A stable MCA learning algorithm

Minor component analysis (MCA) is an important statistical tool for signal processing and data analysis. Neural networks can be used to extract online minor component from input data. Compared with traditional algebraic  approaches, a neural network method has a lower computational complexity. Stability of neural networks learning algorithms is crucial to practical applications. In this paper, we propose a stable MCA neural networks learning algorithm, which has a more satisfactory numerical stability than some existing MCA algorithms. Dynamical behaviors of the proposed algorithm are analyzed via deterministic discrete time (DDT) method and the conditions are obtained to guarantee convergence. Simulations are carried out to illustrate the theoretical results achieved.

Texture recognition by using GLCM and various aggregation functions

We discuss the problem of texture recognition based on the grey level co-occurrence matrix (GLCM). We performed a number of numerical experiments to establish whether the accuracy of classification is optimal when GLCM entries are aggregated into standard metrics like contrast, dissimilarity, homogeneity, entropy, etc., and compared these metrics to several alternative aggregation methods.We conclude that k nearest neighbors classification based on raw GLCM entries typically works better than classification based on the standard metrics for noiseless data, that metrics based on principal component analysis inprove classification, and that a simple change from the arithmetic to quadratic mean in calculating the standard metrics also improves classification.

Ultraviolet plumage colors predict mate preferences in starlings

Avian plumage has long been used to test theories of sexual selection, with humans assessing the colors, However, many birds see in the ultraviolet (<400 nm), to which humans are blind, Consequently, it is important to know whether natural variation in UV reflectance from plumage functions in sexual signaling, We show that female starlings rank males differently when UV wavelengths are present or absent, Principal component analysis of approximate to 1300 reflectance spectra (300-700 nm) taken from sexually dimorphic plumage regions of males predicted preference under the UV+ treatment. Under UV- conditions, females ranked males in a different and nonrandom order, but plumage reflectance in the human visible spectrum did not predict choice, Natural variation in UV reflectance is thus important in avian mate assessment, and the prevailing light environment can have profound effects on observed mating preferences.

Retrospective application of remote sensing to the Tasmanian lakeland

This thesis describes the research undertaken for a degree of Master of Science in a retrospective study of airborne remotely sensed data registered in 1990 and 1993, and field captured data of aquatic humus concentrations for ~ 45 lakes in Tasmania. The aim was to investigate and describe the relationship between the remotely sensed data and the field data and to test the hypothesis that the remotely sensed data would establish further evidence of a limnological corridor of change running north-west to south- east. The airborne remotely sensed data consisted of data captured by the CSIRO Ocean Colour Scanner (OCS) and a newly developed Canadian scanner, a compact airborne spectrographic imager (CASI). The thesis investigates the relationship between the two kinds of data sources. The remotely sensed data was collected with the OCS scanner in 1990 (during one day) and with both the OCS and the CASI in 1993 (during three days). The OCS scanner registers data in 9 wavelength bands between 380 nm and 960 nm with a 10-20 nm bandwidth, and the CASI in 288 wavelength bands between 379.57 nm and 893.5 nm (ie. spectral mode) with a spectral resolution of 2.5 nm. The remotely sensed data were extracted from the original tapes with the help of the CSIRO and supplied software and digital sample areas (band value means) for each lake were subsequently extracted for data manipulation and statistical analysis. Field data was captured concurrently with the remotely sensed data in 1993 by lake hopping using a light aircraft with floats. The field data used for analysis with the remotely sensed data were the laboratory determined g440 values from the 1993 water samples collated with g440 values determined from earlier years. No spectro-radiometric data of the lakes, data of incoming irradiance or ancillary climatic data were captured during the remote sensing missions. The sections of the background chapter in the thesis provide a background to the research both in regards to remote sensing of water quality and the relationship between remotely sensed spectral data and water quality parameters, as well as a description of the Tasmanian lakes flown. The lakes were divided into four groups based on results from previous studies and optical parameters, especially aquatic humus concentrations as measured from field captured data. The four groups consist of the ‘green” clear water lakes mostly situated on the Central Plateau, the ‘brown” highly dystrophic lakes in western Tasmania, the ‘corridor” lakes situated along a corridor of change lying approximately between the two lines denoting the Jurassic edge and 1200 mm isohyet, and the ‘eastern, turbid” lakes make up the fourth group. The analytical part of the research work was mostly concerned with manipulating and analysing the CASI data because of its higher spectral resolution. The research explores methods to apply corrections to this data to reduce the disturbing effects of varying illumination and atmospheric conditions. Three different methods were attempted. In the first method two different standardisation formulas are applied to the data as well as ‘day correction” factors calculated from data from one of the lakes, Lake Rolleston, which had data captured for all three days of the remote sensing operations. The standardisation formulas were also applied to the OCS data. In second method an attempt to reduce the effects of the atmosphere was performed using spectro-radiometric captured in 1988 for one of the lakes flown, Great Lake. All the lake sample data were time normalised using general irradiance data obtained from the University of Tasmania and the sky portion as calculated from Great Lake upwelling irradiance data was then subtracted. The last method involved using two different band ratios to eliminate atmospheric effects. Statistical analysis was applied to the data resulting from the three methods to try to describe the relationship between the remotely sensed data and the field captured data. Discriminant analysis, cluster analysis and factor analysis using principal component analysis (pea) were applied to the remotely sensed data and the field data. The factor scores resulting from the pca were regressed against the field collated data of g440 as were the values resulting from last method. The results from the statistical analysis of the data from the first method show that the lakes group well (100%) against the predetermined groups using discriminant analysis applied to the remotely sensed CASI data. Most variance in the data are contained in the first factor resulting from pca regardless of data manipulation method. Regression of the factor scores against g440 field data show a strong non- linear relationship and a one-sided linear regression test is therefore considered an inappropriate analysis method to describe the dataset relationships. The research has shown that with the available data, correction and analysis methods, and within the scope of the Masters study, it was not possible to establish the relationships between the remotely sensed data and the field measured parameters as hoped. The main reason for this was the failure to retrieve remotely sensed lake signatures adequately corrected for atmospheric noise for comparison with the field data. This in turn is a result of the lack of detailed ancillary information needed to apply available established methods for noise reduction - to apply these methods we require field spectroradiometric measurements and environmental information of the varying conditions both within the study area and within the time frame of capture of the remotely sensed data.

Impact of stresses imposed on macroinvertebrate communities in two suburban streams

The aim of the project was to determine factors which explain the distribution of macroinvertebrates in two Melbourne streams both drastically affected by urbanisation. A detailed description is given of Dandenong Creek, flowing through the south-eastern suburbs, and Darebin Creek, in the northern suburbs, emphasising stream features likely, or known, to influence the drift and benthic fauna. Faunal sampling was carried out in Dandenong Creek from June 1992 until July 1993, and in Darebin Creek from February 1995 until March 1998. Physicochemical parameters were also recorded. The collected data, together with previously existing data, were analysed using multivariate analyses: non-metric multi-dimensional scaling (NMDS); analysis of similarities (ANOSIM); matching biotic and abiotic variables using BIOENV, and principal component analysis (PCA). Various biotic and diversity indices were calculated in an attempt to identify the major factors responsible for the failure of the fauna to recover from previously more seriously degraded water quality. The contribution of drift to the colonisation potential in Dandenong Creek appeared to be impacted by retarding basins, underground barrel-draining and channelization. Results also indicated that increased conductivity adversely affected the fauna in the lower reaches of Dandenong Creek. It was concluded that in Darebin Creek, high nutrient levels, as well as other pollutants, had resulted in low macroinvertebrate diversity in both the drift and benthos. If, as this study suggests, faunal diversity is a valid measure of stream health, the following factors need to be addressed for catchment-wide, stream management: lack of riparian zone vegetation (increasing bank erosion and making the benthic habitat unstable, with greater temperature variability); control of stormwater runoff (flow variability, increased conductivity, nutrient levels, sediment loads, sewage effluent, industrial discharges and heavy metals), and to modify retarding basins to increase stream continuity.

Genetic and environmental influences on variation in balance performance among female twin pairs aged 21-82 years

Genetic and environmental influences on variation in balance performance were measured in 93 monozygous and 83 dizygous female twin pairs aged 21–82 years (mean age, 50.5 years) in Melbourne, Australia, between 1999 and 2003. The authors administered clinical (Lord's Balance Test and Step Test) and laboratory tests of static and dynamic balance from the Chattecx Balance System with and without distractor tasks. The authors conducted factor analysis and estimated genetic and environmental variance components and heritability (defined as additive genetic variance as a proportion of all variance, after adjustment for age) using a multivariate normal model with the statistical package FISHER. Three factors were identified and adjusted for age. Heritability was 46% (standard error (SE), 9) for the "sensory balance tests" factor and 30% (SE, 9) for the "static and dynamic perturbations" factor. For both factors, the remaining variance was attributed to unique environmental effects. There was no evidence that genetic factors influenced variation in the "dynamic weight shift tests" factor, with environmental effects shared by twins accounting for 38% (SE, 7) of variance. Neither genetic nor environmental proportions of variance differed significantly between twin subgroups by age (≤50/>50 years). An age-related decline in performance measures was found across the whole sample. These results imply that balance impairments may have a heritable element.

Feature extraction for animal fiber identification

Fiber identification has been a very important task in many industries such as wool growing, textile processing, archaeology, histochernical engineering, and zoology. Over the years, animal fibers have been identified using physical and chemical approaches. Recently, objective identification of animal fibers has been developed based on the cuticular information of fibers. Effective and accurate extraction of representative features is essential to animal fiber identification and classification. In the current work, two different strategies are developed for this purpose. In the first method, explicit features are extracted using image processing. However, only implicit features are used in the second method with an unsupervised artificial neural network. It is found that the use of explicit features increases the accuracy of fiber identification but requires more effort on processing images and solid knowledge of what features are representative ones.

A unified learning algorithm to extract principal and minor components

Recently, many unified learning algorithms have been developed to solve the task of principal component analysis (PCA) and minor component analysis (MCA). These unified algorithms can be used to extract principal component and if altered simply by the sign, it can also serve as a minor component extractor. This is of practical significance in the implementations of algorithms. Convergence of the existing unified algorithms is guaranteed only under the condition that the learning rates of algorithms approach zero, which is impractical in many practical applications. In this paper, we propose a unified PCA & MCA algorithm with a constant learning rate, and derive the sufficient conditions to guarantee convergence via analyzing the discrete-time dynamics of the proposed algorithm. The achieved theoretical results lay a solid foundation for the applications of our proposed algorithm.

Impact of board composition and ownership on internationalisation strategy amongst small New Zealand firms

Internationalisation strategies are important for company expansion because New Zealand, with its four million people, has such a small market. Nonetheless, there mayor may not exist ;"agency costs" in the use of Outside Directors. Ownership patterns may also influence Internationalization Strategy. Using Binary Correlation, N-Way Cross-Tabulation, and Principal Component Analysis, we find evidence that Outside Directors have less influence on Internationalisation Strategy than Inside Directors. Family ownership also seems to have a greater association than non-family owned companies. Despite substantial limitations, the methods and models proposed seem to have some utility in examining the association of Internationalisation Strategy with Board Composition and Ownership Patterns.

Detection of vehicles with monolithic classifier vis-à-vis a boosted cascaded classifier

This paper describes the comparison of accuracy and performance of two machine learning approaches for visual object detection and tracking vehicles, from an on-road image sequence. The first is a neural network based approach. Where an algorithm of multi resolution technique based on Haar basis functions was used to obtain an image with different scales. Thereafter a classification was carried out with the multilayer feed forward neural network. Principle Component Analysis (PCA) technique was used as a dimension reduction technique to make the classification process much more efficient. The second approach is based on boosting which also yields very good detection rates. In general, boosting is one of the most important developments in classification methodology. It works by sequentially applying a classification algorithm to reweighed versions of the training data, followed by taking a weighted majority vote of the sequence of classifiers thus produced. For this work, a strong classifier was trained by the adaboost algorithm. The results of comparing the two methodologies visà-vis shows the effectiveness of the methods that have been used.

From rural beginnings to statewide roll-out : evaluation of facilitator training for a group-based diabetes prevention program

Objective: To evaluate the approach used to train facilitators for a large-scale group-based diabetes prevention program developed from a rural implementation research project.

Participants: Orientation day was attended by 224 health professionals; 188 submitted the self-learning task; 175 achieved the satisfactory standard for the self-learning task and attended the workshop; 156 completed the pre- and post-training questionnaires.

Main outcome measures: Two pre- and post-training scales were developed to assess knowledge and confidence in group-based diabetes prevention program facilitation. Principal component analysis found four factors for measuring training effectiveness: knowledge of diabetes prevention, knowledge of group facilitation, confidence to facilitate a group to improve health literacy and confidence in diabetes prevention program facilitation. Self-learning task scores, training discontinuation rates and satisfaction scores were also assessed.

Results: There was significant improvement in all four knowledge and confidence factors from pre- to post-training (P < 0.001). The self-learning task mean test score was 88.7/100 (SD = 7.7), and mean assignment score was 72.8/100 (SD = 16.1). Satisfaction with training scores were positive and 'previous training' interacted with 'change in knowledge of diabetes prevention program facilitation' but not with change in 'confidence to facilitate.'

Conclusions: The training program was effective when analysed by change in facilitator knowledge and confidence and the positive mean satisfaction score. Learning task scores suggest tasks were manageable and the requirement contributed to facilitator self-selection. Improvement in confidence scores in facilitating a group-based diabetes prevention program, irrespective of previous training and experience, show that program-specific skill development activities are necessary in curriculum design.

A neural network based human identification framework using ear images

This paper presents a framework that uses ear images for human identification. The framework makes use of Principal Component Analysis (PCA) for ear image feature extraction and Multilayer Feed Forward Neural Network for classification. Framework are proposed to improve recognition accuracy of human identification. The framework was tested on an ear image database to evaluate its reliability and recognition accuracy. The experimental results showed that our framework achieved higher stable recognition accuracy and over-performed other existing methods. The recognition accuracy stability and computation time with respect to different image sizes and factors were investigated thoroughly as well in the experiments.

The relationship between building adaptation and property attributes

Using a database of building adaptation and property attributes this research examined every adaptation event in Melbourne's CBD between 1998 and 2008. The importance of property attributes was derived using a principal component analysis and a weighted index of optimal decision making attributes was proposed; the Preliminary Assessment Adaptation Model (PAAM). The findings indicate the relationship between property attributes is more complex than hitherto held. Overall physical attributes were found to be more important than others such as economic, environmental, legal and social attributes; however physical attributes alone are not important and are closely related to other attributes.

Ecological correlates of the threat of extinction in neotropical bird species

Predicting the threat of extinction aids efficient distribution of conservation resources. This paper utilises a comparative macroecological approach to investigate the threat of extinction in Neotropical birds. Data on ecological variables for 1708 species are analysed using stepwise regression to produce minimum adequate models, first using raw species values and then using independent contrasts (to control for phylogenetic effects). The models differ, suggesting phylogeny has significant effects. The raw species analysis reveals that number of zoogeographical regions occupied, elevational range and utilisation of specialised microhabitats were negatively associated with threat, while minimum elevation and body mass were positively associated, whereas the independent contrasts analysis only identifies zoogeographical regions as important. Confining the analysis to the 582 species restricted to a single zoogeographical region reveals elevational range and number of habitats occupied to be negatively correlated with threat whether the analysis is based on the raw data or on independent contrasts. Analysis of four contrasting zoogeographical regions highlights regional variation in the models. In two Andean regions the threat of extinction declines as the elevation range across which the species occurs increases. In the presence of substantial human populations on high Andean plateaus, a species with a greater elevational range may be more likely to persist at some (relatively) unsettled altitudes. In Central South America, the strongest predictor of threat is minimum elevation of occurrence: species with a lower minimum are less threatened. The minimum elevation result suggests that lowland species experiencing an ecological limit to their minimum elevation (min. elevation >0 m) may be more at risk than those not experiencing such a limit (min. elevation = 0 m). Finally, in southern Amazonia, where there is little altitudinal variation, the only weak predictors of threat are body size, larger species being more threatened, and number of habitats, species occupying more habitats being less threatened. These contrasting results emphasise the importance of undertaking extinction risk analyses at an appropriate geographical scale. Since the models explained only a low percentage of total variance in the data, the effects of human-mediated habitat disturbance across a wide range of habitats may be important.

Intelligent clothing for automated recognition of human physical activities in free-living environment

This paper presents an intelligent clothing framework for human daily activity recognition using a single waist-worn tri-axial accelerometer sensor coupled with a robust pattern recognition system. The activity recognition algorithm is realized to distinguish six different physical activities through three major steps: acceleration signal collection/pre-processing, wavelet-based principle component analysis, and a support vector machine classifier. The proposed activity recognition method has been experimentally validated through two batches of trials with an overall mean classification accuracy of 95.25 and 94.87%, respectively. These results suggest that the intelligent clothing is not only able to learn the activity patterns but also capable of generalizing new data from both known and unknown subjects. This enables the proposed intelligent clothing to be applied in a comfortable and in situ assessment of human physical activities, which would open up new market segments to the textile industry.