Auckland University of Technology

About: Auckland University of Technology is a education organization based out in Auckland, New Zealand. It is known for research contribution in the topics: Population & Context (language use). The organization has 4116 authors who have published 13461 publications receiving 353076 citations. The organization is also known as: AUT & AUT University.

Evaluating prediction systems in software project estimation

Abstract: Context: Software engineering has a problem in that when we empirically evaluate competing prediction systems we obtain conflicting results. Objective: To reduce the inconsistency amongst validation study results and provide a more formal foundation to interpret results with a particular focus on continuous prediction systems. Method: A new framework is proposed for evaluating competing prediction systems based upon (1) an unbiased statistic, Standardised Accuracy, (2) testing the result likelihood relative to the baseline technique of random 'predictions', that is guessing, and (3) calculation of effect sizes. Results: Previously published empirical evaluations of prediction systems are re-examined and the original conclusions shown to be unsafe. Additionally, even the strongest results are shown to have no more than a medium effect size relative to random guessing. Conclusions: Biased accuracy statistics such as MMRE are deprecated. By contrast this new empirical validation framework leads to meaningful results. Such steps will assist in performing future meta-analyses and in providing more robust and usable recommendations to practitioners.

Possible stimuli for strength and power adaptation: acute hormonal responses

Abstract: A great deal of literature has investigated the effects of various resistance training programmes on strength and power changes. Surprisingly, however, our understanding of the stimuli that affect adaptation still remains relatively unexplained. It is thought that strength and power adaptation is mediated by mechanical stimuli, that is the kinematics and kinetics associated with resistance exercise (e.g. forces, contraction duration, power and work), and their interaction with other hormonal and metabolic factors. However, the effect of different combinations of kinematic and kinetic variables and their contribution to adaptation is unclear. The mechanical response to single repetitions has been investigated by a number of researchers; however, it seems problematic to extrapolate the findings of this type of research to the responses associated with a typical resistance training session. That is, resistance training is typified by multiple repetitions, sets and exercises, rest periods of varying durations and different movement techniques (e.g. controlled and explosive). Understanding the mechanical stimuli afforded by such loading schemes would intuitively lead to a better appreciation of how various mechanical stimuli affect adaptation. It will be evident throughout this article that very little research has adopted such an approach; hence our understanding in this area remains rudimentary at best. One should therefore remain cognizant of the limitations that exist in the interpretation of research in this field. We contend that strength and power research needs to adopt a set kinematic and kinetic analysis to improve our understanding of how to optimise strength and power.

Deep-Cascade: Cascading 3D Deep Neural Networks for Fast Anomaly Detection and Localization in Crowded Scenes

Abstract: This paper proposes a fast and reliable method for anomaly detection and localization in video data showing crowded scenes. Time-efficient anomaly localization is an ongoing challenge and subject of this paper. We propose a cubic-patch-based method, characterised by a cascade of classifiers, which makes use of an advanced feature-learning approach. Our cascade of classifiers has two main stages. First, a light but deep 3D auto-encoder is used for early identification of “many” normal cubic patches. This deep network operates on small cubic patches as being the first stage, before carefully resizing the remaining candidates of interest, and evaluating those at the second stage using a more complex and deeper 3D convolutional neural network (CNN). We divide the deep auto-encoder and the CNN into multiple sub-stages, which operate as cascaded classifiers. Shallow layers of the cascaded deep networks (designed as Gaussian classifiers, acting as weak single-class classifiers) detect “simple” normal patches, such as background patches and more complex normal patches, are detected at deeper layers. It is shown that the proposed novel technique (a cascade of two cascaded classifiers) performs comparable to current top-performing detection and localization methods on standard benchmarks, but outperforms those in general with respect to required computation time.

Global, Regional and Country-Specific Burden of Ischaemic Stroke, Intracerebral Haemorrhage and Subarachnoid Haemorrhage: A Systematic Analysis of the Global Burden of Disease Study 2017.

Abstract: Background: Stroke is a leading cause of death and disability in globally and particularly in low- and middle-income countries, and this burden is increasing. The burden of stroke pathological subtypes varies in terms of incidence, disability and mortality. Previous Global Burden of Diseases, Injuries, and Risk Factors Studies (GBD) reports did not provide separate global burden and trends estimates for haemorrhagic stroke by primary intracerebral haemorrhage (PICH) and subarachnoid haemorrhage (SAH). Aim: To summarise the GBD 2017 findings for the burden and 27-year trends for ischaemic stroke (IS), intracerebral haemorrhage and SAH by age, sex and country income level in 21 world regions and associated risk factors. Methods: Data on stroke incidence, prevalence, mortality and disability-adjusted life-years (DALY) lost and the burden of IS, PICH and SAH were derived from all available datasets from the GBD 2017 studies. Data were analysed in terms of absolute numbers and age-standardised rates per 100,000 (95% uncertainty interval [UI]), with estimates stratified by age, sex and economic development level by the World Bank classification. We also analysed changes in the patterns of incidence, mortality and DALYs estimates between 1990 and 2017. Results: In 2017, there were 11.9 million incident (95% UI 11.1–12.8), 104.2 million prevalent (98.6–110.2), 6.2 million fatal (6.0–6.3) cases of stroke and 132.1 million stroke-related DALYs (126.5–137.4). Although stroke incidence, prevalence, mortality and DALY rates declined from 1990 to 2017, the absolute number of people who developed new stroke, died, survived or remained disabled from stroke has almost doubled. The bulk of stroke burden (80% all incident strokes, 77% all stroke survivors, 87% of all deaths from stroke and 89 of all stroke-related DALYs) in 2017 was in low- to middle-income countries. Globally in 2017, IS constituted 65%, PICH –26% and SAH –9% of all incident strokes. Discussion: The latest GBD estimates of stroke burden in 195 countries supersede previous GBD stroke burden findings and provide most accurate data for stroke care planning and resource allocation globally, regionally and for 195 countries. Stroke remains the second leading cause of deaths and disability worldwide. The increased stroke burden continues to exacerbate a huge pressure on people affected by stroke, their families and societies. It is imperative to develop and implement more effective primary prevention strategies to reduce stroke burden and its impact.