Model selection approaches for non-linear system identification: a review

The identification of non-linear systems using only observed finite datasets has become a mature research area over the last two decades. A class of linear-in-the-parameter models with universal approximation capabilities have been intensively studied and widely used due to the availability of many linear-learning algorithms and their inherent convergence conditions. This article presents a systematic overview of basic research on model selection approaches for linear-in-the-parameter models. One of the fundamental problems in non-linear system identification is to find the minimal model with the best model generalisation performance from observational data only. The important concepts in achieving good model generalisation used in various non-linear system-identification algorithms are first reviewed, including Bayesian parameter regularisation and models selective criteria based on the cross validation and experimental design. A significant advance in machine learning has been the development of the support vector machine as a means for identifying kernel models based on the structural risk minimisation principle. The developments on the convex optimisation-based model construction algorithms including the support vector regression algorithms are outlined. Input selection algorithms and on-line system identification algorithms are also included in this review. Finally, some industrial applications of non-linear models are discussed.

Integrated structure selection and parameter optimisation for eng-genes neural models

The eng-genes concept involves the use of fundamental known system functions as activation functions in a neural model to create a 'grey-box' neural network. One of the main issues in eng-genes modelling is to produce a parsimonious model given a model construction criterion. The challenges are that (1) the eng-genes model in most cases is a heterogenous network consisting of more than one type of nonlinear basis functions, and each basis function may have different set of parameters to be optimised; (2) the number of hidden nodes has to be chosen based on a model selection criterion. This is a mixed integer hard problem and this paper investigates the use of a forward selection algorithm to optimise both the network structure and the parameters of the system-derived activation functions. Results are included from case studies performed on a simulated continuously stirred tank reactor process, and using actual data from a pH neutralisation plant. The resulting eng-genes networks demonstrate superior simulation performance and transparency over a range of network sizes when compared to conventional neural models. (c) 2007 Elsevier B.V. All rights reserved.

Genetic programming based formulation for fresh and hardened properties of self-compacting concrete containing pulverised fuel ash

Self-compacting concrete (SCC) flows into place and around obstructions under its own weight to fill the formwork completely and self-compact without any segregation and blocking. Elimination of the need for compaction leads to better quality concrete and substantial improvement of working conditions. This investigation aimed to show possible applicability of genetic programming (GP) to model and formulate the fresh and hardened properties of self-compacting concrete (SCC) containing pulverised fuel ash (PFA) based on experimental data. Twenty-six mixes were made with 0.38 to 0.72 water-to-binder ratio (W/B), 183–317 kg/m³ of cement content, 29–261 kg/m³ of PFA, and 0 to 1% of superplasticizer, by mass of powder. Parameters of SCC mixes modelled by genetic programming were the slump flow, JRing combined to the Orimet, JRing combined to cone, and the compressive strength at 7, 28 and 90 days. GP is constructed of training and testing data using the experimental results obtained in this study. The results of genetic programming models are compared with experimental results and are found to be quite accurate. GP has showed a strong potential as a feasible tool for modelling the fresh properties and the compressive strength of SCC containing PFA and produced analytical prediction of these properties as a function as the mix ingredients. Results showed that the GP model thus developed is not only capable of accurately predicting the slump flow, JRing combined to the Orimet, JRing combined to cone, and the compressive strength used in the training process, but it can also effectively predict the above properties for new mixes designed within the practical range with the variation of mix ingredients.

Effect of Mix Proportions on Rheological and Hardened Properties of Composite Cement Pastes

There is an increasing need to identify the effect of mix composition on the rheological properties of composite cement pastes using simple tests to determine the fluidity, the cohesion and other mechanical properties of grouting applications such as compressive strength. This paper reviews statistical models developed using a fractional factorial design which was carried out to model the influence of key parameters on properties affecting the performance of composite cement paste. Such responses of fluidity included mini-slump, flow time using Marsh cone and cohesion measured by Lombardi plate meter and unit weight, and compressive strength at 3 d, 7 d and 28 d. The models are valid for mixes with 0.35 to 0.42 water-to-binder ratio (W/B), 10% to 40% of pulverised fuel ash (PFA) as replacement of cement by mass, 0.02 to 0.06% of viscosity enhancer admixture (VEA), by mass of binder, and 0.3 to 1.2% of superplasticizer (SP), by mass of binder. The derived models that enable the identification of underlying primary factors and their interactions that influence the modelled responses of composite cement paste are presented. Such parameters can be useful to reduce the test protocol needed for proportioning of composite cement paste. This paper attempts also to demonstrate the usefulness of the models to better understand trade-offs between parameters and compare the responses obtained from the various test methods which are highlighted. The multi parametric optimization is used in order to establish isoresponses for a desirability function of cement composite paste. Results indicate that the replacement of cement by PFA is compromising the early compressive strength and up 26%, the desirability function decreased.

Coupled heat and moisture transfer in multi-layer building materials

A dynamic mathematical model for simulating the coupled heat and moisture migration through multilayer porous building materials was proposed. Vapor content and temperature were chosen as the principal driving potentials. The discretization of the governing equations was done by the finite difference approach. A new experimental set-up was also developed in this study. The evolution of transient temperature and moisture distributions inside specimens were measured. The method for determining the temperature gradient coefficient was also presented. The moisture diffusion coefficient, temperature gradient coefficient, sorption–desorption isotherms were experimentally evaluated for some building materials (sandstone and lime-cement mortar). The model was validated by comparing with the experimental data with good agreement. Another advantage of the method lies in the fact that the required transport properties for predicting the non-isothermal moisture flow only contain the vapor diffusion coefficient and temperature gradient coefficient. They are relatively simple, and can be easily determined.

Two-stage RBF network construction based on particle swarm optimization

The conventional radial basis function (RBF) network optimization methods, such as orthogonal least squares or the two-stage selection, can produce a sparse network with satisfactory generalization capability. However, the RBF width, as a nonlinear parameter in the network, is not easy to determine. In the aforementioned methods, the width is always pre-determined, either by trial-and-error, or generated randomly. Furthermore, all hidden nodes share the same RBF width. This will inevitably reduce the network performance, and more RBF centres may then be needed to meet a desired modelling specification. In this paper we investigate a new two-stage construction algorithm for RBF networks. It utilizes the particle swarm optimization method to search for the optimal RBF centres and their associated widths. Although the new method needs more computation than conventional approaches, it can greatly reduce the model size and improve model generalization performance. The effectiveness of the proposed technique is confirmed by two numerical simulation examples.

Robust automated tumour segmentation on histological and immunohistochemical tissue images

Tissue microarray (TMA) is a high throughput analysis tool to identify new diagnostic and prognostic markers in human cancers. However, standard automated method in tumour detection on both routine histochemical and immunohistochemistry (IHC) images is under developed. This paper presents a robust automated tumour cell segmentation model which can be applied to both routine histochemical tissue slides and IHC slides and deal with finer pixel-based segmentation in comparison with blob or area based segmentation by existing approaches. The presented technique greatly improves the process of TMA construction and plays an important role in automated IHC quantification in biomarker analysis where excluding stroma areas is critical. With the finest pixel-based evaluation (instead of area-based or object-based), the experimental results show that the proposed method is able to achieve 80% accuracy and 78% accuracy in two different types of pathological virtual slides, i.e., routine histochemical H&E and IHC images, respectively. The presented technique greatly reduces labor-intensive workloads for pathologists and highly speeds up the process of TMA construction and provides a possibility for fully automated IHC quantification.

‘What Adult Worker Model? A Critical Look at Recent Social Policy Reform in Europe from a Gender and Family Perspective

Analyses regularly feature claims that European welfare states are in the process of creating an adult worker model. The theoretical and empirical basis of this argument is examined here by looking first at the conceptual foundations of the adult worker model formulation and then at the extent to which social policy reform in western Europe fits with the argument. It is suggested that the adult worker formulation is under-specified. A framework incorporating four dimensions—the treatment of individuals vis-à-vis their family role and status for the purposes of social rights, the treatment of care, the treatment of the family as a social institution, and the extent to which gender inequality is problematized—is developed and then applied. The empirical analysis reveals a strong move towards individualization as social policy promotes and valorizes individual agency and self-sufficiency and shifts some childcare from the family. Yet evidence is also found of continued (albeit changed) familism. Rather than an unequivocal move to an individualized worker model then, a dual earner, gender-specialized, family arrangement is being promoted. The latter is the middle way between the old dependencies and the new “independence.” This makes for complexity and even ambiguity in policy, a manifestation of which is that reform within countries involves concurrent moves in several directions.

The effect of construction pattern and unit interlock on the structural behaviour of block pavements

The maintenance or even replacement of cracked pavements requires considerable financial resources and puts a large burden on the budgets of local councils. In addition to these costs, local councils also face liability claims arising from uneven or cracked pedestrian pavements. These currently cost the Manchester City Council and Preston City Council around £6 million a year each. Design procedures are empirical. A better understanding of the interaction between paving blocks, bedding sand and subbase was necessary in order to determine the mode of failure of pavements under load. Increasing applied stress was found to mobilise ‘‘rotational interlock’’, providing increased pavement stiffness and thus increased load dissipation resulting in lower transmitted stress on the subgrade. The indications from the literature
review were that pavements are designed to fail by excessive deformation and that paving blocks remained uncracked at failure. This was confirmed with experimental data which was obtained from tests on segments of pavements that were laid/constructed in a purpose built test frame in the laboratory.

The Holocene vegetation cover of Britain and Ireland: overcoming problems of scale and discerning patterns of openness

The vegetation of Europe has undergone substantial changes during the course of the Holocene epoch, resulting from range expansion of plants following climate amelioration, competition between taxa and disturbance through anthropogenic activities. Much of the detail of this pattern is understood from
decades of pollen analytical work across Europe, and this understanding has been used to address questions relating to vegetation-climate feedback, biogeography and human impact. Recent advances in modelling the relationship between pollen and vegetation now make it possible to transform pollen
proportions into estimates of vegetation cover at both regional and local spatial scales, using the Landscape Reconstruction Algorithm (LRA), i.e. the REVEALS (Regional Estimates of VEgetation Abundance from Large Sites) and the LOVE (LOcal VEgetation) models. This paper presents the compilation and analysis of 73 pollen stratigraphies from the British Isles, to assess the application of the LRA and describe the pattern of landscape/woodland openness (i.e. the cover of low herb and bushy vegetation) through the Holocene. The results show that multiple small sites can be used as an effective replacement for a single large site for the reconstruction of regional vegetation cover. The REVEALS vegetation estimates imply that the British Isles had a greater degree of landscape/woodland openness at the regional scale than areas on the European mainland. There is considerable spatial bias in the British Isles dataset towards wetland areas and uplands, which may explain higher estimates of landscape openness compared with Europe. Where multiple estimates of regional vegetation are available from within the same region inter-regional differences are greater than intra-regional differences, supporting the use of the REVEALS model to the estimation of regional vegetation from pollen data.

Homological Localisation of Model Categories

One of the most useful methods for studying the stable homotopy category is localising at some spectrum E. For an arbitrary stable model category we introduce a candidate for the E–localisation of this model category. We study the properties of this new construction and relate it to some well–known categories.

The EU economic and social model in the global crisis: Interdisciplinary perspectives:Interdisciplinary perspectives

This book addresses the viability of the EU economic and social model within and after the global economic crisis. It identifies four key issues which warrant further discussion: (1) the asymmetry of the legal and policy framework of the euro and potential recalibration; (2) substantive tensions between the EU 'economic constitution' and its normative aim of social justice and impacts on national policy; (3) the role of civil society, including the two sides of industry in overcoming these tensions and (4) the EU's global aspirations towards the creation of a viable socio-economic model. Its chapters offer two perspectives on each of the four main issues. In drawing these debates together, the book provides a broad understanding as well as starting points for future research. Bringing together different disciplinary approaches, ranging from legal studies to political economy, sociology and macroeconomics, it is a valuable contribution to the debate on the European social model and introduces new insights by focusing on legal and political tensions, the impact of the financial crisis and other economic contexts as well as global dimensions. 

The performance of concrete exposed to marine environments: predictive modelling and use of laboratory/on site test methods

This paper reports an approach by which laboratory based testing and numerical modelling can be combined to predict the long term performance of a range of concretes exposed to marine environments. Firstly, a critical review of the test methods for assessing the chloride penetration resistance of concrete is given. The repeatability of the different test results is also included. In addition to the test methods, a numerical simulation model is used to explore the test data further to obtain long-term chloride ingress trends. The combined use of testing and modelling is validated with the help of long-term chloride ingress data from a North Sea exposure site. In summary, the paper outlines a methodology for determining the long term performance of concrete in marine environments.

Transform both sides model: A parametric approach

A parametric regression model for right-censored data with a log-linear median regression function and a transformation in both response and regression parts, named parametric Transform-Both-Sides (TBS) model, is presented. The TBS model has a parameter that handles data asymmetry while allowing various different distributions for the error, as long as they are unimodal symmetric distributions centered at zero. The discussion is focused on the estimation procedure with five important error distributions (normal, double-exponential, Student's t, Cauchy and logistic) and presents properties, associated functions (that is, survival and hazard functions) and estimation methods based on maximum likelihood and on the Bayesian paradigm. These procedures are implemented in TBSSurvival, an open-source fully documented R package. The use of the package is illustrated and the performance of the model is analyzed using both simulated and real data sets.

Rural engineering infrastructures design and public facility locations

O desenvolvimento das redes de estradas rurais, especialmente em áreas montanhosas, é a intervenção chave para melhorar a acessibilidade às localidades e aos serviços públicos, cobrindo o maior número de localidades e de serviços públicos, otimizando os escassos recursos disponíveis em países em desenvolvimento. Este estudo explora diferentes modelos de organização de redes de estradas rurais considerando a construção de novas ligações ou o melhoramento de estradas existentes. Um método, baseado na cobertura da rede de estradas rurais, é utilizado para identificar os pontos nodais que formam a rede rural base numa específica região, a qual cobrirá um conjunto dos serviços públicos e de localidades. O modelo assenta numa rede rural de estradas típica ("backbone" e "branch") das regiões montanhosas do Nepal. Os modelos propostos fornecem um conjunto de possibilidades de ligações a estabelecer ou a melhorar e oferece soluções para diferentes níveis de orçamento, que otimizam os custos de transporte na rede, considerando diferentes tipos de pavimento (em solo, granular ou asfáltico). Foi realizado separadamente um modelo dedicado a análises multi-objetivo para resolver problemas de melhoramento de ligações dentro da rede considerando dois objectivos, minimizar os custos de operação para o utilizador e maximizar a população coberta pela rede de estradas, considerando ligações pavimentadas e não pavimentadas (em solo, granular ou asfáltico) dentro de um determinado limite orçamental. O modelo dá ao decisor (DM) diferentes alternativas eficientes para que este possa tomar uma decisão final. Estes modelos, desenvolvidos para redes de estradas rurais, são também aplicáveis a outras redes de infraestruturas rurais, tais como, de fornecimento de água, de eletricidade e de telecomunicações. A implementação dos modelos nas redes de estradas rurais dos distritos de Gorkha e Lamjung do Nepal permitiu confirmara sua aplicabilidade. Verifica-se que os modelos propostos são mais práticos e realísticos no estudo de soluções de melhoramento e de desenvolvimento de redes de estradas rurais em regiões montanhosas de países em desenvolvimento.