Regularization and complexity control in feed-forward networks

In this paper we consider four alternative approaches to complexity control in feed-forward networks based respectively on architecture selection, regularization, early stopping, and training with noise. We show that there are close similarities between these approaches and we argue that, for most practical applications, the technique of regularization should be the method of choice.

Topographic mappings and feed-forward neural networks

This thesis is a study of the generation of topographic mappings - dimension reducing transformations of data that preserve some element of geometric structure - with feed-forward neural networks. As an alternative to established methods, a transformational variant of Sammon's method is proposed, where the projection is effected by a radial basis function neural network. This approach is related to the statistical field of multidimensional scaling, and from that the concept of a 'subjective metric' is defined, which permits the exploitation of additional prior knowledge concerning the data in the mapping process. This then enables the generation of more appropriate feature spaces for the purposes of enhanced visualisation or subsequent classification. A comparison with established methods for feature extraction is given for data taken from the 1992 Research Assessment Exercise for higher educational institutions in the United Kingdom. This is a difficult high-dimensional dataset, and illustrates well the benefit of the new topographic technique. A generalisation of the proposed model is considered for implementation of the classical multidimensional scaling (¸mds}) routine. This is related to Oja's principal subspace neural network, whose learning rule is shown to descend the error surface of the proposed ¸mds model. Some of the technical issues concerning the design and training of topographic neural networks are investigated. It is shown that neural network models can be less sensitive to entrapment in the sub-optimal global minima that badly affect the standard Sammon algorithm, and tend to exhibit good generalisation as a result of implicit weight decay in the training process. It is further argued that for ideal structure retention, the network transformation should be perfectly smooth for all inter-data directions in input space. Finally, there is a critique of optimisation techniques for topographic mappings, and a new training algorithm is proposed. A convergence proof is given, and the method is shown to produce lower-error mappings more rapidly than previous algorithms.

Combining the wavelet transform and forecasting models to predict gas forward prices

This paper presents a forecasting technique for forward energy prices, one day ahead. This technique combines a wavelet transform and forecasting models such as multi- layer perceptron, linear regression or GARCH. These techniques are applied to real data from the UK gas markets to evaluate their performance. The results show that the forecasting accuracy is improved significantly by using the wavelet transform. The methodology can be also applied to forecasting market clearing prices and electricity/gas loads.

Chain-store pricing and the structure of retail markets

This paper investigates competition between chain-stores and independents in the UK opticians' industry, using the relationship between the number of outlets present in a local market and the market size. Chain-stores are shown to have a significant effect on local market competition. In addition, the empirical approach is extended to allow inferences on the nature and extent of product differentiation. The results are broadly consistent with a model of vertical product differentiation in which chain-stores adopt national pricing strategies. The evidence suggests that the nature of competition between independent retailers depends on whether a chain-store is present.

Effects of yellow filters on visual acuity, contrast sensitivity and reading under conditions of forward light scatter

Background Yellow filters are sometimes recommended to people with low vision. Our aim was investigate the effects of three commercial yellow filters on visual acuity and contrast sensitivity (with and without glare) and reading (without glare) under conditions of forward light scatter (FLS). Method Fifty-five healthy subjects were assessed with Corning Photochromic Filters (CPFs) 450, 511 and 527 and a filter producing FLS. The effects on log MAR visual acuity, Pelli–Robson contrast sensitivity with and without glare, and reading (measured with MNRead charts) without glare were determined. Results Statistically significant differences were found between the overall effect of glare and between CPFs for visual acuity and contrast sensitivity. A gradual decline in visual acuity, contrast sensitivity and reading with increasing CPF absorption was noted. Conclusion Effects of CPF450, 511, 527 on visual acuity, contrast sensitivity and reading under conditions of FLS were negative but not clinically significant.

Drawing association rules between purchases and in-store behavior:an extension of the market basket analysis

Shopping behavior is often exclusively studied through consumer purchases, since they are an easily measurable ouput. Still, the observation of in-store physical behavior (paths, moves and actions) is crucial, as is the quantification of its impact on purchases. Using an innovative PDA tool to precisely record and time stamp consumer’s moves and gestures, we extend the classical Market Basket Analysis (MBA) by integrating this new kind of information. We draw associations not only from purchases but also from in-store consumer moves and actions. We compare results of our new method with classical MBA results and show a significant improvement.

Drawing association rules between purchases and in-store behaviour:an extension of the market basket analysis

Shopping behavior is often exclusively studied through consumer purchases, since they are an easily measurable ouput. Still, the observation of in-store physical behavior (path, moves and actions) is crucial, as is the quantification of its impact on purchases. Using an innovative PDA tool to precisely record and time stamp consumers' moves and actions, we extend the classical Market Basket Analysis (MBA) by integrating this new information: associations between product categories are measured not only from purchases but also from consumer physical behavior. We compare results of our new method with classical MBA results and show a significant improvement.

Promoting scholarship - the way forward:learning and teaching research in a complex environment – a typology

Scholarship in learning and teaching.

Leadership propensity and sales performance among sales personnel and managers in a speciality retail store setting

We propose that specialty store managers, as well as outside sales personnel attached to the store, have selling responsibilities. In addition, we propose that sales personnel, as well as store managers, should have a propensity for leadership, which reflects an individual's enduring disposition to exhibit leadership within the context of his or her organizational roles. In two studies, we develop a new individual difference measure of propensity to lead and investigate its nomological validity within a specialty retail store environment. As predicted, leadership propensity was predictive of self-rated sales performance and a proclivity to identify prospects through cold calls to close sales, to reveal customer orientation, and to exhibit organizational citizenship behavior. We found that propensity to lead did not differ between salespeople and retail store managers, but we found that the respondent's role moderated the relationship between propensity to lead and supervisor performance ratings. Study limitations and managerial implications of this heretofore unidentified trait of salespeople are discussed.

Evaluation of a heating system using a heat pump and heat store

DUE TO COPYRIGHT RESTRICTIONS ONLY AVAILABLE FOR CONSULTATION AT ASTON UNIVERSITY LIBRARY AND INFORMATION SERVICES WITH PRIOR ARRANGEMENT

Feed-forward neural networks and topographic mappings for exploratory data analysis

A recent novel approach to the visualisation and analysis of datasets, and one which is particularly applicable to those of a high dimension, is discussed in the context of real applications. A feed-forward neural network is utilised to effect a topographic, structure-preserving, dimension-reducing transformation of the data, with an additional facility to incorporate different degrees of associated subjective information. The properties of this transformation are illustrated on synthetic and real datasets, including the 1992 UK Research Assessment Exercise for funding in higher education. The method is compared and contrasted to established techniques for feature extraction, and related to topographic mappings, the Sammon projection and the statistical field of multidimensional scaling.