Using toponymy to reconstruct past land use: a case study of ‘brüsáda’ (burn) in southern Switzerland

Toponyms are frequently the outcome of a creative process, a subjective interpretation by the local inhabitants at the time of naming. They often survive changes in the local landscape, thus becoming historical documents of landscape dynamics or changes in land use. In this contribution we provide a systematic synchronic analysis of the toponym brüsáda (burn) in Canton Ticino. In total, 182 instances of place names containing brüsáda were recorded, of which 102 cases (56.0%) were localised. Based on oral information provided by the local people or on archival documents, it was possible to group the selected toponyms into different meaning classes such as burnt building, exposed sites, pasture-land and others without any indication (classified here as generic). On the basis of etymological, topographic, documentary, bibliographic, and historic evidence, as well as proxy-archives (such as charcoal in lake sediments), we hypothesise that most of the toponyms associated with pasture-land – and probably also a large portion of the generic toponyms – are directly or indirectly related to the habit of using fire to clear brushwood to improve pasture-land or to eliminate trees. In the study area, the practice of pasture-fires appears to have been a very well controlled and targeted activity in medieval times. In contrast, pasture-fires became very frequent and uncontrolled in the second half of the nineteenth century up until the Second World War. Our study confirms the usefulness and suitability of toponymy, in combination with other disciplines, for ecohistorical reconstructions of past environmental and land-use changes.     

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

This paper attempts to develop a mathematically rigid framework for minimizing the cross-entropy function in an error backpropagating framework. In doing so, we derive the backpropagation formulae for evaluating the partial derivatives in a computationally efficient way. 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 classification 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 classification accuracy 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 unacceptable instabilities in the classification results.     

Imbonga and Batalimo: ceramic evidence for early settlement of the equatorial rain forest

This paper summarizes four seasons of archaeological field work in the equatorial rain forest of Zaïre. It discusses the current state of research on the earliest settlement of the forest by pottery-producing populations. Pottery from the oldest ceramic horizon, named after its type-site Imbonga on the Momboyo, is compared with that of the so-called Batalimo-Maluba horizon, named after the sites of Batalimo on the Lobaye and Maluba on the Lua.

---

Résumé Cet article résume les résultats de quatre saisons de recherches archéologiques dans la forêt humide équatoriale du Zaïre. On y discute de l'état actuel de la recherche sur l'occupation la plus ancienne de la forêt par des populations fabriquant de la céramique. On compare la céramique de l'horizon le plus ancien, qui porte le nom du site-type d'Imbonga sur la Momboyo, avec celle de l'horizon Batalimo-Maluba, qui porte les noms des gisements de Batalimo sur la Lobaye et de Maluba sur la Lua.

     

Telecommunication Media Choice Behavior in Academia: An Austrian-Swiss Comparison

A general framework to analyze communication media choice behavior in the university setting is proposed which integrates a stated preference experimental design procedure into a discrete choice modeling framework. The framework is empirically tested using hypothetical choice experiments in which traditional mail, courier mail, telephone, facsimile, and electronic mail services were choice options to carry out information communication tasks. For this purpose face-to-face interviews were conducted in six universities in Austria and Switzerland. The choice modeling approach developed emphasizes the influence of communication context specific characteristics, individual and organizational characteristics of the communication initiator as well as the individual's perceptions and feelings about the communication media on the formation of preferences. Empirical results are presented using stated preference models of communication media choice behavior for a series of communication situations. Specific emphasis is laid on cross-national differences in choice behavior.     

Learning in Single Hidden-Layer Feedforward Network Models: Backpropagation in a Spatial Interaction Modeling Context

Learning in neural networks has attracted considerable interest in recent years. Our focus is on learning in single hidden-layer feedforward networks which is posed as a search in the network parameter space for a network that minimizes an additive error function of statistically independent examples. We review first the class of single hidden-layer feedforward networks and characterize the learning process in such networks from a statistical point of view. Then we describe the backpropagation procedure, the leading case of gradient descent learning algorithms for the class of networks considered here, as well as an efficient heuristic modification. Finally, we analyze the applicability of these learning methods to the problem of predicting interregional telecommunication flows. Particular emphasis is laid on the engineering judgment, first, in choosing appropriate values for the tunable parameters, second, on the decision whether to train the network by epoch or by pattern (random approximation), and, third, on the overfitting problem. In addition, the analysis shows that the neural network model whether using either epoch-based or pattern-based stochastic approximation outperforms the classical regression approach to modeling telecommunication flows.