An optimisation of Gaussian mixture models for integer processing units

This paper investigates sub-integer implementations of the adaptive Gaussian mixture model (GMM) for background/foreground segmentation to allow the deployment of the method on low cost/low power processors that lack Floating Point Unit (FPU). We propose two novel integer computer arithmetic techniques to update Gaussian parameters. Specifically, the mean value and the variance of each Gaussian are updated by a redefined and generalised "round'' operation that emulates the original updating rules for a large set of learning rates. Weights are represented by counters that are updated following stochastic rules to allow a wider range of learning rates and the weight trend is approximated by a line or a staircase. We demonstrate that the memory footprint and computational cost of GMM are significantly reduced, without significantly affecting the performance of background/foreground segmentation.

Behavioural context of multi-scale species distribution models assessed by radio-tracking

Incorporating ecological processes and animal behaviour into Species Distribution Models (SDMs) is difficult. In species with a central resting or breeding place, there can be conflict between the environmental requirements of the 'central place' and foraging habitat. We apply a multi-scale SDM to examine habitat trade-offs between the central place, roost sites, and foraging habitat in . Myotis nattereri. We validate these derived associations using habitat selection from behavioural observations of radio-tracked bats. A Generalised Linear Model (GLM) of roost occurrence using land cover variables with mixed spatial scales indicated roost occurrence was positively associated with woodland on a fine scale and pasture on a broad scale. Habitat selection of radio-tracked bats mirrored the SDM with bats selecting for woodland in the immediate vicinity of individual roosts but avoiding this habitat in foraging areas, whilst pasture was significantly positively selected for in foraging areas. Using habitat selection derived from radio-tracking enables a multi-scale SDM to be interpreted in a behavioural context. We suggest that the multi-scale SDM of . M. nattereri describes a trade-off between the central place and foraging habitat. Multi-scale methods provide a greater understanding of the ecological processes which determine where species occur and allow integration of behavioural processes into SDMs. The findings have implications when assessing the resource use of a species at a single point in time. Doing so could lead to misinterpretation of habitat requirements as these can change within a short time period depending on specific behaviour, particularly if detectability changes depending on behaviour. © 2011 Gesellschaft für ökologie.

Higher incidence of childhood-onset type 1 diabetes mellitus in remote areas: a UK regional small-area analysis

Aims/hypothesis: We investigated the association between the incidence of type 1 diabetes mellitus and remoteness (a proxy measure for exposure to infections) using recently developed techniques for statistical analysis of small-area data.

Subjects, materials and methods: New cases in children aged 0 to 14 years in Northern Ireland were prospectively registered from 1989 to 2003. Ecological analysis was conducted using small geographical units (582 electoral wards) and area characteristics including remoteness, deprivation and child population density. Analysis was conducted using Poisson regression models and Bayesian
hierarchical models to allow for spatially correlated risks that were potentially caused by unmeasured explanatory variables.

Results: In Northern Ireland between 1989 and 2003, there were 1,433 new cases of type 1 diabetes, giving a directly standardised incidence rate of 24.7 per 100,000 personyears. Areas in the most remote fifth of all areas had a significantly (p=0.0006) higher incidence of type 1 diabetes mellitus (incidence rate ratio=1.27 [95% CI 1.07, 1.50]) than those in the most accessible fifth of all areas. There was also a higher incidence rate in areas that were less deprived (p<0.0001) and less densely populated (p=0.002). After adjustment for deprivation and additional adjustment for child population density the association between diabetes and remoteness remained significant (p=0.01 and p=0.03, respectively).

Conclusions/interpretation: In Northern Ireland, there is evidence that remote areas experience higher rates of type 1 diabetes mellitus. This could reflect a reduced or delayed exposure to infections, particularly early in life, in these areas.

Dealing with non-linearity in shape modelling of articulated objects

We address the problem of non-linearity in 2D Shape modelling of a particular articulated object: the human body. This issue is partially resolved by applying a different Point Distribution Model (PDM) depending on the viewpoint. The remaining non-linearity is solved by using Gaussian Mixture Models (GMM). A dynamic-based clustering is proposed and carried out in the Pose Eigenspace. A fundamental question when clustering is to determine the optimal number of clusters. From our point of view, the main aspect to be evaluated is the mean gaussianity. This partitioning is then used to fit a GMM to each one of the view-based PDM, derived from a database of Silhouettes and Skeletons. Dynamic correspondences are then obtained between gaussian models of the 4 mixtures. Finally, we compare this approach with other two methods we previously developed to cope with non-linearity: Nearest Neighbor (NN) Classifier and Independent Component Analysis (ICA).

Generating Realistic Labelled, Weighted Random Graphs

Generative algorithms for random graphs have yielded insights into the structure and evolution of real-world networks. Most networks exhibit a well-known set of properties, such as heavy-tailed degree distributions, clustering and community formation. Usually, random graph models consider only structural information, but many real-world networks also have labelled vertices and weighted edges. In this paper, we present a generative model for random graphs with discrete vertex labels and numeric edge weights. The weights are represented as a set of Beta Mixture Models (BMMs) with an arbitrary number of mixtures, which are learned from real-world networks. We propose a Bayesian Variational Inference (VI) approach, which yields an accurate estimation while keeping computation times tractable. We compare our approach to state-of-the-art random labelled graph generators and an earlier approach based on Gaussian Mixture Models (GMMs). Our results allow us to draw conclusions about the contribution of vertex labels and edge weights to graph structure.

A comparative study of Type II-P and II-L supernova rise times as exemplified by the case of LSQ13cuw

We report on our findings based on the analysis of observations of the Type II-L supernova LSQ13cuw within the framework of currently accepted physical predictions of core-collapse supernova explosions. LSQ13cuw was discovered within a day of explosion, hitherto unprecedented for Type II-L supernovae. This motivated a comparative study of Type II-P and II-L supernovae with relatively well-constrained explosion epochs and rise times to maximum (optical) light. From our sample of twenty such events, we find evidence of a positive correlation between the duration of the rise and the peak brightness. On average, SNe II-L tend to have brighter peak magnitudes and longer rise times than SNe II-P. However, this difference is clearest only at the extreme ends of the rise time versus peak brightness relation. Using two different analytical models, we performed a parameter study to investigate the physical parameters that control the rise time behaviour. In general, the models qualitatively reproduce aspects of the observed trends. We find that the brightness of the optical peak increases for larger progenitor radii and explosion energies, and decreases for larger masses. The dependence of the rise time on mass and explosion energy is smaller than the dependence on the progenitor radius. We find no evidence that the progenitors of SNe II-L have significantly smaller radii than those of SNe II-P.

Using the opinions of coronary heart disease patients in designing a health education booklet for use in general practice consultations.

Objective This study explored how coronary heart disease (CHD) patients’ views would inform the design of an information booklet aimed at providing patients and practitioners with a resource to help influence positively patients’ health behaviour outcomes. Methods Opinions of patients (N=23) with CHD about their information needs, particularly lifestyle advice, were explored using a qualitative approach in four general practices. This information was used in designing a booklet for a pilot study intervention aimed at promoting healthy lifestyle behaviours and medication adherence among people with CHD. Subsequent focus groups explored patients’ (N=17) opinions about the booklet’s ‘fitness for purpose’; semi-structured interviews with practitioners (N=10) examined their views on the booklet’s usefulness. Results In initial focus groups patients identified gaps in their information provision regarding coping with stress, available local community support and medication purpose. A booklet, prepared on the basis of previous literature, was modified to address these gaps. Pilot study patients were satisfied with the re-designed booklet and practitioners reported that its use in consultations enabled change implementation and facilitated patients’ understanding of connections between lifestyle and health outcomes. Conclusion Acknowledging the opinions of CHD patients in producing health information booklets which emphasised a patient centred approach supported practitioner-patient partnerships for choosing healthy lifestyle choices.

Application of Jacobi algorithm for ISI channels

Here, the Jacobi iterative algorithm is applied to combat intersymbol interference (ISI) caused by frequency-selective channels. The performance bound of the equaliser is analysed in order to gain an insight into its asymptotic behaviour. Because of the error propagation problem, the potential of this algorithm is not reached in an uncoded system. However, its extension to a coded system with the application of the turbo-processing principle results in a new turbo equalisation algorithm, which demonstrates comparable performance with reduced complexity compared with some existing filter-based turbo equalisation schemes; and superior performance compared with some frequency domain solutions, such as orthogonal frequency division multiplexing and single-carrier frequency domain equalisation.

Do non-native invasive fish support elevated lamprey populations?

Summary
1: Managing populations of predators and their prey to achieve conservation or resource management goals is usually technically challenging and frequently socially controversial. This is true even in the simplest ecosystems but can be made much worse when predator–prey relationships are in?uenced by complex interactions, such as biological invasions, population trends or animal movements.
2: Lough Neagh in Northern Ireland is a European stronghold for pollan Coregonus autumnalis, a coregonine ?sh and for river lampreyLampetra ?uviatilis, which feeds parasitically as an adult. Both species are of high conservation importance. Lampreys are known to consume pollan but detailed knowledge of their interactions is scant. While pollan is well known to be a landlocked species in Ireland, the life cycle of normally anadromous river lamprey in Lough Neagh has been unclear. The Lough is also a highly perturbed ecosystem, supporting several invasive, non-native ?sh species that have the potential to in?uence lamprey–pollan interactions.
3: We applied stable isotope techniques to resolve both the movement patterns of lamprey and trophic interactions in this complex community. Recognizing that stable isotope studies are often hampered by high-levels of variability and uncertainty in the systems of interest, we employed novel Bayesian mixing models, which incorporate variability and uncertainty.
4: Stable isotope analyses identi?ed troutSalmo trutta and non-native breamAbramis brama as the main items in lamprey diet. Pollan only represented a major food source for lamprey between May and July.
5: Stable isotope ratios of carbon in tissues from 71 adult lamprey showed no evidence of marine carbon sources, strongly suggesting that Lough Neagh is host to a highly unusual, nonanadromous freshwater population. This ?nding marks out the Lough’s lamprey population as of particular scienti?c interest and enhances the conservation signi?cance of this feature of the Lough.
6: Synthesis and applications.Our Bayesian isotopic mixing models illustrate an unusual pattern of animal movement, enhancing conservation interest in an already threatened population. We have also revealed a complex relationship between lamprey and their food species that is suggestive of hyperpredation, whereby non-native species may sustain high lamprey populations that may in turn be detrimental to native pollan.Long-term conservation of lamprey and pollan in this system is likely to require management intervention, but in light of this exceptional complexity, no simple management options are currently supported. Conservation plans will require better characterization ofpopulation-level interactions and simulation modelling of interventions. More generally, our study demonstrates the importance of considering a full range of possible trophic interactions, particularly in complex ecosystems, and highlights Bayesian isotopic mixing models as powerful tools in resolving trophic relationships.
Key-words: Bayesian, conservation dilemma, Coregonus autumnalis, hyperpredation, Lampetra ?uviatilis, pollan, potamodromous, River lamprey, stable isotope analysis in R, stable isotope

The Design Process- Making It Relevant For Students

Within the ever-changing arenas of architectural design and education, the core element of architectural education remains: that of the design process. The consideration of how to design in addition to what to design presents architectural educators with that most constant and demanding challenge of how do we best teach the design process?

This challenge is arguably most acute at a student's early stages of their architectural education. In their first years in architecture, students will commonly concentrate on the end product rather than the process. This is, in many ways, understandable. A great deal of time, money and effort go into their final presentations. They believe that it is what is on the wall that is going to be assessed. Armed with new computer skills, they want to produce eye-catching graphics that are often no more than a celebration of a CAD package. In an era of increasing speed, immediacy of information and powerful advertising it is unsurprising that students want to race quickly to presenting an end-product.

Recognising that trend, new teaching methods and models were introduced into the second year undergraduate studio over the past two years at Queen's University Belfast, aimed at promoting student self-reflection and making the design process more relevant to the students. This paper will first generate a critical discussion on the difficulties associated with the design process before outlining some of the methods employed to help promote the following; an understanding of concept, personalisation of the design process for the individual student; adding realism and value to the design process and finally, getting he students to play to their strengths in illustrating their design process like an element of product. Frameworks, examples, outcomes and student feedback will all be presented to help illustrate the effectiveness of the new strategies employed in making the design process firstly, more relevant and therefore secondly, of greater value, to the architecture student.

Convergence of partial-wave expansions for energies, scattering amplitudes and positron annihilation rates

We use many-body theory to find the asymptotic behaviour of second-order correlation corrections to the energies and positron annihilation rates in many- electron systems with respect to the angular momenta l of the single-particle orbitals included. The energy corrections decrease as 1/(l+1/2)4, in agreement with the result of Schwartz, whereas the positron annihilation rate has a slower 1/(l+1/2)2 convergence rate. We illustrate these results by numerical calculations of the energies of Ne and Kr and by examining results from extensive con?guration-interaction calculations of PsH binding and annihilation.

Three-photon detachment of electrons from the fluorine negative ion

Absolute three-photon detachment cross sections are calculated for the fluorine negative ion within the lowest-order perturbation theory. The Dyson equation of the atomic many-body theory is used to obtain the ground-state 2p wavefunction with correct asymptotic behaviour, corresponding to the true (experimental) binding energy. We show that in accordance with the adiabatic theory this is crucial for obtaining absolute values of the multiphoton cross sections. Comparisons with other calculations and experimental data are presented.

Computational learning of the conditional phase-type (C-Ph) distribution: Learning C-Ph distributions

This paper presents a new algorithm for learning the structure of a special type of Bayesian network. The conditional phase-type (C-Ph) distribution is a Bayesian network that models the probabilistic causal relationships between a skewed continuous variable, modelled by the Coxian phase-type distribution, a special type of Markov model, and a set of interacting discrete variables. The algorithm takes a dataset as input and produces the structure, parameters and graphical representations of the fit of the C-Ph distribution as output.The algorithm, which uses a greedy-search technique and has been implemented in MATLAB, is evaluated using a simulated data set consisting of 20,000 cases. The results show that the original C-Ph distribution is recaptured and the fit of the network to the data is discussed.

Real-time GPU color-based segmentation of football players

In this paper, we propose a multi-camera application capable of processing high resolution images and extracting features based on colors patterns over graphic processing units (GPU). The goal is to work in real time under the uncontrolled environment of a sport event like a football match. Since football players are composed for diverse and complex color patterns, a Gaussian Mixture Models (GMM) is applied as segmentation paradigm, in order to analyze sport live images and video. Optimization techniques have also been applied over the C++ implementation using profiling tools focused on high performance. Time consuming tasks were implemented over NVIDIA's CUDA platform, and later restructured and enhanced, speeding up the whole process significantly. Our resulting code is around 4-11 times faster on a low cost GPU than a highly optimized C++ version on a central processing unit (CPU) over the same data. Real time has been obtained processing until 64 frames per second. An important conclusion derived from our study is the scalability of the application to the number of cores on the GPU. © 2011 Springer-Verlag.