Modelling water dynamics with DNDC and DAISY in a soil of the North China Plain: A comparative study

The performance of the DNDC and Daisy model to simulate the water dynamics in a floodplain soil of the North China Plain was tested and compared. While the DNDC model uses a simple cascade approach, the Daisy model applies the physically based Richard's equation for simulating water movement in soil. For model testing a three years record of the soil water content from the Dong Bei Wang experimental station near Beijing was used. There, the effect of nitrogen fertilization, irrigation and straw removal on soil water and nitrogen dynamics was investigated in a three factorial field experiment applying a split-split-plot design with 4 replications. The dataset of one treatment was used for model testing and calibration. Two other independent datasets from further treatments were employed for validating the models. For both models, the simulation results were not satisfying using default parameters. After parameter optimisation and the use of site-specific van Genuchten parameters, however, the Daisy model performed well. But, for the DNDC model, parameter optimisation failed to improve the simulation result. Owing to the fact that many biological processes such as plant growth, nitrification or denitrification depend strongly on the soil water content, our findings bring us to the conclusion that the site-specific suitability of the DNDC model for simulating the soil water dynamics should be tested before further simulation of other processes.     

Statistical relational learning of trust

The learning of trust and distrust is a crucial aspect of social interaction among autonomous, mentally-opaque agents. In this work, we address the learning of trust based on past observations and context information. We argue that from the truster’s point of view trust is best expressed as one of several relations that exist between the agent to be trusted (trustee) and the state of the environment. Besides attributes expressing trustworthiness, additional relations might describe commitments made by the trustee with regard to the current situation, like: a seller offers a certain price for a specific product. We show how to implement and learn context-sensitive trust using statistical relational learning in form of a Dirichlet process mixture model called Infinite Hidden Relational Trust Model (IHRTM). The practicability and effectiveness of our approach is evaluated empirically on user-ratings gathered from eBay. Our results suggest that (i) the inherent clustering achieved in the algorithm allows the truster to characterize the structure of a trust-situation and provides meaningful trust assessments; (ii) utilizing the collaborative filtering effect associated with relational data does improve trust assessment performance; (iii) by learning faster and transferring knowledge more effectively we improve cold start performance and can cope better with dynamic behavior in open multiagent systems. The later is demonstrated with interactions recorded from a strategic two-player negotiation scenario.     

Scaling Kernel-Based Systems to Large Data Sets

In the form of the support vector machine and Gaussian processes, kernel-based systems are currently very popular approaches to supervised learning. Unfortunately, the computational load for training kernel-based systems increases drastically with the size of the training data set, such that these systems are not ideal candidates for applications with large data sets. Nevertheless, research in this direction is very active. In this paper, I review some of the current approaches toward scaling kernel-based systems to large data sets.     

Transforming smoothers for PDE constrained optimization problems

Transforming smoothers are known as a successful approach to the multigrid treatment of saddlepoint problems resulting from variational problems. In this paper we analyze similar multigrid methods in the context of the practically important class of optimization problems with partial differential equation constraint, which lead to a different kind of saddle point problems. We prove level independent convergence properties for the resulting multigrid methods and demonstrate this also in numerical investigations for a relevant model problem.     

Kernel fisher discriminants for outlier detection

The problem of detecting atypical objects or outliers is one of the classical topics in (robust) statistics. Recently, it has been proposed to address this problem by means of one-class SVM classifiers. The method presented in this letter bridges the gap between kernelized one-class classification and gaussian density estimation in the induced feature space. Having established the exact relation between the two concepts, it is now possible to identify atypical objects by quantifying their deviations from the gaussian model. This model-based formalization of outliers overcomes the main conceptual shortcoming of most one-class approaches, which, in a strict sense, are unable to detect outliers, since the expected fraction of outliers has to be specified in advance. In order to overcome the inherent model selection problem of unsupervised kernel methods, a cross-validated likelihood criterion for selecting all free model parameters is applied. Experiments for detecting atypical objects in image databases effectively demonstrate the applicability of the proposed method in real-world scenarios.     

Generation of time delays: simplified models of intracellular signalling in cerebellar Purkinje cells

In many neuronal systems, information is encoded in temporal spike patterns. The recognition and storage of temporal patterns requires the generation and modulation of time delays between inputs and outputs. In cerebellar Purkinje cells, stimulation of metabotropic glutamate receptors (mGluRs) results in a delayed calcium and voltage response that has been implicated in classical conditioning and temporal pattern recognition. Here, we analyse and simplify a complex model of the intracellular signalling network that has been proposed as a substrate for this delayed response. We systematically simplify the original model, present a minimal model of time delay generation, and show that a delayed response can be produced by the combination of negative feedback and autocatalysis, without any intervening signalling steps that would contribute additive delays. The minimal model is analysed using phase plane methods, and classified as an excitable system. We discuss the implication of excitability for computations performed by intracellular signalling networks in general.     

Service level management with agent technology

This paper outlines a generic evaluation methodology for multimedia and real time applications. It concentrates on the application layer and the service aspect. Principles of software evaluation for quality assessment and measurement for networked services and distributed applications are used to present a specific method for measuring important characteristics. The method and the measurement procedure allow comparison of the actual characteristics of service quality with the required characteristics and thresholds. The proof of concepts will be made by the implementation of quality measurement agents following the approach outlined in this paper. Our approach of performing service level management (SLM) with agent technology is distributed, slim, minimizable to the maximum, independent in its methodology and offers comparable, objective results.     

On the number of term orders

By an admissible order on a finite subsetQ of ⁿ we mean the restriction toQ of a linear order on ⁿ compatible with the group structure of ⁿ and such that ⁿ is contained in the positive cone of the order. We first derive upper and lower bounds on the number of admissible orders on a given setQ in terms of the dimensionn and the cardinality ofQ. Better estimates are possible if the setQ enjoys symmetry properties and in the case whereQ is a discrete hyperbox of the form In the latter case, we also give asymptotic results as d _k for fixedn. We finally present algorithms which compute the set of admissible orders that extend a given binary relation onQ and their number. The algorithms are useful in connection with the construction of universal Gröbner bases.AMS Classification: primary 06F20 secondary 06-04, 11N25