the realistic factors, such as the actuator effects and the gear backlash, to represent the real system accurately. All simulations throughout this thesis are carried out on this model. A model predictive control strategy for robot trajectory tracking is also introduced in this thesis. The feasibility of the proposed MPC control method is studied based on a perfect prediction model, a model with uncertainties, and when the frequency band of the MPC controller is limited. Furthermore, a new method of using neural networks for robot dynamics modelling is introduced. This method is developed on the basis of a numerical differential technique that eliminates the explicit requirement of robot joint accelerations. Therefore, this method can be easily implemented on physical systems. As the measurements of the robot joint positions, velocities, and torques collected from operating the robot can be used to train the neural network, a more accurate dynamic model can be obtained. Finally, the MPC control method and the neural network model are combined together to form a neural network based MPC controller. The validity of this method is verified by using simulation on the simulated robot system / Master of Engineering (Hons)

This research investigates the use of conducting polymer coatings on platinum (Pt) electrodes for use in neuroprostheses. Conducting polymers aim to provide an environment conducive to neurite outgrowth and attachment at the electrode sites, producing intimate contact between neural cells and stimulating electrodes. Conducting polymers were electropolymerised onto model Pt electrodes. Conventional polymers polypyrrole (PPy) and poly-3,4-ethylenedioxythiphene (PEDOT) doped with polystyrenesulfonate (PSS) and para-toluenesulfonate (pTS)were investigated. Improvement of material properties was assessed through the layering of polymers with multi-walled carbon nanotubes (MWNTs). The ability to incorporate cell attachment bioactivity into polymers was examined through the doping of PEDOT with anionic laminin peptides DCDPGYIGSR and DEDEDYFQRYLI. Finally, nerve growth factor (NGF), was entrapped in PEDOT during polymerisation and tested for neurite outgrowth bioactivity against the PC12 cell line. Each polymer modification was assessed for electrical performance over multiple reduction-oxidation cycles, conductivity and impedance spectroscopy, mechanical adherence and hardness, and biological response. Scanning electron microscopy was used to visualise film topography and x-ray photon spectroscopy was employed to examine chemical constitution of the polymers. For application of electrode coatings to neural prostheses, optimal bioactive conducting polymer PEDOT/pTS/NGF was deposited on electrode arrays intended for implantation. PC12s were used to assess the bioactivity of NGF functionalised PEDOT when electrode size was micronised. Flexibility of the design was tested by tailoring PEDOT bioactivity for the cloned retinal ganglion cell, RGC-5, differentiated via staurasporine. It was established that PEDOT films had superior electrical and cell growth characteristics, but only PPy was able to benefit from incorporation of MWNTs. Bioactive polymers were produced through inclusion of both laminin peptides and NGF, but the optimum film constitution was found to be PEDOT doped with pTS with NGF entrapped during electrodeposition. Application of this polymer to an implant device was confirmed through positive neurite outgrowth on vision prosthesis electrode arrays. The design was shown to be flexible when tailored for RGC-5s, with differentiation occurring on both PEDOT/pTS and PEDOT/DEDEDYFQRYLI. Conducting polymers demonstrate the potential to improve electrode-cell interactions. Future work will focus on the effect of electrical stimulation and design of bioactive polymers with improved cell attachment properties.

This thesis explores the relationship between two classification models: decision trees and multilayer perceptrons. Decision trees carve up databases into box-shaped regions, and make predictions based on the majority class in each box. They are quick to build and relatively easy to interpret. Multilayer perceptrons (MLPs) are often more accurate than decision trees, because they are able to use soft, curved, arbitrarily oriented decision boundaries. Unfortunately MLPs typically require a great deal of effort to determine a good number and arrangement of neural units, and then require many passes through the database to determine a good set of connection weights. The cost of creating and training an MLP is thus hundreds of times greater than the cost of creating a decision tree, for perhaps only a small gain in accuracy. The following scheme is proposed for reducing the computational cost of creating and training MLPs. First, build and prune a decision tree to generate prior knowledge of the database. Then, use that knowledge to determine the initial architecture and connection weights of an MLP. Finally, use a training algorithm to refine the knowledge now embedded in the MLP. This scheme has two potential advantages: a suitable neural network architecture is determined very quickly, and training should require far fewer passes through the data. In this thesis, new algorithms for initialising MLPs from decision trees are developed. The algorithms require just one traversal of a decision tree, and produce four-layer MLPs with the same number of hidden units as there are nodes in the tree. The resulting MLPs can be shown to reach a state more accurate than the decision trees that initialised them, in fewer training epochs than a standard MLP. Employing this approach typically results in MLPs that are just as accurate as standard MLPs, and an order of magnitude cheaper to train.

Thesis (M.S.)--Oregon Graduate Center, 1988.

Typescript (photocopy)

<p>In this work new artificial learning and innate control mechanisms are proposed for application</p><p>in autonomous behavioral systems for mobile robots. An autonomous system (for mobile robots)</p><p>existent in the literature is enhanced with respect to its capacity of exploring the environment and</p><p>avoiding risky configurations (that lead to collisions with obstacles even after learning). The</p><p>particular autonomous system is based on modular hierarchical neural networks. Initially,the</p><p>autonomous system does not have any knowledge suitable for exploring the environment (and</p><p>capture targets œ foraging). After a period of learning,the system generates efficientobstacle</p><p>avoid ance and target seeking behaviors. Two particular deficiencies of the forme rautonomous</p><p>system (tendency to generate unsuitable cyclic trajectories and ineffectiveness in risky</p><p>configurations) are discussed and the new learning and controltechniques (applied to the</p><p>autonomous system) are verified through simulations. It is shown the effectiveness of the</p><p>proposals: theautonomous system is able to detect unsuitable behaviors (cyclic trajectories) and</p><p>decrease their probability of appearance in the future and the number of collisions in risky</p><p>situations is significantly decreased. Experiments also consider maze environments (with targets</p><p>distant from each other) and dynamic environments (with moving objects).</p>

This thesis describes an investigation of retinal directional selectivity. We show intracellular (whole-cell patch) recordings in turtle retina which indicate that this computation occurs prior to the ganglion cell, and we describe a pre-ganglionic circuit model to account for this and other findings which places the non-linear spatio-temporal filter at individual, oriented amacrine cell dendrites. The key non-linearity is provided by interactions between excitatory and inhibitory synaptic inputs onto the dendrites, and their distal tips provide directionally selective excitatory outputs onto ganglion cells. Detailed simulations of putative cells support this model, given reasonable parameter constraints. The performance of the model also suggests that this computational substructure may be relevant within the dendritic trees of CNS neurons in general.

The computation of a piecewise smooth function that approximates a finite set of data points may be decomposed into two decoupled tasks: first, the computation of the locally smooth models, and hence, the segmentation of the data into classes that consist on the sets of points best approximated by each model, and second, the computation of the normalized discriminant functions for each induced class. The approximating function may then be computed as the optimal estimator with respect to this measure field. We give an efficient procedure for effecting both computations, and for the determination of the optimal number of components.

M.S. / Computer Science & Engineering / The Oregon Graduate Center's Cognitive Architecture Project (CAP) is developing a flexible architecture to evaluate and implement several types of neural networks. Wafer-scale integrated silicon is the targeted technology, allowing higher density and larger networks to be implemented more cheaply than with discrete components. The large size of networks implemented in wafer-scale technology makes it difficult to assess the effects of manufacturing faults on network behavior. Since neural networks degrade gracefully in the presence of faults, and since in larger networks faults tend to interact with each other, it is difficult to determine these effects analytically. This paper discusses a program, FltSim, that simulates wafer manufacturing faults. By building an abstract model of the CAP architecture, the effects of these manufacturing faults can be determined long before proceeding to implementation. In addition, the effects of architectural design trade-offs can be studied during the design process.