Visualization of learning in multilayer perceptron networks using principal component analysis

This paper is concerned with the use of scientific visualization methods for the analysis of feedforward neural networks (NNs). Inevitably, the kinds of data associated with the design and implementation of neural networks are of very high dimensionality, presenting a major challenge for visualization. A method is described using the well-known statistical technique of principal component analysis (PCA). This is found to be an effective and useful method of visualizing the learning trajectories of many learning algorithms such as back-propagation and can also be used to provide insight into the learning process and the nature of the error surface.

Supervisory control of wastewater treatment plants by combining principal component analysis and fuzzy c-means clustering

In this paper a methodology for integrated multivariate monitoring and control of biological wastewater treatment plants during extreme events is presented. To monitor the process, on-line dynamic principal component analysis (PCA) is performed on the process data to extract the principal components that represent the underlying mechanisms of the process. Fuzzy c-means (FCM) clustering is used to classify the operational state. Performing clustering on scores from PCA solves computational problems as well as increases robustness due to noise attenuation. The class-membership information from FCM is used to derive adequate control set points for the local control loops. The methodology is illustrated by a simulation study of a biological wastewater treatment plant, on which disturbances of various types are imposed. The results show that the methodology can be used to determine and co-ordinate control actions in order to shift the control objective and improve the effluent quality.

Drying process of microcrystalline cellulose studied by attenuated total reflection IR spectroscopy with two-dimensional correlation spectroscopy and principal component analysis

Molecular interactions between microcrystalline cellulose (MCC) and water were investigated by attenuated total reflection infrared (ATR/IR) spectroscopy. Moisture-content-dependent IR spectra during a drying process of wet MCC were measured. In order to distinguish overlapping O–H stretching bands arising from both cellulose and water, principal component analysis (PCA) and, generalized two-dimensional correlation spectroscopy (2DCOS) and second derivative analysis were applied to the obtained spectra. Four typical drying stages were clearly separated by PCA, and spectral variations in each stage were analyzed by 2DCOS. In the drying time range of 0–41 min, a decrease in the broad band around 3390 cm−1 was observed, indicating that bulk water was evaporated. In the drying time range of 49–195 min, decreases in the bands at 3412, 3344 and 3286 cm−1 assigned to the O6H6cdots, three dots, centeredO3′ interchain hydrogen bonds (H-bonds), the O3H3cdots, three dots, centeredO5 intrachain H-bonds and the H-bonds in Iβ phase in MCC, respectively, were observed. The result of the second derivative analysis suggests that water molecules mainly interact with the O6H6cdots, three dots, centeredO3′ interchain H-bonds. Thus, the H-bonding network in MCC is stabilized by H-bonds between OH groups constructing O6H6cdots, three dots, centeredO3′ interchain H-bonds and water, and the removal of the water molecules induces changes in the H-bonding network in MCC.

Non-linear principal components analysis: an alternative method for finding patterns in environmental data

The main purpose of this article is to gain an insight into the relationships between variables describing the environmental conditions of the Far Northern section of the Great Barrier Reef, Australia, Several of the variables describing these conditions had different measurement levels and often they had non-linear relationships. Using non-linear principal component analysis, it was possible to acquire an insight into these relationships. Furthermore. three geographical areas with unique environmental characteristics could be identified. Copyright (c) 2005 John Wiley & Sons, Ltd.

Distribution of major, minor, and trace metals in lake environments of Antarctica

The concentrations of major, minor and trace metals were measured in water samples collected from five shallow Antarctic lakes (Carezza, Edmonson Point (No 14 and 15a), Inexpressible Island and Tarn Flat) found in Terra Nova Bay (northern Victoria Land, Antarctica) during the Italian Expeditions of 1993-2001. The total concentrations of a large suite of elements (Al, As, Ba, Ca, Cd, Ce, Co, Cr, Cs, Cu, Fe, Ga, Gd, K, La, Li, Mg, Mn, Mo, Na, Nd, Ni, Pb, Pr, Rb, Sc, Si, Sr, Ta, Ti, U, V, Y, W, Zn and Zr) were determined using spectroscopic techniques (ICP-AES, GF-AAS and ICP-MS). The results are similar to those obtained for the freshwater lakes of the Larsemann Hills, East Antarctica, and for the McMurdo Dry Valleys. Principal Component Analysis (PCA) and Cluster Analysis (CA) were performed to identify groups of samples with similar characteristics and to find correlations between the variables. The variability observed within the water samples is closely connected to the sea spray input; hence, it is primarily a consequence of geographical and meteorological factors, such as distance from the ocean and time of year. The trace element levels, in particular those of heavy metals, are very low, suggesting an origin from natural sources rather than from anthropogenic contamination.

Using biplots to interpret gene expression patterns in plants

Sum: Plant biologists in fields of ecology, evolution, genetics and breeding frequently use multivariate methods. This paper illustrates Principal Component Analysis (PCA) and Gabriel's biplot as applied to microarray expression data from plant pathology experiments. Availability: An example program in the publicly distributed statistical language R is available from the web site (www.tpp.uq.edu.au) and by e-mail from the contact. Contact: scott.chapman@csiro.au.

A comparison between three and five factor models of Pakistani personality data

The Eysenck Personality Questionnaire-Revised (EPQ-R), the Eysenck Personality Profiler Short Version (EPP-S), and the Big Five Inventory (BFI-V4a) were administered to 135 postgraduate students of business in Pakistan. Whilst Extraversion and Neuroticism scales from the three questionnaires were highly correlated, it was found that Agreeableness was most highly correlated with Psychoticism in the EPQ-R and Conscientiousness was most highly correlated with Psychoticism in the EPP-S. Principal component analyses with varimax rotation were carried out. The analyses generally suggested that the five factor model rather than the three-factor model was more robust and better for interpretation of all the higher order scales of the EPQ-R, EPP-S, and BFI-V4a in the Pakistani data. Results show that the superiority of the five factor solution results from the inclusion of a broader variety of personality scales in the input data, whereas Eysenck's three factor solution seems to be best when a less complete but possibly more important set of variables are input. (C) 2001 Elsevier Science Ltd. All rights reserved.

Carapace dentition patterns, morphometrics and allozyme differentiation amongst two toothed freshwater crab species (Potamonautes warreni and P-unispinus) (Decapoda : Brachyura : Potamonautidae) from river systems in South Africa

The taxonomic relationship between two toothed South African river crabs, Potamonautes warreni and P. unispinus, is unclear. The problem stems from the widespread variation in carapace dentition patterns amongst P. warreni individuals over its biogeographic range, where single toothed individuals may appear similar in carapace morphology to P. unispinus. Ten populations of P. warreni and 18 populations of P. unispinus were collected and the morphometric and genetic differentiation between the two taxa quantified. Patterns of morphometric and genetic variation were examined using multivariate statistics and protein gel electrophoresis, respectively. Principal component analyses of carapace characters showed that the two species are morphologically indistinguishable. However, discriminate functions analyses and additional statistical results corroborate the morphological distinction between the two taxa. Allozyme electrophoresis of 17 protein coding loci, indicated a close genetic similarity between the two species (I = 0.92). A fixed allelic difference at one locus (LT-2) and extensive genetic variability at another locus (PGM-1) indicate that two gene pools are present and that the two taxa are genetically isolated. Intraspecific genetic I values for both species were > 0.97 and indicated no apparent genetic structuring on a micro or macro-geographic scale. The variation in carapace dentition among P. warreni populations possesses no genetic basis and may possibly toe the product of ecogenesis. The value of dentition patterns in the systematics of river crabs is discussed. Dentition patterns among river crab species appear to be conserved and reliable as species specific diagnostic markers, but should ideally be used in combination with other morphological data sets and genetic evidence.

Treatment outcome in individuals with chronic pain: is the Pain Stages of Change Questionnaire (PSOCQ) a useful tool? Jenny StrongCorresponding Author Contact Information, E-mail The Corresponding Author, a, Kym Westburya, Glen Smithb, Ian McKenziec and William Ryan

The efficacy of psychological treatments emphasising a self-management approach to chronic pain has been demonstrated by substantial empirical research. Nevertheless, high drop-out and relapse rates and low or unsuccessful engagement in self-management pain rehabilitation programs have prompted the suggestion that people vary in their readiness to adopt a self-management approach to their pain. The Pain Stages of Change Questionnaire (PSOCQ) was developed to assess a patient's readiness to adopt a self-management approach to their chronic pain. Preliminary evidence has supported the PSOCQ's psychometric properties. The current study was designed to further examine the psychometric properties of the PSOCQ, including its reliability, factorial structure and predictive validity. A total of 107 patients with an average age of 36.2 years (SD = 10.63) attending a multi-disciplinary pain management program completed the PSOCQ, the Pain Self-Efficacy Questionnaire (PSEQ) and the West Haven-Yale Multidimensional Pain Inventory (WHYMPI) pre-admission and at discharge from the program. Initial data analysis found inadequate internal consistencies of the precontemplation and action scales of the PSOCQ and a high correlation (r = 0.66, P < 0.01) between the action and maintenance scales. Principal component analysis supported a two-factor structure: 'Contemplation' and 'Engagement'. Subsequent analyses revealed that the PSEQ was a better predictor of treatment outcome than the PSOCQ scales. Discussion centres upon the utility of the PSOCQ in a clinical pain setting in light of the above findings, and a need for further research. (C) 2002 International Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved.

A framework for extreme-event control in wastewater treatment

In this paper an approach to extreme event control in wastewater treatment plant operation by use of automatic supervisory control is discussed. The framework presented is based on the fact that different operational conditions manifest themselves as clusters in a multivariate measurement space. These clusters are identified and linked to specific and corresponding events by use of principal component analysis and fuzzy c-means clustering. A reduced system model is assigned to each type of extreme event and used to calculate appropriate local controller set points. In earlier work we have shown that this approach is applicable to wastewater treatment control using look-up tables to determine current set points. In this work we focus on the automatic determination of appropriate set points by use of steady state and dynamic predictions. The performance of a relatively simple steady-state supervisory controller is compared with that of a model predictive supervisory controller. Also, a look-up table approach is included in the comparison, as it provides a simple and robust alternative to the steady-state and model predictive controllers, The methodology is illustrated in a simulation study.

Genetic variability in cultivated common bean beyond the two major gene pools

It is generally accepted that two major gene pools exist in cultivated common bean (Phaseolus vulgaris L.), a Middle American and an Andean one. Some evidence, based on unique phaseolin morphotypes and AFLP analysis, suggests that at least one more gene pool exists in cultivated common bean. To investigate this hypothesis, 1072 accessions from a common bean core collection from the primary centres of origin, held at CIAT, were investigated. Various agronomic and morphological attributes (14 categorical and 11 quantitative) were measured. Multivariate analyses, consisting of homogeneity analysis and clustering for categorical data, clustering and ordination techniques for quantitative data and nonlinear principal component analysis for mixed data, were undertaken. The results of most analyses supported the existence of the two major gene pools. However, the analysis of categorical data of protein types showed an additional minor gene pool. The minor gene pool is designated North Andean and includes phaseolin types CH, S and T; lectin types 312, Pr, B and K; and mostly A5, A6 and A4 types alpha-amylase inhibitor. Analysis of the combined categorical data of protein types and some plant categorical data also suggested that some other germplasm with C type phaseolin are distinguished from the major gene pools.

Spatial and temporal patterns in Australian wheat yield and their relationship with ENSO

Spatial and temporal variability in wheat production in Australia is dominated by rainfall occurrence. The length of historical production records is inadequate, however, to analyse spatial and temporal patterns conclusively. In this study we used modelling and simulation to identify key spatial patterns in Australian wheat yield, identify groups of years in the historical record in which spatial patterns were similar, and examine association of those wheat yield year groups with indicators of the El Nino Southern Oscillation (ENSO). A simple stress index model was trained on 19 years of Australian Bureau of Statistics shire yield data (1975-93). The model was then used to simulate shire yield from 1901 to 1999 for all wheat-producing shires. Principal components analysis was used to determine the dominating spatial relationships in wheat yield among shires. Six major components of spatial variability were found. Five of these represented near spatially independent zones across the Australian wheatbelt that demonstrated coherent temporal (annual) variability in wheat yield. A second orthogonal component was required to explain the temporal variation in New South Wales. The principal component scores were used to identify high- and low-yielding years in each zone. Year type groupings identified in this way were tested for association with indicators of ENSO. Significant associations were found for all zones in the Australian wheatbelt. Associations were as strong or stronger when ENSO indicators preceding the wheat season (April-May phases of the Southern Oscillation Index) were used rather than indicators based on classification during the wheat season. Although this association suggests an obvious role for seasonal climate forecasting in national wheat crop forecasting, the discriminatory power of the ENSO indicators, although significant, was not strong. By examining the historical years forming the wheat yield analog sets within each zone, it may be possible to identify novel climate system or ocean-atmosphere features that may be causal and, hence, most useful in improving seasonal forecasting schemes.

Are wing size, wing shape and asymmetry related to field fitness of Trichogramma egg parasitoids?

The effects of wing shape, wing size, and fluctuating asymmetry in these measures oil the field fitness of T. nr. brassicae and T. pretiosum were investigated. Trichogramma wasps mass-reared on eggs of the factitious host Sitotroga cerealella were released in tomato paddocks and those females ovipositing on Helicoverpo spp. eggs were recaptured. Comparisons of the recaptured group with a sample from the release population were used to assess fitness. Wing data were obtained by positioning landmarks on mounted forewings. Size was then measured as the centroid size computed from landmark distances, while Procrustes analysis followed by principal component analysis was used to assess wing shape. Similar findings were obtained for both Trichogramma species: fitness of wasps was strongly related to wing size and some shape dimensions, but not to the asymmetries of these measures. Wasps which performed well in the field had larger wings and a different wing shape compared to wasps from the mass reared population. Both size and the shape dimensions were linearly associated with fitness although there was also some evidence for non-linear selection on shape. The results suggest that wing shape and wing size are reliable predictors of field fitness for these Trichogramma wasps.

On measuring quality of a probabilistic commodity forecast for a system that incorporates seasonal climate forecasts

Regional commodity forecasts are being used increasingly in agricultural industries to enhance their risk management and decision-making processes. These commodity forecasts are probabilistic in nature and are often integrated with a seasonal climate forecast system. The climate forecast system is based on a subset of analogue years drawn from the full climatological distribution. In this study we sought to measure forecast quality for such an integrated system. We investigated the quality of a commodity (i.e. wheat and sugar) forecast based on a subset of analogue years in relation to a standard reference forecast based on the full climatological set. We derived three key dimensions of forecast quality for such probabilistic forecasts: reliability, distribution shift, and change in dispersion. A measure of reliability was required to ensure no bias in the forecast distribution. This was assessed via the slope of the reliability plot, which was derived from examination of probability levels of forecasts and associated frequencies of realizations. The other two dimensions related to changes in features of the forecast distribution relative to the reference distribution. The relationship of 13 published accuracy/skill measures to these dimensions of forecast quality was assessed using principal component analysis in case studies of commodity forecasting using seasonal climate forecasting for the wheat and sugar industries in Australia. There were two orthogonal dimensions of forecast quality: one associated with distribution shift relative to the reference distribution and the other associated with relative distribution dispersion. Although the conventional quality measures aligned with these dimensions, none measured both adequately. We conclude that a multi-dimensional approach to assessment of forecast quality is required and that simple measures of reliability, distribution shift, and change in dispersion provide a means for such assessment. The analysis presented was also relevant to measuring quality of probabilistic seasonal climate forecasting systems. The importance of retaining a focus on the probabilistic nature of the forecast and avoiding simplifying, but erroneous, distortions was discussed in relation to applying this new forecast quality assessment paradigm to seasonal climate forecasts. Copyright (K) 2003 Royal Meteorological Society.