An investigation into the use of physical modelling for the prediction of various feature types visible from different view points

This paper describes a general purpose flexible technique which uses physical modelling techniques for determining the features of a 3D object that are visible from any predefined view. Physical modelling techniques are used to determine which of many different types of features are visible from a complete set of viewpoints. The power of this technique lies in its ability to detect and parameterise object features, regardless of object complexity. Raytracing is used to simulate the physical process by which object features are visible so that surface properties (eg specularity, transparency) as well as object boundaries can be used in the recognition process. Using this technique occluding and non-occluding edge based features are extracted using image processing techniques and then parameterised. Features caused by specularity are also extracted and qualitative descriptions for these are defined.

Coupled modelling of hydrodynamics and mass transfer in membrane filtration

This research produced a novel predictive computational model for the water treatment processes of nanofiltration and reverse osmosis. This model combined commercial computational fluid dynamics codes with numerical mass transfer models developed by the candidate to provide a rigorous description of these processes’ hydrodynamic and pollutant removal behaviour.

Modelling attribute dependencies in single unit game combat settings

The relative strengths of individual units, described as attributes, determine the outcome during unit conflicts involving units in the armies of opponents in large strategy games. We describe a process for developing predictive models of the outcome for such conflicts where combat involves single unit adversaries where each unit is defined in terms of 4 fundamental attributes. These attributes: range, speed, health and damage are common to unit descriptions in many successful real-time strategy games. Our analysis process identifies three phases of game play with invariant properties that hold throughout each phase. We demonstrate the utility of this analysis by creating a model that predicts game outcomes. Predictions are validated against the game simulation. The game outcomes are explicitly related to the attributes of the two units involved, highlighting the significance and role of each attribute.

Intervention-driven predictive framework for modeling healthcare data

Assessing prognostic risk is crucial to clinical care, and critically dependent on both diagnosis and medical interventions. Current methods use this augmented information to build a single prediction rule. But this may not be expressive enough to capture differential effects of interventions on prognosis. To this end, we propose a supervised, Bayesian nonparametric framework that simultaneously discovers the latent intervention groups and builds a separate prediction rule for each intervention group. The prediction rule is learnt using diagnosis data through a Bayesian logistic regression. For inference, we develop an efficient collapsed Gibbs sampler. We demonstrate that our method outperforms baselines in predicting 30-day hospital readmission using two patient cohorts - Acute Myocardial Infarction and Pneumonia. The significance of this model is that it can be applied widely across a broad range of medical prognosis tasks. © 2014 Springer International Publishing.

A feasibility study for modelling tie strength with the facebook API

This thesis examines the concept of tie strength and investigates how it can be determined on the fly in the Facebook Social Network Service (SNS) by a system constructed using the standard developer API. We analyze and compare two different models: the first one is an adaptation of previous literature (Gilbert & Karahalios, 2009), the second model is built from scratch and based on a dataset obtained from an online survey. This survey took the form of a Facebook application that collected subjective ratings of the strength of 1642 ties (friendships) from 85 different participants. The new tie strength model was built based on this dataset by using a multiple regression method. We saw that the new model performed slightly better than the original adapted model, plus it had the advantage of being easier to implement. In conclusion, this thesis has shown that tie strength models capable of serving as useful friendship predictors are easily implementable in a Facebook application via standard API calls. In addition to a new tie strength model, the methodology adopted in this work permitted observation of the weights of each predictive variable used in the model, increasing the visibility of the factors that affects peoples’ relationships in online social networks.

Fluid dynamic analysis and modelling of industrial chemical equipment

The research is aimed at contributing to the identification of reliable fully predictive Computational Fluid Dynamics (CFD) methods for the numerical simulation of equipment typically adopted in the chemical and process industries. The apparatuses selected for the investigation, specifically membrane modules, stirred vessels and fluidized beds, were characterized by a different and often complex fluid dynamic behaviour and in some cases the momentum transfer phenomena were coupled with mass transfer or multiphase interactions. Firs of all, a novel modelling approach based on CFD for the prediction of the gas separation process in membrane modules for hydrogen purification is developed. The reliability of the gas velocity field calculated numerically is assessed by comparison of the predictions with experimental velocity data collected by Particle Image Velocimetry, while the applicability of the model to properly predict the separation process under a wide range of operating conditions is assessed through a strict comparison with permeation experimental data. Then, the effect of numerical issues on the RANS-based predictions of single phase stirred tanks is analysed. The homogenisation process of a scalar tracer is also investigated and simulation results are compared to original passive tracer homogenisation curves determined with Planar Laser Induced Fluorescence. The capability of a CFD approach based on the solution of RANS equations is also investigated for describing the fluid dynamic characteristics of the dispersion of organics in water. Finally, an Eulerian-Eulerian fluid-dynamic model is used to simulate mono-disperse suspensions of Geldart A Group particles fluidized by a Newtonian incompressible fluid as well as binary segregating fluidized beds of particles differing in size and density. The results obtained under a number of different operating conditions are compared with literature experimental data and the effect of numerical uncertainties on axial segregation is also discussed.

Analysis and modelling of the performance of technologies for flue gas treatment in Waste-to-Energy processes

Waste management represents an important issue in our society and Waste-to-Energy incineration plants have been playing a significant role in the last decades, showing an increased importance in Europe. One of the main issues posed by waste combustion is the generation of air contaminants. Particular concern is present about acid gases, mainly hydrogen chloride and sulfur oxides, due to their potential impact on the environment and on human health. Therefore, in the present study the main available technological options for flue gas treatment were analyzed, focusing on dry treatment systems, which are increasingly applied in Municipal Solid Wastes (MSW) incinerators. An operational model was proposed to describe and optimize acid gas removal process. It was applied to an existing MSW incineration plant, where acid gases are neutralized in a two-stage dry treatment system. This process is based on the injection of powdered calcium hydroxide and sodium bicarbonate in reactors followed by fabric filters. HCl and SO2 conversions were expressed as a function of reactants flow rates, calculating model parameters from literature and plant data. The implementation in a software for process simulation allowed the identification of optimal operating conditions, taking into account the reactant feed rates, the amount of solid products and the recycle of the sorbent. Alternative configurations of the reference plant were also assessed. The applicability of the operational model was extended developing also a fundamental approach to the issue. A predictive model was developed, describing mass transfer and kinetic phenomena governing the acid gas neutralization with solid sorbents. The rate controlling steps were identified through the reproduction of literature data, allowing the description of acid gas removal in the case study analyzed. A laboratory device was also designed and started up to assess the required model parameters.

Elderly age is not a negative predictive factor for virological response to therapy with pegylated interferon-α and ribavirin in chronic hepatitis C virus patients

BACKGROUND & AIMS: Age is frequently discussed as negative host factor to achieve a sustained virological response (SVR) to antiviral therapy of chronic hepatitis C. However, elderly patients often show advanced fibrosis/cirrhosis as known negative predictive factor. The aim of this study was to assess age as an independent predictive factor during antiviral therapy. METHODS: Overall, 516 hepatitis C patients were treated with pegylated interferon-α and ribavirin, thereof 66 patients ≥60 years. We analysed the impact of host factors (age, gender, fibrosis, haemoglobin, previous hepatitis C treatment) and viral factors (genotype, viral load) on SVR per therapy course by performing a generalized estimating equations (GEE) regression modelling, a matched pair analysis and a classification tree analysis. RESULTS: Overall, SVR per therapy course was 42.9 and 26.1%, respectively, in young and elderly patients with hepatitis C virus (HCV) genotypes 1/4/6. The corresponding figures for HCV genotypes 2/3 were 74.4 and 84%. In the GEE model, age had no significant influence on achieving SVR. In matched pair analysis, SVR was not different in young and elderly patients (54.2 and 55.9% respectively; P = 0.795 in binominal test). In classification tree analysis, age was not a relevant splitting variable. CONCLUSIONS: Age is not a significant predictive factor for achieving SVR, when relevant confounders are taken into account. As life expectancy in Western Europe at age 60 is more than 20 years, it is reasonable to treat chronic hepatitis C in selected elderly patients with relevant fibrosis or cirrhosis but without major concomitant diseases, as SVR improves survival and reduces carcinogenesis.

TCAD for PV: a fast method for accurately modelling metal impurity evolution during solar cell processing

Coupled device and process silumation tools, collectively known as technology computer-aided design (TCAD), have been used in the integrated circuit industry for over 30 years. These tools allow researchers to quickly converge on optimized devide designs and manufacturing processes with minimal experimental expenditures. The PV industry has been slower to adopt these tools, but is quickly developing competency in using them. This paper introduces a predictive defect engineering paradigm and simulation tool, while demonstrating its effectiveness at increasing the performance and throughput of current industrial processes. the impurity-to-efficiency (I2E) simulator is a coupled process and device simulation tool that links wafer material purity, processing parameters and cell desigh to device performance. The tool has been validated with experimental data and used successfully with partners in industry. The simulator has also been deployed in a free web-accessible applet, which is available for use by the industrial and academic communities.

A Finite Element Numerical Algorithm for Modelling and Data Fitting in Complex Systems

Numerical modelling methodologies are important by their application to engineering and scientific problems, because there are processes where analytical mathematical expressions cannot be obtained to model them. When the only available information is a set of experimental values for the variables that determine the state of the system, the modelling problem is equivalent to determining the hyper-surface that best fits the data. This paper presents a methodology based on the Galerkin formulation of the finite elements method to obtain representations of relationships that are defined a priori, between a set of variables: y = z(x1, x2,...., xd). These representations are generated from the values of the variables in the experimental data. The approximation, piecewise, is an element of a Sobolev space and has derivatives defined in a general sense into this space. The using of this approach results in the need of inverting a linear system with a structure that allows a fast solver algorithm. The algorithm can be used in a variety of fields, being a multidisciplinary tool. The validity of the methodology is studied considering two real applications: a problem in hydrodynamics and a problem of engineering related to fluids, heat and transport in an energy generation plant. Also a test of the predictive capacity of the methodology is performed using a cross-validation method.

Predictive vaccinology: Optimisation of predictions using support vector machine classifiers

Promiscuous human leukocyte antigen (HLA) binding peptides are ideal targets for vaccine development. Existing computational models for prediction of promiscuous peptides used hidden Markov models and artificial neural networks as prediction algorithms. We report a system based on support vector machines that outperforms previously published methods. Preliminary testing showed that it can predict peptides binding to HLA-A2 and -A3 super-type molecules with excellent accuracy, even for molecules where no binding data are currently available.

Modelling predicts positive and negative interactions between three Australian tropical tree species in monoculture and binary mixture

Computer modelling promises to. be an important tool for analysing and predicting interactions between trees within mixed species forest plantations. This study explored the use of an individual-based mechanistic model as a predictive tool for designing mixed species plantations of Australian tropical trees. The 'spatially explicit individually based-forest simulator' (SeXI-FS) modelling system was used to describe the spatial interaction of individual tree crowns within a binary mixed-species experiment. The three-dimensional model was developed and verified with field data from three forest tree species grown in tropical Australia. The model predicted the interactions within monocultures and binary mixtures of Flindersia brayleyana, Eucalyptus pellita and Elaeocarpus grandis, accounting for an average of 42% of the growth variation exhibited by species in different treatments. The model requires only structural dimensions and shade tolerance as species parameters. By modelling interactions in existing tree mixtures, the model predicted both increases and reductions in the growth of mixtures (up to +/- 50% of stem volume at 7 years) compared to monocultures. This modelling approach may be useful for designing mixed tree plantations. (c) 2006 Published by Elsevier B.V.

Modelling broccoli development, yield and quality

Broccoli is a vegetable crop of increasing importance in Australia, particularly in south-east Queensland and farmers need to maintain a regular supply of good quality broccoli to meet the expanding market. A predictive model of ontogeny, incorporating climatic data including frost risk, would enable farmers to predict harvest maturity date and select appropriate cultivar – sowing date combinations. To develop procedures for predicting ontogeny, yield and quality, field studies using three cultivars, ‘Fiesta’, ‘Greenbelt’ and ‘Marathon’, were sown on eight dates from 11 March to 22 May 1997, and grown under natural and extended (16 h) photoperiods at the University of Queensland, Gatton Campus. Cultivar, rather than the environment, mainly determined head quality attributes of head shape and branching angle. Yield and quality were not influenced by photoperiod. A better understanding of genotype and environmental interactions will help farmers optimise yield and quality, by matching cultivars with time of sowing. The estimated base and optimum temperature for broccoli development were 0°C and 20 °C, respectively, and were consistent across cultivars, but thermal time requirements for phenological intervals were cultivar specific. Differences in thermal time requirement from floral initiation to harvest maturity between cultivars were small and of little importance, but differences in thermal time requirement from emergence to floral initiation were large. Sensitivity to photoperiod and solar radiation was low in the three cultivars used. This research has produced models to assist broccoli farmers in crop scheduling and cultivar selection in south-east Queensland.

Trait physiology and crop modelling to link phenotypic complexity to underlying genetic systems

New tools derived from advances in molecular biology have not been widely adopted in plant breeding because of the inability to connect information at gene level to the phenotype in a manner that is useful for selection. We explore whether a crop growth and development modelling framework can link phenotype complexity to underlying genetic systems in a way that strengthens molecular breeding strategies. We use gene-to-phenotype simulation studies on sorghum to consider the value to marker-assisted selection of intrinsically stable QTLs that might be generated by physiological dissection of complex traits. The consequences on grain yield of genetic variation in four key adaptive traits – phenology, osmotic adjustment, transpiration efficiency, and staygreen – were simulated for a diverse set of environments by placing the known extent of genetic variation in the context of the physiological determinants framework of a crop growth and development model. It was assumed that the three to five genes associated with each trait, had two alleles per locus acting in an additive manner. The effects on average simulated yield, generated by differing combinations of positive alleles for the traits incorporated, varied with environment type. The full matrix of simulated phenotypes, which consisted of 547 location-season combinations and 4235 genotypic expression states, was analysed for genetic and environmental effects. The analysis was conducted in stages with gradually increased understanding of gene-to-phenotype relationships, which would arise from physiological dissection and modelling. It was found that environmental characterisation and physiological knowledge helped to explain and unravel gene and environment context dependencies. We simulated a marker-assisted selection (MAS) breeding strategy based on the analyses of gene effects. When marker scores were allocated based on the contribution of gene effects to yield in a single environment, there was a wide divergence in rate of yield gain over all environments with breeding cycle depending on the environment chosen for the QTL analysis. It was suggested that knowledge resulting from trait physiology and modelling would overcome this dependency by identifying stable QTLs. The improved predictive power would increase the utility of the QTLs in MAS. Developing and implementing this gene-to-phenotype capability in crop improvement requires enhanced attention to phenotyping, ecophysiological modelling, and validation studies to test the stability of candidate QTLs.

Modelling the dynamics of genetic algorithms using statistical mechanics

A formalism for modelling the dynamics of Genetic Algorithms (GAs) using methods from statistical mechanics, originally due to Prugel-Bennett and Shapiro, is reviewed, generalized and improved upon. This formalism can be used to predict the averaged trajectory of macroscopic statistics describing the GA's population. These macroscopics are chosen to average well between runs, so that fluctuations from mean behaviour can often be neglected. Where necessary, non-trivial terms are determined by assuming maximum entropy with constraints on known macroscopics. Problems of realistic size are described in compact form and finite population effects are included, often proving to be of fundamental importance. The macroscopics used here are cumulants of an appropriate quantity within the population and the mean correlation (Hamming distance) within the population. Including the correlation as an explicit macroscopic provides a significant improvement over the original formulation. The formalism is applied to a number of simple optimization problems in order to determine its predictive power and to gain insight into GA dynamics. Problems which are most amenable to analysis come from the class where alleles within the genotype contribute additively to the phenotype. This class can be treated with some generality, including problems with inhomogeneous contributions from each site, non-linear or noisy fitness measures, simple diploid representations and temporally varying fitness. The results can also be applied to a simple learning problem, generalization in a binary perceptron, and a limit is identified for which the optimal training batch size can be determined for this problem. The theory is compared to averaged results from a real GA in each case, showing excellent agreement if the maximum entropy principle holds. Some situations where this approximation brakes down are identified. In order to fully test the formalism, an attempt is made on the strong sc np-hard problem of storing random patterns in a binary perceptron. Here, the relationship between the genotype and phenotype (training error) is strongly non-linear. Mutation is modelled under the assumption that perceptron configurations are typical of perceptrons with a given training error. Unfortunately, this assumption does not provide a good approximation in general. It is conjectured that perceptron configurations would have to be constrained by other statistics in order to accurately model mutation for this problem. Issues arising from this study are discussed in conclusion and some possible areas of further research are outlined.