873 publications from this institution
In this paper we view learning as an unconstrained non-linear minimization problem in which the objective function is defined by the negative log-likelihood function and the search space by the parameter space of an origin constrained product unit neural spatial interaction model. We consider Alopex based global search, as opposed to local search based upon backpropagation of gradient descents, each in combination with the bootstrapping pairs approach to solve the maximum likelihood learning problem. Interregional telecommunication traffic flow data from Austria are used as test bed for comparing the performance of the two learning procedures. The study illustrates the superiority of Alopex based global search, measured in terms of Kullback and Leibler's information criterion.
The main aim of the present paper is to survey some major trends in current research in the field of discrete choice modelling, with particular emphasis on dynamic approaches. The paper is organized as follows. Section 2 provides a brief overview of static disaggregate choice modelling and random utility maximization, based inter alia on multinomial logit and/or probit models, generalized extreme value models and nested logit models. Particular attention is given here to model representation issues, sampling and estimation issues and model performance issues. Next, section 3 is devoted to some recent developments in the rapidly growing new field of dynamic discrete choice modelling. In contrast to stochastic panel data models of buying behaviour, dynamic discrete choice models incorporate explanatory variables and take adaptive behaviour explicitly into account (i.e., the effect of past experience on choice behaviour). Several dynamic discrete choice model approaches are summarized. Special attention is paid to the seminal work of Heckman. In the final section, complementary and alternative approaches to dynamic choice modelling are discussed, such as the human activity constraint approach, the computational process modelling approach and the master equation approach. It is concluded that contextual effects, multi-actor or synergetic interactions and shifting individual preferences based on learning principles are of primary importance in dynamic discrete choice modelling.
Inter-organizational linkages, often referred to as network relationships, are considered to be of increasing importance for the competitive performance of firms, industries and nations. Two hypotheses about the conditions for the emergence of network relationships are derived from the transaction cost approach and discussed in view of two case studies relying on a medium sized machine-tool firm and a medium sized cement firm in Austria. The paper clearly illustrates the necessity to go beyond the transaction cost approach and to take into account factors such as strategic orientation, management skills and organizational issues. (authors' abstract)
No abstract is provided for this article.
Parameter estimation is one of the central issues in neural spatial interaction modelling. Current practice is dominated by gradient based local minimization techniques. They find local minima efficiently and work best in unimodal minimization problems, but can get trapped in multimodal problems. Global search procedures provide an alternative optimization scheme that allows to escape from local minima. Differential evolution has been recently introduced as an efficient direct search method for optimizing real-valued multi-modal objective functions (Storn and Price 1997). The method is conceptually simple and attractive, but little is known about its behaviour in real world applications. This paper explores this method as an alternative to current practice for solving the parameter estimation task, and attempts to assess ist robustness, measured in terms of in-sample and out-of-sample performance. A benchmark comparison against backpropagation of conjugate gradients is based on Austrian interregional telecommunication traffic data. (authors' abstract)
We report a prospective, stratified study of 60 PCA-cups and 60 RM-polyethylene cups which have been followed for a median time of 90 months, with annual radiography. The radiological migration of cups was measured by the computer-assisted EBRA method. A number of threshold migration rates from 1 mm in the first year to 1 mm in five years have been assessed and related to clinically determined revision rates. A total of 28 cups showed a total migration of 1 mm or more within the first two years; 13 of these cups have required revision and been exchanged. The survival curves of cups which had previously shown early migration were considerably different from those without early migration. For cups with a migration of less than 1 mm within the first two years the mean survival at 96 months was 0.96 +/- 0.02; for migrating cups, it was 0.63 +/- 0.11 (log-rank test, p=0.0001; chi-square value=39.4). Early migration is a good predictor for late loosening of hip sockets.
Spatial interaction models of the gravity type are widely used to describe origin-destination flows. They draw attention to three types of variables to explain variation in spatial interactions across geographic space: variables that characterize the origin region of interaction, variables that characterize the destination region of interaction, and variables that measure the separation between origin and destination regions. A violation of standard minimal assumptions for least squares estimation may be associated with two problems: spatial autocorrelation within the residuals, and spatial autocorrelation within explanatory variables. This paper compares a spatial econometric solution with the spatial statistical Moran eigenvector spatial filtering solution to accounting for spatial autocorrelation within model residuals. An example using patent citation data that capture knowledge flows across 257 European regions serves to illustrate the application of the two approaches.
No abstract is provided for this article.
Spectral pattern recognition deals with classifications that utilize pixel-by-pixel spectral information from satellite imagery. The literature on neural network applications in this area is relatively new, dating back only about six to seven years. The first studies...
Unknown
The focus of this study is on regional knowledge production in Europe, with special emphasis on the interplay between intra- and inter-regional research collaboration. The objective is to identify and measure effects of research collaboration on knowledge production...
The focus of this paper is on the neural network modelling approach that has gained increasing recognition in GIScience in recent years. The novelty about neural networks lies in their ability to model non-linear processes with few, if any, a priori assumptions about the nature of the data-generating process. The paper discusses some important issues that are central for successful application development. The scope is limited to feedforward neural networks, the leading example of neural networks. It is argued that failures in applications can usually be attributed to inadequate learning and/or inadequate complexity of the network model. Parameter estimation and a suitably chosen number of hidden units are, thus, of crucial importance for the success of real world neural network applications. The paper views network learning as an optimization problem, reviews two alternative approaches to network learning, and provides insights into current best practice to optimize complexity so to perform well on generalization tasks.
In this paper, we distinguish three constrained variants of the gravity model of spatial interaction: doubly constrained, production constrained and attraction constrained exponential gravity models. These model variants include origin- and/or destination-specific balancing factors that act as constraints to ensure that the estimated rows and columns of the flow data matrix sum to the observed row and column totals. Because flows are typically counts, the Poisson rather than the normal probability model specification furnishes the appropriate statistical distribution, and parameter estimation can be achieved via Poisson regression. This probability model specification motivates the use of origin and/or destination fixed effects or—under certain conditions—the use of origin- and/or destination-specific random effects for model estimation. The paper establishes theoretical connections between balancing factors, fixed effects represented by binary indicator variables and random effects. The results pertaining to both the doubly and singly constrained cases of spatial interaction are illustrated with an empirical example while accounting for spatial dependence between flows from locations neighbouring both the origins and destinations during estimation.
This article reviews some of the special properties of spatial data and the ways in which these properties have influenced developments in spatial data analysis. The discussion focuses on exploratory and model-driven (confirmatory) modes of analyzing spatial data, in particular spatial point patterns and area data, which occupy a prominent position in social science research. The article concludes with a brief discussion of some key areas in which the field of spatial data analysis has been growing in recent years.
In this paper we view learning as an unconstrained non-linear minimization problem in which the objective function is defined by the negative log-likelihood function and the search space by the parameter space of an origin constrained product unit neural spatial interaction model. We consider Alopex based global search, as opposed to local search based upon backpropagation of gradient descents, each in combination with the bootstrapping pairs approach to solve the maximum likelihood learning problem. Interregional telecommunication traffic flow data from Austria are used as test bed for comparing the performance of the two learning procedures. The study illustrates the superiority of Alopex based global search, measured in terms of Kullback and Leibler's information criterion.
Aus dem Vereinsleben: Berichtszeitraum März 2020 bis Februar 2021
This paper focuses on Austrian outbound foreign direct investment (FDI, measured by sales of Austrian affiliates abroad) in Europe over the period 2009-2013, using a spatial Durbin panel data model specification with fixed effects, and a spatial weight matrix based on the first-order contiguity relationship of the countries and normalised by its largest eigenvalue. Third-country effects essentially enter the empirical analysis in two major ways: first, by the endogenous spatial lag on FDI (measured by FDI into markets nearby the host country), and, second, by including an exogenous market potential variable that measures the size of markets nearby the FDI host country in terms of gross domestic product. The question whether the empirical result is compatible with horizontal, vertical, export-platform or complex vertical FDI then depends on the sign and significance levels of both the coefficient of the spatial lag on FDI and the direct impact estimate of the market potential variable. The paper yields robust results that provide significant empirical evidence for horizontal FDI as the main driver of Austrian outbound FDI in Europe. This result is strengthened by the indirect impact estimate of the mark et potential variable indicating that spatial spillovers do not matter. (authors' abstract)
No abstract is provided for this article.
Trans-Alpine freight transport suffers because of severe road congestion due to under-capacity of roads and causes considerable environmental damages in the transit countries, particularly in Austria and Switzerland, but also in Italy. The obvious strategy of shifting trans-Alpine flows from road to rail is hampered by several factors such as Jack of coordination among the various national railway companies, lack of provision of competitive services (e.g., blocktrains, etc.), and insufficient efforts to increase and promote combined transport. One of the key problems is the disintegrated policy approach which does not conceive road and rail as complementary modes of transport, but still as competitive ones. Using the five dimensions - hardware, software, orgware (institutions), tinware, and the environment - as a checklist the current paper makes a modest attempt to identify areas of major concern for road and rail. The paper looks more specifically at the case of combined road/rail freight transport as part of a single, virtual mobility network for trans-Alpine freight transport. (authors' abstract)
