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.

Designing teams for first-of-a-kind, complex systems using the initial phases of cognitive work analysis: Case study

We present a technique for team design based on cognitive work analysis (CWA). We first develop a rationale for this technique by discussing the limitations of conventional approaches for team design in light of the special characteristics of first-of-a-kind, complex systems. We then introduce the CWA-based technique for team design and provide a case study of how we used this technique to design a team for a first-of-a-kind, complex military system during the early stages of its development. In addition to illustrating the CWA-based technique by example, the case study allows us to evaluate the technique. This case study demonstrates that the CWA-based technique for team design is both feasible and useful, although empirical validation of the technique is still necessary. Applications of this work include the design of teams for first-of-a-kind, complex systems in military, medical, and industrial domains.

Regulation of the gene encoding glutathione S-transferase M1 (GSTM1) by the Myb oncoprotein

The identification of Myb 'target' genes will not only aid in the understanding of how overexpression of Myb, or expression of activated forms of Myb, leads to cellular transformation but will also shed light on its role in normal cells. Using a combination of an estrogen-regulated Myb-transformed cell line (ERMYB) and PCR-based subtractive hybridization, we have identified the gene (GSTM1) encoding the detoxification enzyme glutathione S-transferase M1 as being transcriptionally upregulated by Myb. Functional analysis of the GSTM1 promoter using reporter assays indicated that both the DNA binding and transactivation domains of Myb were required for transcriptional activation. Mutational analysis of consensus Myb-binding sites (MBS) in the promoter and electrophoretic mobility gel shift analysis indicated that one of the three potential MBS can bind Myb protein, and is the primary site involved in the regulation of this promoter by Myb.

Screening for leprosy in immigrants: A decision analysis model

Almost all leprosy cases reported in industrialized countries occur amongst immigrants or refugees from developing countries where leprosy continues to be an important health issue. Screening for leprosy is an important question for governments in countries with immigration and refugee programmes. A decision analysis framework is used to evaluate leprosy screening. The analysis uses a set of criteria and parameters regarding leprosy screening, and available data to estimate the number of cases which would be detected by a leprosy screening programme of immigrants from countries with different leprosy prevalences, compared with a policy of waiting for immigrants who develop symptomatic clinical diseases to present for health care. In a cohort of 100,000 immigrants from high leprosy prevalence regions (3.6/10,000), screening would detect 32 of the 42 cases which would arise in the destination country over the 14 years after migration; from medium prevalence areas (0.7/10,000) 6.3 of the total 8.1 cases would be detected, and from low prevalence regions (0.2/10,600) 1.8 of 2.3 cases. Using Australian data, the migrant mix would produce 74 leprosy cases from 10 years intake; screening would detect 54, and 19 would be diagnosed subsequently after migration. Screening would only produce significant case-yield amongst immigrants from regions or social groups with high leprosy prevalence. Since the number of immigrants to Australia from countries of higher endemnicity is not large routine leprosy screening would have a small impact on case incidence.

Characterisation of the three-dimensional structure of earthworm burrow systems using image analysis and mathematical morphology

The aim of this work was to exemplify the specific contribution of both two- and three-dimensional (31)) X-ray computed tomography to characterise earthworm burrow systems. To achieve this purpose we used 3D mathematical morphology operators to characterise burrow systems resulting from the activity of an anecic (Aporrectodea noctunia), and an endogeic species (Allolobophora chlorotica), when both species were introduced either separately or together into artificial soil cores. Images of these soil cores were obtained using a medical X-ray tomography scanner. Three-dimensional reconstructions of burrow systems were obtained using a specifically developed segmentation algorithm. To study the differences between burrow systems, a set of classical tools of mathematical morphology (granulometries) were used. So-called granulometries based on different structuring elements clearly separated the different burrow systems. They enabled us to show that burrows made by the anecic species were fatter, longer, more vertical, more continuous but less sinuous than burrows of the endogeic species. The granulometry transform of the soil matrix showed that burrows made by A. nocturna were more evenly distributed than those of A. chlorotica. Although a good discrimination was possible when only one species was introduced into the soil cores, it was not possible to separate burrows of the two species from each other in cases where species were introduced into the same soil core. This limitation, partly due to the insufficient spatial resolution of the medical scanner, precluded the use of the morphological operators to study putative interactions between the two species.

Analysis of soil wetting and solute transport in subsurface trickle irrigation

The increased use of trickle or drip irrigation is seen as one way of helping to improve the sustainability of irrigation systems around the world. However, soil water and solute transport properties and soil profile characteristics are often not adequately incorporated in the design and management of trickle systems. In this paper, we describe results of a simulation study designed to highlight the impacts of soil properties on water and solute transport from buried trickle emitters. The analysis addresses the influence of soil hydraulic properties, soil layering, trickle discharge rate, irrigation frequency, and timing of nutrient application on wetting patterns and solute distribution. We show that (1) trickle irrigation can improve plant water availability in medium and low permeability fine-textured soils, providing that design and management are adapted to account for their soil hydraulic properties, (2) in highly permeable coarse-textured soils, water and nutrients move quickly downwards from the emitter, making it difficult to wet the near surface zone if emitters are buried too deep, and (3) changing the fertigation strategy for highly permeable coarse-textured soils to apply nutrients at the beginning of an irrigation cycle can maintain larger amounts of nutrient near to and above the emitter, thereby making them less susceptible to leaching losses. The results demonstrate the need to account for differences in soil hydraulic properties and solute transport when designing irrigation and fertigation management strategies. Failure to do this will result in inefficient systems and lost opportunities for reducing the negative environmental impacts of irrigation.

A medial to lateral shift in premovement cortical activity in hemi-Parkinson's disease: Studies of event-related potentials and whole-scalp magneto encephalography

Cortical activity associated with voluntary movement is shifted from medial to lateral premotor areas in Parkinson's disease. This occurs bilaterally, even for unilateral movements. We have used both EEG and MEG to further investigate medial and lateral premotor activity in patients with hemi-Parkinson's disease, in whom basal ganglia impairment is most pronounced in one hemisphere. The CNV, recorded from 21 scalp positions in a Go/NoGo task, was maximal over central medial regions in control subjects. For hemi-Parkinson's disease subjects, activity was shifted more frontally, reduced in the midline and lateralised towards the side of greatest basal ganglia impairment. With 143 channel whole-scalp magneto encephalography (MEG) we are further examining asymmetries in supplementary motor/premotor cortical activity prior to self-paced voluntary movement. In preliminary results, one hemi-Parkinson's disease patient with predominantly left-side symptoms showed strong medial activity consistent with a dominant source in the left supplementary motor area (SMA). Three patients showed little medial activity, but early bilateral sources within lateral premotor cortex. Results suggest greater involvement of lateral premotor rather than the SMA prior to movement in Parkinson's disease and provide evidence for asymmetric function of the SMA in hemi- Parkinson's disease, with reduced activity on the side of greatest basal ganglia deficit.

Event-related FMRI analysis with maximum correlation modelling

Functional magnetic resonance imaging (FMRI) analysis methods can be quite generally divided into hypothesis-driven and data-driven approaches. The former are utilised in the majority of FMRI studies, where a specific haemodynamic response is modelled utilising knowledge of event timing during the scan, and is tested against the data using a t test or a correlation analysis. These approaches often lack the flexibility to account for variability in haemodynamic response across subjects and brain regions which is of specific interest in high-temporal resolution event-related studies. Current data-driven approaches attempt to identify components of interest in the data, but currently do not utilise any physiological information for the discrimination of these components. Here we present a hypothesis-driven approach that is an extension of Friman's maximum correlation modelling method (Neurolmage 16, 454-464, 2002) specifically focused on discriminating the temporal characteristics of event-related haemodynamic activity. Test analyses, on both simulated and real event-related FMRI data, will be presented.

Identification of a minimal promoter sequence for the human N-acetyltransferase Type I gene that binds AP-1 (activator protein 1) and YY-1 (Yin and Yang 1)

Human N-acetyltransferase Type I (NAT1) catalyses the acetylation of many aromatic amine and hydrazine compounds and it has been implicated in the catabolism of folic acid. The enzyme is widely expressed in the body, although there are considerable differences in the level of activity between tissues. A search of the mRNA databases revealed the presence of several NAT1 transcripts in human tissue that appear to be derived from different promoters. Because little is known about NAT1 gene regulation, the present study was undertaken to characterize one of the putative promoter sequences of the NAT1 gene located just upstream of the coding region. We show with reverse-transcriptase PCR that mRNA transcribed from this promoter (Promoter 1) is present in a variety of human cell-lines, but not in quiescent peripheral blood mononuclear cells. Using deletion mutant constructs, we identified a 20 bp sequence located 245 bases upstream of the translation start site which was sufficient for basal NAT1 expression. It comprised an AP-1 (activator protein 1)-binding site, flanked on either side by a TCATT motif. Mutational analysis showed that the AP-1 site and the 3' TCATT sequence were necessary for gene expression, whereas the 5' TCATT appeared to attenuate promoter activity. Electromobility shift assays revealed two specific bands made up by complexes of c-Fos/Fra, c-Jun, YY-1 (Yin and Yang 1) and possibly Oct-1. PMA treatment enhanced expression from the NAT1 promoter via the AP-1-binding site. Furthermore, in peripheral blood mononuclear cells, PMA increased endogenous NAT1 activity and induced mRNA expression from Promoter I, suggesting that it is functional in vivo.