OWDEAH: Online Web Data Extraction based on Access History

Web data extraction systems are the kernel of information mediators between users and heterogeneous Web data resources. How to extract structured data from semi-structured documents has been a problem of active research. Supervised and unsupervised methods have been devised to learn extraction rules from training sets. However, trying to prepare training sets (especially to annotate them for supervised methods), is very time-consuming. We propose a framework for Web data extraction, which logged usersrsquo access history and exploit them to assist automatic training set generation. We cluster accessed Web documents according to their structural details; define criteria to measure the importance of sub-structures; and then generate extraction rules. We also propose a method to adjust the rules according to historical data. Our experiments confirm the viability of our proposal.

Automated identification of lung nodules

A system that can automatically detect nodules within lung images may assist expert radiologists in interpreting the abnormal patterns as nodules in 2D CT lung images. A system is presented that can automatically identify nodules of various sizes within lung images. The pattern classification method is employed to develop the proposed system. A random forest ensemble classifier is formed consisting of many weak learners that can grow decision trees. The forest selects the decision that has the most votes. The developed system consists of two random forest classifiers connected in a series fashion. A subset of CT lung images from the LIDC database is employed. It consists of 5721 images to train and test the system. There are 411 images that contained expert- radiologists identified nodules. Training sets consisting of nodule, non-nodule, and false-detection patterns are constructed. A collection of test images are also built. The first classifier is developed to detect all nodules. The second classifier is developed to eliminate the false detections produced by the first classifier. According to the experimental results, a true positive rate of 100%, and false positive rate of 1.4 per lung image are achieved.

Learning from large data : bias, variance, sampling, and learning curves

One of the fundamental machine learning tasks is that of predictive classification. Given that organisations collect an ever increasing amount of data, predictive classification methods must be able to effectively and efficiently handle large amounts of data. However, it is understood that present requirements push existing algorithms to, and sometimes beyond, their limits since many classification prediction algorithms were designed when currently common data set sizes were beyond imagination. This has led to a significant amount of research into ways of making classification learning algorithms more effective and efficient. Although substantial progress has been made, a number of key questions have not been answered. This dissertation investigates two of these key questions. The first is whether different types of algorithms to those currently employed are required when using large data sets. This is answered by analysis of the way in which the bias plus variance decomposition of predictive classification error changes as training set size is increased. Experiments find that larger training sets require different types of algorithms to those currently used. Some insight into the characteristics of suitable algorithms is provided, and this may provide some direction for the development of future classification prediction algorithms which are specifically designed for use with large data sets. The second question investigated is that of the role of sampling in machine learning with large data sets. Sampling has long been used as a means of avoiding the need to scale up algorithms to suit the size of the data set by scaling down the size of the data sets to suit the algorithm. However, the costs of performing sampling have not been widely explored. Two popular sampling methods are compared with learning from all available data in terms of predictive accuracy, model complexity, and execution time. The comparison shows that sub-sampling generally products models with accuracy close to, and sometimes greater than, that obtainable from learning with all available data. This result suggests that it may be possible to develop algorithms that take advantage of the sub-sampling methodology to reduce the time required to infer a model while sacrificing little if any accuracy. Methods of improving effective and efficient learning via sampling are also investigated, and now sampling methodologies proposed. These methodologies include using a varying-proportion of instances to determine the next inference step and using a statistical calculation at each inference step to determine sufficient sample size. Experiments show that using a statistical calculation of sample size can not only substantially reduce execution time but can do so with only a small loss, and occasional gain, in accuracy. One of the common uses of sampling is in the construction of learning curves. Learning curves are often used to attempt to determine the optimal training size which will maximally reduce execution time while nut being detrimental to accuracy. An analysis of the performance of methods for detection of convergence of learning curves is performed, with the focus of the analysis on methods that calculate the gradient, of the tangent to the curve. Given that such methods can be susceptible to local accuracy plateaus, an investigation into the frequency of local plateaus is also performed. It is shown that local accuracy plateaus are a common occurrence, and that ensuring a small loss of accuracy often results in greater computational cost than learning from all available data. These results cast doubt over the applicability of gradient of tangent methods for detecting convergence, and of the viability of learning curves for reducing execution time in general.

A chemotactic-based model for spatial activity recognition

Spatial activity recognition in everyday environments is particularly challenging due to noise incorporated during video-tracking. We address the noise issue of spatial recognition with a biologically inspired chemotactic model that is capable of handling noisy data. The model is based on bacterial chemotaxis, a process that allows bacteria to survive by changing motile behaviour in relation to environmental dynamics. Using chemotactic principles, we propose the chemotactic model and evaluate its classification performance in a smart house environment. The model exhibits high classification accuracy (99%) with a diverse 10 class activity dataset and outperforms the discrete hidden Markov model (HMM). High accuracy (>89%) is also maintained across small training sets and through incorporation of varying degrees of artificial noise into testing sequences. Importantly, unlike other bottom–up spatial activity recognition models, we show that the chemotactic model is capable of recognizing simple interwoven activities.

Spatial activity recognition in a smart home environment using a chemotactic model

Spatial activity recognition is challenging due to the amount of noise incorporated during video tracking in everyday environments. We address the spatial recognition problem with a biologically-inspired chemotactic model that is capable of handling noisy data. The model is based on bacterial chemotaxis, a process that allows bacteria to change motile behaviour in relation to environmental gradients. Through adoption of chemotactic principles, we propose the chemotactic model and evaluate its performance in a smart house environment. The model exhibits greater than 99% recognition performance with a diverse six class dataset and outperforms the Hidden Markov Model (HMM). The approach also maintains high accuracy (90-99%) with small training sets of one to five sequences. Importantly, unlike other low-level spatial activity recognition models, we show that the chemotactic model is capable of recognising simple interwoven activities.

Statistical detection of online drifting twitter spam

Spam has become a critical problem in online social networks. This paper focuses on Twitter spam detection. Recent research works focus on applying machine learning techniques for Twitter spam detection, which make use of the statistical features of tweets. We observe existing machine learning based detection methods suffer from the problem of Twitter spam drift, i.e., the statistical properties of spam tweets vary over time. To avoid this problem, an effective solution is to train one twitter spam classifier every day. However, it faces a challenge of the small number of imbalanced training data because labelling spam samples is time-consuming. This paper proposes a new method to address this challenge. The new method employs two new techniques, fuzzy-based redistribution and asymmetric sampling. We develop a fuzzy-based information decomposition technique to re-distribute the spam class and generate more spam samples. Moreover, an asymmetric sampling technique is proposed to re-balance the sizes of spam samples and non-spam samples in the training data. Finally, we apply the ensemble technique to combine the spam classifiers over two different training sets. A number of experiments are performed on a real-world 10-day ground-truth dataset to evaluate the new method. Experiments results show that the new method can significantly improve the detection performance for drifting Twitter spam.

Enhancing flexible business training: learners and enterprises

Focuses on flexible training and its implementation in the workplace; examines the barriers for those learners on vocational programmes involved in self-directed independent learning. Discusses ways to develop learners self-direction and to improve learners ability to use and access a wide range of learning media. Highlights the need to establish structures to support learning through flexible training. Puts forward a two-dimensional diagram describing business vocational learner preferences, and outlines three sets of strategies to develop effective enterprise support for flexible training.

Motor unit synchronization measured by cross-correlation is not influenced by short-term strength training of a hand muscle

The purpose of the study was to quantify the strength of motor unit synchronization and coherence from pairs of concurrently active motor units before and after short-term (4–8 weeks) strength training of the left first dorsal interosseous (FDI) muscle. Five subjects (age 24.8 ± 4.3 years) performed a training protocol three times/week that consisted of six sets of ten maximal isometric index finger abductions, whereas three subjects (age 27.3 ± 6.7 years) acted as controls. Motor unit activity was recorded from pairs of intramuscular electrodes in the FDI muscle with two separate motor unit recording sessions obtained before and after strength training (trained group) or after 4 weeks of normal daily activities that did not involve training (control group). The training intervention resulted in a 54% (45.2 ± 8.3 to 69.5 ± 13.8 N, P = 0.001) increase in maximal index finger abduction force, whereas there was no change in strength in the control group. A total of 163 motor unit pairs (198 single motor units) were examined in both subject groups, with 52 motor unit pairs obtained from 10 recording sessions before training and 51 motor unit pairs from 10 recording sessions after training. Using the cross-correlation procedure, there was no change in the strength of motor unit synchronization following strength training (common input strength index; 0.71 ± 0.41 to 0.67 ± 0.43 pulses/s). Furthermore, motor unit coherence z scores at low (0–10 Hz; 3.9 ± 0.3 before to 4.4 ± 0.4 after) or high (10–30 Hz; 1.7 ± 0.1 before to 1.9 ± 0.1 after) frequencies were not influenced by strength training. These motor unit data indicate that increases in strength following several weeks of training a hand muscle are not accompanied by changes in motor unit synchronization or coherence, suggesting that these features of correlated motor unit activity are not important in the expression of muscle strength.

The OECD-ENSI project and its relevance for teacher training concepts in environmental education

This article presents an account of the Environment and School Initiatives (ENSI) project. The philosophical basis of the project is outlined, and case studies from within the project provided as a basis for a consideration of some current issues in the field. It is concluded that one of the distinctive features of the ENSI project is its encouragement of praxiological, action research-based to curriculum and professional development. The case studies of suggest that some of the more conventionally prescribed charcteristics such as agreed sets of goals and professional competencies might need to be revisited.

Aerobic exercise and resistance training for the management of type 2 diabetes mellitus

BACKGROUND
Implementation of a structured physical exercise program can improve glycemic control in patients with type 2 diabetes mellitus.

OBJECTIVE
To evaluate the efficacy of aerobic exercise and resistance training (either alone or in combination) in the management of type 2 diabetes mellitus.

DESIGN AND INTERVENTION
DARE (Diabetes Aerobic and Resistance Exercise) was a 26-week, single-center, parallel-group, randomized, controlled trial of patients with type 2 diabetes mellitus of >6 months' duration. Participants were aged 39-70 years with a baseline [HbA.sub.1c] level 6.6-9.9%. Exclusion criteria included current insulin therapy, regular exercise regime and blood pressure >160/95 mmHg. All participants underwent a 4-week run-in period that comprised 12 sessions of combined aerobic exercise and resistance training; participants who attended [greater than or equal to] 10 sessions were eligible to enter the study. Eligible participants were randomly allocated to one of four groups: aerobic exercise alone; resistance training alone; combined aerobic exercise and resistance training; and no intervention (control group). Exercise was performed three times weekly. The aerobic exercise group progressed from 15-20 min on a treadmill or bicycle ergometer per session at 60% of the maximum heart rate to 45 min per session at 75% of the maximum heart rate. The resistance training group performed 7 different exercises on weight machines per 45 min session, and progressed to 2-3 sets of each exercise at the maximum weight that could be lifted 7-9 times. The combined exercise group performed the full aerobic exercise program plus the full resistance training program. Participants in the control group reverted to their pre-study exercise levels.

OUTCOME MEASURES
The primary outcome measure was the change in [HbA.sub.1c] from baseline. Secondary outcome measures included changes in blood pressure, lipid profile, and body composition.

RESULTS
A total of 251 participants were eligible for intervention. The median session attendance was 80% (aerobic exercise), 85% (resistance training) and 86% (combined exercise). When compared with the control group, the HbA1c levels were reduced by 0.50% in the aerobic exercise group (P = 0.007) and by 0.38% in the resistance training group (P = 0.038). The combined exercise group had an additional reduction of 0.46% when compared with the aerobic exercise group (P = 0.014) and of 0.59% when compared with the resistance training group (P = 0.001). Decreases in [HbA.sub.1c] levels were greatest for participants with a baseline [HbA.sub.1c] level = 7.5% (P <0.001). For participants with a baseline level [HbA.sub.1c] <7.5%, significant improvements in glycemic control were observed in the combined exercise group only (P = 0.002). Changes in blood pressure and lipid profiles did not differ between the groups. By contrast, participation in a structured exercise program improved body composition.

CONCLUSION
Although aerobic exercise or resistance training alone improved glycemic control, additional improvements were observed with the combined exercise regimen.

Levy grant training scheme evolution : focus on Fiji and Singapore

This thesis examines the evolution of national training policy in Fiji since 1973 with a particular emphasis on the national levy-grant scheme that was introduced in Fiji in that same year. Developments in the Fiji National Training Council (FNTC) levy-grants scheme since its inception, including substantive amendments to the legislation in late 2002, form part of the scholarship. The thesis will provide an analytical narration of the training policy objectives and their transformation over a time span of almost three decades in the context of a small island nation. To inform this study, it was considered essential to compare the Fiji experience of levy grants schemes with other levy grants scheme. The author decided to use as the focal comparative benchmark the case of the Skills Development Fund (SDF) in Singapore. The SDF has been increasingly portrayed, by the World Bank, the International Labour Organisation and other influential agencies, as the best practice case when it comes to managing a training levy grants scheme. The thesis adopted a qualitative approach that utilized elements of case study, historical research, and key person interviews. The challenges of doing 'insider* research were explored because of its pertinence to the study. Because the study also involved the comparison of the policy experiences of two distinct countries, it was imperative to consider the issues and challenges of undertaking comparative research with particular reference to training matters- Given that training is often enmeshed with other human resources management issues, cognisance was taken of some of the broader debates in this regard. Following consideration of the methodological issues, the research paper explores the objectives of national training strategies and, in particular, issues relating to national competitiveness and skills development. The purpose is to situate the issue of training and skills development within the broader discourse of national development. Alternative approaches to the strategic role of training are considered both at the national and organisational level and some of the classic and current debates surrounding human capital investment are visited. The thesis then proceeds to examine the forms of, and rationale for government interventions in the area of training. One of the challenges both in practice and theoretically is to arrive at a consensual definition of training because of the constantly evolving context and boundaries in which training policies are fashioned. This provides the setting to examine the role that governments can and do play in skills development and how levy-grant schemes, in particular, contribute to the process. Three forms of levy grants schemes are identified and examined: levy-generating; levy-exemption; and levy-grant and reimbursement schemes. The levy-grant and reimbursement variant is the basic thrust of this thesis. In this regard, the UK experience with the levy-grant system from 1964 to 1981 is also reviewed. Some of the issues in relation to training levies are scrutinized including the levy as a sheltered source of training finance, levy rates, duration of levy, impact of levy on the quality and quantity of training, benefits to small businesses, links between training and strategic business objectives, repackaging of training to qualify for grants, and the process by which training levy policies are devised. In looking at the policy formulation, it was necessary to unpack the processes involved and explore the role of the state further. In relation to policy development and implementation, the consultation processes, role of bureaucrats, the policy context, and approaches to policy transfer are examined. In looking at the role of the state in policy development, the alternative roles of government are explored and the concepts of the 'developmental state' and the 'corporatist state* evaluated. The notion of the developmental state has particular relevance to this study given the emphasis placed by the Singaporean government on human resource development policies. This sets the scene for a detailed examination of the role of levy-grant training schemes in Fiji and Singapore. The Skills Development Fund in Singapore was developed as an integral component of national economic policy when the Singaporean government decided to break out of the 'low-skills' trap and move the economy towards a higher value adding structure. The levy-grant system was designed to complement the strategy by focusing on upgrading the skills of employees on lower incomes, the assumption being that employees on lower remuneration were more likely to need skills upgrading. The study notes that the early objectives of the SDF were displaced when it was revealed that the bulk of SDF expenditure was directed at higher level supervisory and management training. As a result, the SDF had to refocus its activities on small and medium enterprises and the workers who were likely to miss out on formal training opportunities. The Singaporean context also shows trade unions playing a significant role in worker education and literacy programmes financed under the SDF. To understand this requires some understanding of the historical linkages between the present Singaporean government and trade union leadership. Another aspect of the development of the SDF has been the constant shifting of the institutional responsibility for the scheme. As late as September 2003, the SDF was again moved, this time to the newly created Singapore Workforce Development Agency, with the focus turning to lifelong learning and assisting Singaporeans who are unemployed or made redundant as a result of the economic restructuring. The Fiji experience with the FNTC scheme is different. It evolved in the context of perceived skills shortages but there was a degree of ambiguity over its objectives. There were no specific linkages with economic policy. Relationships with other public training institutions and more recently, private training providers, have been fraught with difficulties. The study examines the origins of the policy, the early difficulties including perceived employer grievances, and the numerous external assessments of the Fiji levy-grant scheme noting that some of them were highly critical. The thesis also examines an attempted reform of the scheme in 1992-93 that proved unsuccessful and the more recent legislative reforms to the scheme in 2002 that have expanded the role of the scheme to encompass, inter alia, national occupational standards and accreditation activities. The thesis concludes by comparing the two schemes noting that the SDF is well entrenched as a policy instrument in Singapore whilst the FNTC is facing a struggle to assert its legitimacy in Fiji.

Inflammation, hepatic enzymes and resistance training in individuals with metabolic risk factors

Aims Increases in inflammatory markers, hepatic enzymes and physical inactivity are associated with the development of the metabolic syndrome (MetS). We examined whether inflammatory markers and hepatic enzymes are correlated with traditional risk factors for MetS and studied the effects of resistance training (RT) on these emerging risk factors in individuals with a high number of metabolic risk factors (HiMF, 2.9 ± 0.8) and those with a low number of metabolic risk factors (LoMF, 0.5 ± 0.5).

Methods Twenty-eight men and 27 women aged 50.8 ± 6.5 years (mean ± sd) participated in the study. Participants were randomized to four groups, HiMF training (HiMFT), HiMF control (HiMFC), LoMF training (LoMFT) and LoMF control (LoMFC). Before and after 10 weeks of RT [3 days/week, seven exercises, three sets with intensity gradually increased from 40–50% of one repetition maximum (1RM) to 75–85% of 1RM], blood samples were obtained for the measurement of pro-inflammatory cytokines, C-reactive protein (CRP), -glutamyltransferase (GGT) and alanine aminotransferase (ALT).

Results At baseline, HiMF had higher interleukin-6 (33.9%), CRP (57.1%), GGT (45.2%) and ALT (40.6%) levels, compared with LoMF (all P < 0.05). CRP, GGT and ALT correlated with the number of risk factors (r = 0.48, 0.51 and 0.57, respectively, all P < 0.01) and with other anthropometric and clinical measures (r range from 0.26 to 0.60, P < 0.05). RT did not significantly alter inflammatory markers or hepatic enzymes (all P > 0.05).

Conclusions HiMF was associated with increased inflammatory markers and hepatic enzyme concentrations. RT did not reduce inflammatory markers and hepatic enzymes in individuals with HiMF.

Corticospinal excitability following short-term motor imagery training of a strength task

Motor imagery and actual movement engage similar neural structures, however, whether they produce similar training-related corticospinal adaptations has yet to be established. The aim of this study was to compare changes in strength and corticospinal excitability following short-term motor imagery strength training and short-term strength training. Transcranial magnetic stimulation (TMS) was applied over the contralateral motor cortex (M1) to elicit motor-evoked potentials in the dominant biceps brachii muscle prior to and following 3-week strength training using actual bicep curls or motor imagery of bicep curls. The strength training (n = 6) and motor imagery (n = 6) groups underwent three supervised training sessions per week for 3 weeks. Participants completed four sets of six to eight repetitions (actual or imagined) at a training load of 80% of their one-repetition maximum. The control group (n = 6) were required to maintain their current level of physical activity. Both training groups exhibited large performance gains in strength (p < 0.001; strength training 39% improvement, imagery 16% improvement), which were significantly different between groups (p = 0.027). TMS revealed that the performance improvements observed in both imagery and strength training were accompanied by increases in corticospinal excitability (p < 0.001), however, these differences were not significantly different between groups (p = 0.920). Our findings suggest that both strength training and motor imagery training utilised similar neural substrates within the primary M1, however, strength training resulted in greater gains in strength than motor imagery strength training. This difference in strength increases may be attributed to adaptations during strength training that are not confined to the primary M1. These findings have theoretical implications for functional equivalent views of motor imagery as well as important therapeutic implications.

Portfolio careers for medical graduates: implications for postgraduate training and workforce planning.

Portfolio careers in medicine can be defined as significant involvement in one or more portfolios of activity beyond a practitioner's primary clinical role, either concurrently or in sequence. Portfolio occupations may include medical education, research, administration, legal medicine, the arts, engineering, business and consulting, leadership, politics and entrepreneurship. Despite significant interest among junior doctors, portfolios are poorly integrated with prevocational and speciality training programs in Australia. The present paper seeks to explore this issue. More formal systems for portfolio careers in Australia have the potential to increase job satisfaction, flexibility and retention, as well as diversify trainee skill sets. Although there are numerous benefits from involvement in portfolio careers, there are also risks to the trainee, employing health service and workforce modelling. Formalising pathways to portfolio careers relies on assessing stakeholder interest, enhancing flexibility in training programs, developing support programs, mentorship and coaching schemes and improving support structures in health services.

Motor cortex excitability is not differentially modulated following skill and strength training

AIM: A single session of skill or strength training can modulate the primary motor cortex (M1), which manifests as increased corticospinal excitability (CSE) and decreased short-latency intra-cortical inhibition (SICI). We tested the hypothesis that both skill and strength training can propagate the neural mechanisms mediating cross-transfer and modulate the ipsilateral M1 (iM1). METHODS: Transcranial magnetic stimulation (TMS) measured baseline CSE and SICI in the contralateral motor cortex (cM1) and iM1. Participants completed 4 sets of unilateral training with their dominant arm, either visuomotor tracking, metronome-paced strength training (MPST), self-paced strength training (SPST) or control. Immediately post training, TMS was repeated in both M1s. RESULTS: Motor-evoked potentials (MEPs) increased and inhibition was reduced for skill and MPST training from baseline in both M1s. Self-paced strength training and control did not produce changes in CSE and SICI when compared to baseline in both M1s. After training, skill and MPST increased CSE and decreased SICI in cM1 compared to SPST and control. Skill and MPST training decreased SICI in iM1 compared to SPST and control post intervention; however, CSE in iM1 was not different across groups post training. CONCLUSION: Both skill training and MPST facilitated an increase in CSE and released SICI in iM1 and cM1 compared to baseline. Our results suggest that synchronizing to an auditory or a visual cue promotes neural adaptations within the iM1, which is thought to mediate cross transfer.