Publications

Optimization in an Error Backpropagation Neural Network Environment with a Performance Test on a Pattern Classification Problem

Various techniques of optimizing the multiple class cross-entropy error function to train single hidden layer neural network classifiers with softmax output transfer functions are investigated on a real-world multispectral pixel-by-pixel classification problem that is of fundamental importance in remote sensing. These techniques include epoch-based and batch versions of backpropagation of gradient descent, PR-conjugate gradient and BFGS quasi-Newton errors. The method of choice depends upon the nature of the learning task and whether one wants to optimize learning for speed or generalization performance. It was found that, comparatively considered, gradient descent error backpropagation provided the best and most stable out-of-sample performance results across batch and epoch-based modes of operation. If the goal is to maximize learning speed and a sacrifice in generalisation is acceptable, then PR-conjugate gradient error backpropagation tends to be superior. If the training set is very large, stochastic epoch-based versions of local optimizers should be chosen utilizing a larger rather than a smaller epoch size to avoid inacceptable instabilities in the generalization results.

EVALUATING NEURAL SPATIAL INTERACTION. MODELLING BY BOOTSTRAPPING

This paper exposes problems of the commonly used technique of splitting the available data in neural spatial interaction modelling into training, validation, and test sets that are held fixed and warns about drawing too strong conclusions from such static splits. Using a bootstrapping procedure, we compare the uncertainty in the solution stemming from the data splitting with model specific uncertainties such as parameter initialization. Utilizing the Austrian interregional telecommunication traffic data and the differential evolution method for solving the parameter estimation task for a fixed topology of the network model [ i.e. J = 9] this paper illustrates that the variation due to different resamplings is significantly larger than the variation due to different parameter initializations. This result implies that it is important to not over-interpret a model, estimated on one specific static split of the data.

Innovation, Networks, and Knowledge Spillovers: Selected Essays

Technological change and the nature of the firm.- Innovation and Technological Change.- Innovation and Technological Change: An Austrian-British Comparison.- Technology, Organisation and Export-driven Research and Development in Austria's Electronics Industry.- Information Processing, Technological Progress, and Retail Market Dynamics.- Innovation and Network Activities.- The New Economy and Networking.- The Innovation Process and Network Activities of Manufacturing Firms.- Knowledge Interactions between Universities and Industry in Austria: Sectoral Patterns and Determinants.- Knowledge Creation, Diffusion and Spillovers.- Innovation, Knowledge Creation and Systems of Innovation.- The Role of Space in the Creation of Knowledge in Austria: An Exploratory Spatial Data Analysis.- Spatial Knowledge Spillovers and University Research: Evidence from Austria.- Patents, Patent Citations and the Geography of Knowledge Spillovers in Europe.

Dual X-ray absorptiometry—cross-calibration and normative reference ranges for the spine: Results of a European Community Concerted Action

Bone density measurements by dual X-ray absorptiometry (DXA) of the spine can now be made precisely, but there is no uniformity in reporting results and in presenting reference data. A European Union Concerted Action therefore devised a uniform procedure for cross-calibrating and standardizing instruments, using the European spine phantom (ESP) prototype. This phantom differs in a number of respects from the final version of the ESP. Eighteen centers in nine countries obtained 1619 records (1035 women) from Caucasian subjects, aged 20–80 years, drawn from normal populations. The DXA machines used were made by the Hologic, Lunar, and Norland companies. Highly statistically significant differences were evident between populations, both in apparent rates of bone loss with age and in the spread of values about the age-adjusted means. There were small residual differences in the results obtained with the three machine brands which could have been due to the relatively large between-center population differences we observed. The alternative or additional explanation that they were attributable, in part, to the design differences between the ESP prototype and the definitive ESP, which became available after this study was completed, was shown to be a valid possibility. Results from postmenopausal women reported in relation to the years that have elapsed since menopause showed reduced population variance when compared with conventional reporting in relation to age. After cross-calibration, the center with the highest age-adjusted normal density value averaged 23% more than the center with the lowest. It is therefore crucially important to select appropriate reference data in clinical and epidemiological studies. These results provide a basis for designing protocols for multicenter studies using currently installed densitometers. (Bone 17:247–254; 1995)

Regional Income Convergence in Europe, 1995-2000: A Spatial Econometric Perspective

Questions of convergence have received increasing attention in recent years, in light of the pressure for greater integration and enlargement of the European Union [EU] to countries in Central and Eastern Europe [CEE]. This paper looks at the evidence for convergence of per capita income between regions in Europe in the second half of the 1990s, when economic recovery in CEE gathered pace. The analysis is based on the simplest of the models, the unconditional s-convergence model and shows that the classical test methodology is ill-designed due to two reasons. First, it cannot identify groupings of regional economies that are converging. Second, it neglects spatial effects that represent interregional interactions and spatial spillovers. The paper suggests a much richer and theoretically more satisfactory approach that is in line with both the notions of club convergence and spatial dependence, and reflects recent developments in spatial econometrics. The two-club spatial error convergence model with groupwise heteroskedasticity is found to be most appropriate for the data at hand. Two empirical key findings are worthwhile to note. The first is that the data provide much support for unconditional beta-convergence in Europe. The second is that the usual convergence conclusions hold. But they do so for reasons that are not revealed by the classical test equation that is typical in mainstream economics literature.

Systems of Innovation:A Novel Conceptual Frameworkfor Innovation Analysis

The objective of this paper is to increase our ability to understand the systems of innovation approach as a flexible and useful conceptual framework for innovation analysis. It presents an effort to develop some missing links and to decrease the conceptual noise often present in the discussions on national innovation systems. The paper specifies elements and relations that seem to be essential to the conceptual core of the framework and argues that there is no a priori reason to emphasise the national over the subnational [regional] scale as an appropriate mode for analysis, irrespective of time and place. Localised input-output relations between the actors of the system, knowledge spillovers and their untraded interdependencies lie at the centre of the arguments. The paper is organised as follows. It introduces the reader, first, to some basic elements and concepts that are central to understanding the approach. The characteristics of the innovation process are examined: its nature, sources and some of the factors shaping its development. Particular emphasis is laid on the role of knowledge creation and dissemination based on the fundamental distinction between codified and tacit forms. These concepts recur throughout the paper and particularly in discussions on the nature and specifications of the systems approach. The paper concludes by summarising some of the major findings of the discussion and pointing to some directions for future research activities. (author's abstract)

Middle Managers - Leading for Performance The case of a major Australian retail business

The Centre for Workplace Leadership was invited by a major Australian retail business to conduct research on the company. The CEO wished to have a solid basis of evidence on which to improve communication, innovation and decisionmaking in the company. The research took place over a year between mid-2014 and mid-2015. It involved interviews with staff at all levels, from CEO to frontline business staff. Key Findings: The analysis of communication, innovation and decision-making processes found three main issues that were impacting organisational effectiveness. These were: • shifting to hierarchical leadership had reduced employee engagement; • increased bureaucracy had reinforced organisational silos; • top-down decision-making had crowded out collaboration and innovation. Each of these findings is described in more detail below, along with quotes from the interviews.

2022 JGS best paper award and the editors’ choice paper volume 25(1)

Expert Systems and Artificial Neural Networks for Spatial Analysis and Modelling. Essential Components for Knowledge-Based Geographical Information Systems

This article outlines the general architecture of a knowledge based GISystem that has the potential to intelligently support decision making in a GIS environment. The efficient and effective integration of spatial data, spatial analytic procedures and models, procedural and declarative knowledge is through fuzzy logic, expert systems and neural network technologies. A specific focus of the discussion is on the expert system and neural network components of the system, technologies which had been relatively unknown in the GIS community at the time this chapter was written.

[Osteotomy of tuberositas tibiae in supracondylar femoral fractures].

No abstract is provided for this article.

Fabrication and measurement of optical freeforms made of fused silica with novel manufacturing technologies

The production of freeform components is a promising field for numerous optical applications. The utilization of novel manufacturing processes is shown as well as the surface quality measurement on the example of a steep freeform-geometry.

The regional transmission of uncertainty shocks on income inequality in the United States

This paper explores the relationship between household income inequality and macro-economic uncertainty in the United States. Using a novel large-scale macroeconometric model, we shed light on regional disparities of inequality responses to a national uncertainty shock. The results suggest that income inequality decreases in most states, with a pronounced degree of heterogeneity in terms of the dynamic responses. By contrast, some few states, mostly located in the Midwest, display increasing levels of income inequality over time. Forecast error variance and historical decompositions highlight the importance of uncertainty shocks in explaining income inequality in most regions considered. Finally, we explain differences in the responses of income inequality by means of a simple regression analysis. These regressions reveal that the income composition as well as labor market fundamentals determine the directional pattern of the dynamic responses.

A novel modular product unit neural network for modelling constrained spatial interaction flows

A novel product unit neural network approach is presented for modelling origin constrained spatial interaction flows. We adopt the Alopex procedure, a global search procedure, in combination with the bootstrapping pairs approach to address the issue of maximum likelihood estimation of the parameters. A benchmark comparison illustrates the generalisation performance of the model approach in terms of Kullback and Leibler's information criterion.

Entrepreneurship and Regional Development

This chapter offers a review on modern entrepreneurship analysis, against the background of regional development. After a conceptual discussion on the importance and the measurement of entrepreneurship, the contribution discusses success factors and determinants of entrepreneurships. Next, focus is laid on the geography of entrepreneurship, while also due attention is paid to the relevance of networks for modern entrepreneurship. The chapter concludes with some retrospective and prospective remarks.

A MULTIREGIONAL LABOUR SUPPLY MODEL FOR AUSTRIA: THE EFFECTS OF DIFFERENT REGIONALISATIONS IN MULTIREGIONAL LABOUR MARKET MODELLING

The design of a spatial Framework in multi- and interregional modelling is a crucial element of the research process. In this paper an attempt is made to present some empirical evidence with respect to two hypotheses, the scale hyothesis and the aggregation hypothesis. This will be achieved by estimating the parameters and testing the performance of a multiregional labour supply model for Austria. Five regionalisation approaches are specified for delineating (functional) labour market regions. The paper shows that it is by no means admissible to ignore possible effects of the spatial representation choice. The fundamental question of which regionalisation should be chosen should be decided on the basis of relevant evaluation criteria.

Hemobahn-Endoprothese: 4-Jahres-Erfahrungen mit der perkutanen Anwendung bei Stenosen und Verschlüssen der Arteria femoralis superficialis

In einer prospektiven, nichtrandomisierten klinischen Studie wurde der Wert der Hemobahn-Endoprothese für die Behandlung von Verschlüssen der Arteria femoralis superficialis untersucht. In den letzten vier Jahren wurden bei 52 Patienten an 55 Extremitäten insgesamt 78 Hemobahn-Endoprothesen zur Behandlung von Stenosen (15 %) oder Verschlüssen (85 %) der Arteria femoralis superficialis perkutan platziert. Die mittlere Verschlusslänge betrug 11,6 cm, die mittlere Länge der platzierten Hemobahn 15,9 cm und der mittlere Endoprothesendurchmesser 6,1 mm. In 45 % der Fälle wurde in Cross-over-Technik, in 55 % der Fälle ipsilateral antegrad vorgegangen. Klinisch bestanden die Fontaine-Stadien: II b (89 %), III (2 %) und IV (9 %); 15 Patienten waren Diabetiker. Die Patienten wurden primär und in jährlichen Intervallen angiologisch kontrolliert; das mittlere Nachbeobachtungsintervall betrug 33 Monate. Nach 1 Jahr „Follow-up” waren die primären/sekundären Offenheitsraten bezogen auf alle nachuntersuchten Extremitäten 64/78 % (n = 50), nach 2 Jahren 61/76 % (n = 46), nach 3 Jahren 53/66 % (n = 32) und nach 4 Jahren 29/50 % (n = 14). Nach den ersten Erfahrungen von 1998 mit 10 Verschlüssen bei 14 insgesamt behandelten Extremitäten wurden die Indikationen geprüft und nachfolgend lediglich bei definierter, so genannter „Idealindikation” mit der Hemobahn behandelt. Als „Idealindikationen” galten: ≥ 1 cm unauffälliges vor- und nachgeschaltetes Gefäßsegment, fehlende Läsion im 1. Poplitealsegment, mindestens eine offene Unterschenkelarterie und das Fehlen ausgeprägter Verkalkungen. Bei diesen Patienten betrugen die primären/sekundären Offenheitsraten nach 1 Jahr 76/90 % (n = 36), nach 2 Jahren 74/89 % (n = 32) und nach 3 Jahren 71/83 % (n = 24); nach 4 Jahren waren von 8 behandelten Arterien 4 primär und 7 sekundär offen. Die Hemobahn- Endoprothese eignet sich zur Therapie ausgewählter komplexer femoraler Gefäßläsionen und längerstreckiger Verschlüsse. Die mittelfristige Offenheitsrate liegt über jener der alleinigen Ballondilatation oder Stenttherapie unter Nutzung ungecoverter Stents.

The Speed of Income Convergence in Europe: A case for Bayesian Model Averaging with Eigenvector Filtering

The speed of income convergence in Europe remains one of the hot topics in regional economics. Recently Bayesian Model Averaging (BMA) applied to spatial autoregressive models seems to have gained more popularity. BMA averages over some predetermined number of so called top models, ranked by the model's posterior likelihood. We regard two approaches for especially noteworthy: First Crespo-Cuaresma and Feldkircher (2012) employ BMA to a spatial autoregressive model, where spatial eigenvector filtering is used in order to tackle the econometric problems caused by the spatial lag. However, spatial filtering has its drawbacks. It relies on a model approximation and no partial derivatives of interest associated with the model can be computed. This means, that it is impossible to derive direct and indirect effects. Second LeSage and Fischer (2008) rely on BMA applied to a Spatial Durbin Model (SDM), where the model posterior is calculated without any model approximation. Although it can be computationally burdensome, it allows for a proper model interpretation if the underlying data generating process (DGP) is of SDM form. One virtue of spatial filtering, as shown by Pace et al. (2011), is that it estimates some of the model coefficients efficiently for various spatial autocorrelated DGPs. Hence, the likelihoods associated with spatial filtering are more robust against model misspecification. Since our preliminary results show that the top models' (posterior) likelihoods obtained from a spatial filtering BMA exercise and (non spatial filtering) BMA applied to a Spatial Durbin Model differ, it is most likely that the DGP is not of SDM form, i.e. misspecified. This leads us to the conclusion that, even though the methodology employed by Crespo-Cuaresma and Feldkircher (2012) cannot be used for a proper model interpretation, the results obtained by spatial filtering BMA do not suffer from model misspecification. References: Crespo-Cuaresma, J., Feldkircher, M. (2012), `Spatial Filtering, Model Uncertainty and the Speed of Income Convergence in Europe', Journal of Applied Econometrics, forthcoming. LeSage, J., Fischer, M. (2008), `Spatial Growth Regressions, Model Specification, Estimation, and Interpretation', Spatial Economic Analysis 3, 275-304. Pace, R., LeSage, J., Zhu, S. (2011), `Interpretation and Computation of Estimates from Regression Models using Spatial Filtering', written for Spatial Econometrics Association 2011. Keywords: Model uncertainty, spatial ltering, determinants of economic growth, European regions. JEL Classi cations: C11, C21, O11, R11

Local Treatment of Inflammatory Joint Diseases

This book presents a systematic survey of different local treatments in inflammatory joint diseases and their potential side effects and complications. Both surgical therapies and pharmaceutical strategies involving local or systemic application of anti-inflammatory and immunomodulating drugs are addressed. Special emphasis is placed on nuclear medicine therapy by intra-articular instillation of beta radiation emitters, known as radiosynoviorthesis or radiation synovectomy. Possible complications, in particular radionecrosis, infection, and thromboembolism, are described and available treatment strategies are examined in detail. In addition, radiation safety considerations are critically reviewed and relevant clinical questions are discussed.While radiosynoviorthesis is an effective and safe treatment if a proper indication is secured and if the application is performed by an experienced nuclear medicine physician, local complications can never be excluded with certainty. This book will support physicians who are engaged in the care and treatment of patients with inflammatory joint disease, approximately 30,000 of whom undergo radiosynoviorthesis in Europe each year.