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.

Extreme learning machine for predicting HLA-peptide binding

Machine learning techniques have been recognized as powerful tools for learning from data. One of the most popular learning techniques, the Back-Propagation (BP) Artificial Neural Networks, can be used as a computer model to predict peptides binding to the Human Leukocyte Antigens (HLA). The major advantage of computational screening is that it reduces the number of wet-lab experiments that need to be performed, significantly reducing the cost and time. A recently developed method, Extreme Learning Machine (ELM), which has superior properties over BP has been investigated to accomplish such tasks. In our work, we found that the ELM is as good as, if not better than, the BP in term of time complexity, accuracy deviations across experiments, and most importantly - prevention from over-fitting for prediction of peptide binding to HLA.

Real-Time Gene-Expression Extraction with Fast Classification (FC) Networks

Fast Classification (FC) networks were inspired by a biologically plausible mechanism for short term memory where learning occurs instantaneously. Both weights and the topology for an FC network are mapped directly from the training samples by using a prescriptive training scheme. Only two presentations of the training data are required to train an FC network. Compared with iterative learning algorithms such as Back-propagation (which may require many hundreds of presentations of the training data), the training of FC networks is extremely fast and learning convergence is always guaranteed. Thus FC networks may be suitable for applications where real-time classification is needed. In this paper, the FC networks are applied for the real-time extraction of gene expressions for Chlamydia microarray data. Both the classification performance and learning time of the FC networks are compared with the Multi-Layer Proceptron (MLP) networks and support-vector-machines (SVM) in the same classification task. The FC networks are shown to have extremely fast learning time and comparable classification accuracy.

Does anticipation of back pain predispose to back trouble?

Limb movement imparts a perturbation to the body. The impact of that perturbation is limited via anticipatory postural adjustments. The strategy by which the CNS controls anticipatory postural adjustments of the trunk muscles during limb movement is altered during acute back pain and in people with recurrent back pain, even when they are pain free. The altered postural strategy probably serves to protect the spine in the short term, but it is associated with a cost and is thought to predispose spinal structures to injury in the long term. It is not known why this protective strategy might occur even when people are pain free, but one possibility is that it is caused by the anticipation of back pain. In eight healthy subjects, recordings of intramuscular EMG were made from the trunk muscles during single and repetitive arm movements. Anticipation of experimental back pain and anticipation of experimental elbow pain were elicited by the threat of painful cutaneous stimulation. There was no effect of anticipated experimental elbow pain on postural adjustments. During anticipated experimental back pain, for single arm movements there was delayed activation of the deep trunk muscles and augmentation of at least one superficial trunk muscle. For repetitive arm movements, there was decreased activity and a shift from biphasic to monophasic activation of the deep trunk muscles and increased activity of superficial trunk muscles during anticipation of back pain. In both instances, the changes were consistent with adoption of an altered strategy for postural control and were similar to those observed in patients with recurrent back pain. We conclude that anticipation of experimental back pain evokes a protective postural strategy that stiffens the spine. This protective strategy is associated with compressive cost and is thought to predispose to spinal injury if maintained long term. © Guarantors of Brain 2004; all rights reserved

(FAB07_002) From the 'Curry to the 'Weal: Aboriginal town camps and compounds of the back-blocks

My involvement with Aboriginal people began in 1972 in my final year of architecture, when a small group of students were asked to advise on some Aboriginal building projects in Mt Isa and Cloncurry. This led to my Doctoral research and grew into the Aboriginal Environments Research Centre now well established at the university of Queensland. Although the personnel of this Centre have completed over 140 field trips in the last 20 years, it is a set of data collected largely from the first ten field trips in 1972-76 that will be presented in this paper.

Stress corrosion crack propagation in AerMet 100

Fracture mechanics tests were carried out for AerMet 100 in distilled water and NaCl (3.5 and 35 gl(-1)). The initiation period at higher values of the stress intensity factor indicated that load application in the stress corrosion cracking (SCC) environment is a necessary but not sufficient factor for SCC and that time is needed for some other factor (e.g., the local hydrogen concentration) to reach an appropriate value. The threshold stress intensity factor, K-ISSC, was found to increase with decreasing NaCl concentration. The plateau stress corrosion crack velocity was 2 x 10(-8) ms(-1) for NaCl (3.5 and 35 gl(-1)). The fracture mode was transgranular with small areas of an intergranular nature. (C) 1998 Chapman & Hall.

Delayed postural contraction of transversus abdominis in low back pain associated with movement of the lower limb

The temporal parameters of the response of the trunk muscles associated with movement of the lower limb were investigated in people with and without low back pain (LBP). The weight shift component of the task was completed voluntarily prior to a stimulus to move to allow investigation of the movement component of the response. In the control subjects the onset of electromyographic (EMG) activity of all trunk muscles preceded that of the muscle responsible for limb movement, thus contributing to the feed forward postural response. The EMG onset of transversus abdominis was delayed in the LBP subjects with movement in each direction, while the EMG onsets of rectus abdominis, erector spinae, and oblique abdominal muscles were delayed with specific movement directions. This result provides evidence of a change in the postural control of the trunk in people with LBP.

Incorporating cognitive-behavioral therapy with occupational therapy: A comparative study with patients with low back pain

This paper reports on the implementation of a psychoeducational program utilizing cognitive-behavioral principles. The efficacy of this psychoeducational treatment program in modifying dysfunctional attitudes in patients with chronic low back pain was examined using a two-group pretest posttest design with a follow-lip at 3 months Thirty patients (average age = 44.37 SD = 13.71) participated in the study, with 15 in the psychoeducational treatment group and 15 in the placebo control group. These two conditions were added on to an existing eclectic inpatient pain management program. After assessment on the IPAM (The Integrated Psychosocial Assessment Model), scores were reduced to multivariate composite scores on the factors of illness behavior depressed and negative cognitions, and acute pain strategies. Results of a group x time repeated measures analysis of variance for the three pain factors revealed a significant main effect for group (F(23,1) = 5.00 p < .04), tempered by a significant interaction between group and rime on the 'depressed and negative' pain factor (F(23,1) = 4.77 p < .04). Patients in the treatment group improved significantly over time and significantly more than the placebo control group patients at posttreatment. Results provide support for the program in increasing patients' feelings of control over their pain and the use of positive coping strategies, while reducing perceived helplessness, depression, disability, and pain intensity.

Analytical solution for tidal propagation in a coupled semi-confined/phreatic coastal aquifer

An analytical solution is derived for tidal fluctuations in a coupled coastal aquifer system consisting of a semi-confined aquifer, a thin semi-permeable layer and a phreatic aquifer. Based on the solution, we study the interactions (via leakage) between the confined and unconfined aquifers in response to tides. The results show that, under certain conditions, leakage from the confined aquifer can affect considerably the tidal water table fluctuation in the phreatic aquifer and vice versa. Ignoring these effects could lead to errors in estimating aquifer properties based on tidal signals. (C) 2002 Elsevier Science Ltd. All rights reserved.

Mode of action of back-line application of spinosad, a non-systemic parasiticide, on sheep

Spinosad, applied as a jetting solution or dip is an efficacious, non-systemic treatment for the control of Bovicola ovis in sheep. This paper describes the effect of back-line treatment width and group housing of animals on the efficacy of spinosad for the control of lice. A 0.4 mg/kg liveweight dose was found to be the suboptimal dose of spinosad for the control of body lice in a dose titration study and was used to investigate application and housing effects in a second study. Lousy Merino sheep were treated with either a narrow 3-cm application of spinosad or with a wider 25-cm swathe. After treatment they were either kept alone or in groups of 6 sheep per pen. Lice were counted at day 0 and every 14 days to 70 days after treatment before estimation of the percentage of lice control and analysis of treatment effects. A much higher percentage of lice control was achieved with 0.4 mg/kg in the second study than in the first, possibly because of differences in formulation used. The wider application width gave significantly higher (P < 0.05) control of lice than the narrow application when sheep were either housed alone or in groups up to day 42 post-treatment. Greater control of lice was seen in group-housed sheep compared with sheep housed individually (P < 0.05) up to day 70. Using broader application widths combined with holding the animals together after treatment with pour-on formulations may optimise the delivery and efficacy of ectoparasiticides for livestock.

Multifidus muscle recovery is not automatic after resolution of acute, first-episode low back pain

Study Design. A clinical study was conducted on 39 patients with acute, first-episode, unilateral low back pain and unilateral, segmental inhibition of the multifidus muscle. Patients were allocated randomly to a control or treatment group. Objectives. To document the natural course of lumbar multifidus recovery and to evaluate the effectiveness of specific, localized, exercise therapy on muscle recovery. Summary of Background Data. Acute low back pain usually resolves spontaneously, but the recurrence rate is high. Inhibition of multifidus occurs with acute, first-episode, low back pain, and pathologic changes in this muscle have been linked with poor outcome and recurrence of symptoms. Methods. Patients in group 1 received medical treatment only. Patients in group 2 received medical treatment and specific, localized, exercise therapy. Outcome measures for both groups included 4 weekly assessments of pain, disability, range of motion, and size of the multifidus cross-sectional area. Independent examiners were blinded to group allocation. Patients were reassessed at a 10-week follow-up examination. Results. Multifidus muscle recovery was not spontaneous on remission of painful symptoms in patients in group 1. Muscle recovery was more rapid and more complete in patients in group 2 who received exercise therapy (P = 0.0001). Other outcome measurements were similar for the two groups at the 4-week examination. Although they resumed normal levels of activity, patients in group 1 still had decreased multifidus muscle size at the 10-week follow-up examination. Conclusions. Multifidus muscle recovery is not spontaneous on remission of painful symptoms. Lack of localized, muscle support may be one reason for the high recurrence rate of low back pain following the initial episode.

Critical sets in direct products of back circulant Latin squares

To date very Few families of critical sets for latin squares are known. The only previously known method for constructing critical sets involves taking a critical set which is known to satisfy certain strong initial conditions and using a doubling construction. This construction can be applied to the known critical sets in back circulant latin squares of even order. However, the doubling construction cannot be applied to critical sets in back circulant latin squares of odd order. In this paper a family of critical sets is identified for latin squares which are the product of the latin square of order 2 with a back circulant latin square of odd order. The proof that each element of the critical set is an essential part of the reconstruction process relies on the proof of the existence of a large number of latin interchanges.