Robust output feedback control of nonlinear stochastic systems using neural networks

We present an adaptive output feedback controller for a class of uncertain stochastic nonlinear systems. The plant dynamics is represented as a nominal linear system plus nonlinearities. In turn, these nonlinearities are decomposed into a part, obtained as the best approximation given by neural networks, plus a remaining part which is treated as uncertainties, modeling approximation errors, and neglected dynamics. The weights of the neural network are tuned adaptively by a Lyapunov design. The proposed controller is obtained through robust optimal design and combines together parameter projection, control saturation, and high-gain observers. High performances are obtained in terms of large errors tolerance as shown through simulations.     

Least Squares and Robust Estimation of Local Image Structure

Linear scale space methodology uses Gaussian probes at scale s to observe the differential structure. In observing the differential image structure through the Gaussian derivative probes at scale s we implicitly construct the Taylor series expansion of the smoothed image. The Gaussian facet model, as a generalization of the classic Haralick facet model, constructs a polynomial approximation of the unsmoothed image. The measured differential structure therefore is closer to the ‘real’ structure than the differential structure measured using Gaussian derivatives.At the points in an image where the differential structure changes abruptly (because of discontinuities in the imaging conditions, e.g. a material change, or a depth discontinuity) both the Gaussian derivatives and the Gaussian facet model diffuse the information from both sides of the discontinuity (smoothing across the edge).Robust estimators that are classically meant to deal with statistical outliers can also be used to deal with these ‘mixed model distributions’. In this paper we introduce the robust estimators of local image structure. Starting with the Gaussian facet model where we replace the quadratic error norm with a robust (Gaussian) error norm leads to a robust Gaussian facet model.We will show examples of using the robust differential structure estimators for luminance and color images, for zero and higher order differential structure. Furthermore we look at a ‘robustified’ structure tensor that forms the basis of robust orientation estimation.First online version published in June, 2005     

On Nirenberg-type conditions for higher-order systems on time scales

We study the existence of periodic solutions for a nonlinear system of nth-order differential equations on time scales. Assuming a suitable Nirenberg-type condition, we prove the existence of at least one solution of the problem using Mawhin’s coincidence degree.     

Optimized H.264 Video Encoding and Packetization for Video Transmission Over Pipeline Forwarding Networks

Previous works showed that the quality-of-service (QoS) requirements of multimedia applications can be optimally satisfied by pipeline forwarding (PF) by providing end-to-end delay guarantees as well as high network resource utilization. However, the unavoidable mismatch between reserved resources and the unpredictable traffic profile of a video stream has an impact on the resulting application layer quality. Therefore, a new low-complexity H.264 video encoding and packetization scheme based on a distortion-optimized macroblock grouping technique is designed here to maximize the performance of video transmission on PF networks. The scheme considers the perceptual importance of the different parts of the video data to group the most important information in few packets that are the natural candidates to receive the deterministic service provided by PF. Results show peak signal-to-noise ratio (PSNR) gains up to 2.5 dB over traditional video encoding and packetization schemes, as well as more graceful degradation in case of high network load.     

Avatar mobility in user-created networked virtual worlds: measurements, analysis, and implications

We collected mobility traces of avatars spanning multiple regions in Second Life, a popular user-created virtual world. We analyzed the traces to characterize the dynamics of the avatars’ mobility and behavior, both temporally and spatially. We discuss the implications of our findings on the design of peer-to-peer architecture, interest management, mobility modeling of avatars, server load balancing and zone partitioning, caching, and prefetching for user-created virtual worlds.     

CSDL: a language for cooperative systems design

The aim of a cooperative system is to coordinate and support group activities. Cooperative Systems Design Language (CSDL) is an experimental language designed to support the development of cooperative systems from specification to implementation. In CSDL, a system is defined as a collection of reusable entities implementing floor control disciplines and shared workspaces. CSDL tries to address the difficulties of integrating different aspects of cooperative systems: cooperation control, communication, and system modularization. This paper presents CSDL as a specification language. Basic units are coordinators that can be combined hierarchically. A coordinator is composed of a specification, a body, and a context. The specification defines the cooperation policy; the body controls the underlying communication channels; and the context defines coordinators' interaction in modular systems     

Adaptation in Multimedia Systems

Multimedia systems can profit a lot from personalization. Such a personalization is essential to give users the feeling that the system is easily accessible especially if it is done automatically. The way this adaptive personalization works is very dependent on the adaptation model that is chosen.We introduce a generic two-dimensional classification framework for user modeling systems. This enables us to clarify existing as well as new applications in the area of user modeling. In order to illustrate our framework we evaluate push and pull based user modeling in user modeling systems.Paul de Vrieze received his Masters degree in Information Science in 2002 from the University Of Tilburg, The Netherlands. He is currently junior researcher at the University of Nijmegen. His main research interests include adaptive systems and user modelling.Patrick van Bommel received his Masters degree in Computer Science in 1990, and the degree of Ph.D in Mathematics and Computer Science, from the University of Nijmegen, the Netherlands in 1995. He is currently assistant professor at the University of Nijmegen. His main research interests include information modelling and information retrieval.Prof.Dr.Ir. Th.P. van der Weide received his masters degree from the Technical University Eindhoven, the Netherlands in 1975, and the degree of Ph.D in Mathematics and Physics from the University of Leiden, the Netherlands in 1980. He is currently professor at the University of Nijmegen, the Netherlands. His main research interests include information systems, information retrieval, hypertext and knowledge based systems.     

ENTEROBACTER AGGLOMERANS—A NEW BACTERIAL CONTAMINANT ISOLATED FROM LAGER BEER BREWERIES

From a large number of bacterial strains isolated from all stages of the brewing process in four different beer breweries in South Africa, fifty-five bacterial strains were identified as Enterobacteriaceae by phenotypic analysis. All enterobacterial species previously reported in brewery samples, as well as eighteen strains of Enterobacter agglomerans were found. E. agglomerans strains were isolated from pitching yeast and fermenting wort samples. The occurrence of E. agglomerans in pitching yeast is significant and indicates that these bacteria survive the brewing process. E. agglomerans is regarded as a potential beer spoilage contaminant in lager beer breweries.     

Evolutionary Computation in Structural Design

The paper provides the results of preliminary research on the application of evolutionary computation to integrated structural design in which a complex design support tool automatically conducts both conceptual and detailed design. In the paper, a brief overview of the state of the art in evolutionary computation and its applications to structural design is provided. Next, Inventor 2000 is described, a unique research and structural design tool developed by the authors at George Mason University that combines an evolutionary computation component with a system for wind forces analysis, and a system for the analysis, design and optimisation of steel structures. The paper also presents the results of four structural design experiments conducted with Inventor 2000. The objective of experiments was to investigate various forms of evolutionary computation as applied to structure design. Finally, the paper provides the initial research conclusions and recommendations for further research.