Direction-Dependent Learning Approach for Radial Basis Function Networks

Direction-dependent scaling, shaping, and rotation of Gaussian basis functions are introduced for maximal trend sensing with minimal parameter representations for input output approximation. It is shown that shaping and rotation of the radial basis functions helps in reducing the total number of function units required to approximate any given input-output data, while improving accuracy. Several alternate formulations that enforce minimal parameterization of the most general radial basis functions are presented. A novel "directed graph" based algorithm is introduced to facilitate intelligent direction based learning and adaptation of the parameters appearing in the radial basis function network. Further, a parameter estimation algorithm is incorporated to establish starting estimates for the model parameters using multiple windows of the input-output data. The efficacy of direction-dependent shaping and rotation in function approximation is evaluated by modifying the minimal resource allocating network and considering different test examples. The examples are drawn from recent literature to benchmark the new algorithm versus existing methods     

Endothelial selectins and vascular cell adhesion molecule-1 promote hematopoietic progenitor homing to bone marrow

The adhesive mechanisms allowing hematopoietic progenitor cells (HPC) homing to the bone marrow (BM) after BM transplantation are poorly understood. We investigated the role of endothelial selectins and vascular cell adhesion molecule-1 (VCAM-1) in this process. Lethally irradiated recipient mice deficient in both P-and E-selectins (P/E-/-), reconstituted with minimal numbers (相似文献   

Mechanical properties of biorenewable fiber/plastic composites

Plastic fiber composites, consisting of polypropylene (PP) or polyethylene (PE), and pinewood, big blue stem (BBS), soybean hulls, or distillers dried grain and solubles (DDGS), were prepared by extrusion. Young's modulus, tensile and flexural strengths, melt flow, shrinkage, and impact energy, with respect to the type, amount, and size of fiber on composites, were evaluated. Young's moduli under tensile load of wood, BBS, and soybean‐hull fiber composites, compared with those of pure plastic controls, were either comparable or higher. Tensile strength significantly decreased for all the PP/fiber composites when compared with that of the control. Strength of BBS fiber composites was higher than or comparable to that of wood. When natural fibers were added there was a significant decrease in the melt flow index for both plastic/fiber composites. There was no significant difference in the shrinkage of all fiber/plastic composites compared to that of controls. BBS/PE plastic composites resulted in higher notched impact strength than that of wood or soybean‐hull fiber composites. There was significant reduction in the unnotched impact strength compared to that of controls. BBS has the potential to be used as reinforcing materials for low‐cost composites. © 2004 Wiley Periodicals, Inc. J Appl Polym Sci 93: 2484–2493, 2004     

Preparation and surface- active properties of polyoxyethylene-glycol (600) monoesters of fatty acids

Polyoxyethyleneglycol (PEG-600) monoesters of undecylenic, lauric, myristic, palmitic, stearic, oleic, ricinoleic and 12-hydroxystearic acids were prepared in 80-85% yield by reacting PEG with boric acid, esterifying the resultant borate with fatty acid, and selectively hydrolyzing the borate ester; their surface active properties were evaluated. Increase in acyl chain length increased the surface tension. The presence of a double bond or hydroxyl group in C₁₈ series improved the wetting and emulsifying proper-ties. Unsaturation in the fatty acid chain reduced the foaming power.     

Sequential Monte Carlo in reachability heuristics for probabilistic planning

Some of the current best conformant probabilistic planners focus on finding a fixed length plan with maximal probability. While these approaches can find optimal solutions, they often do not scale for large problems or plan lengths. As has been shown in classical planning, heuristic search outperforms bounded length search (especially when an appropriate plan length is not given a priori). The problem with applying heuristic search in probabilistic planning is that effective heuristics are as yet lacking.In this work, we apply heuristic search to conformant probabilistic planning by adapting planning graph heuristics developed for non-deterministic planning. We evaluate a straight-forward application of these planning graph techniques, which amounts to exactly computing a distribution over many relaxed planning graphs (one planning graph for each joint outcome of uncertain actions at each time step). Computing this distribution is costly, so we apply Sequential Monte Carlo (SMC) to approximate it. One important issue that we explore in this work is how to automatically determine the number of samples required for effective heuristic computation. We empirically demonstrate on several domains how our efficient, but sometimes suboptimal, approach enables our planner to solve much larger problems than an existing optimal bounded length probabilistic planner and still find reasonable quality solutions.     

H-Infinity Static Output-feedback Control for Rotorcraft

The problem of stabilization of an autonomous rotorcraft platform in a hover configuration subject to external disturbances is addressed. Necessary and sufficient conditions are presented for static output-feedback control of linear time-invariant systems using the H-Infinity approach. Simplified conditions are given which only require the solution of two coupled matrix design equations. This paper also proposes a numerically efficient solution algorithm for the coupled design equations to determine the output-feedback gain. A major contribution is that an initial stabilizing gain is not needed. The efficacy of the control law and the disturbance accommodation properties are shown on a rotorcraft design example. The helicopter dynamics do not decouple as in the fixed-wing aircraft case, so that the design of helicopter flight controllers with a desirable intuitive structure is not straightforward. In this paper an output feedback approach is given that allows one to selectively close prescribed multivariable feedback loops using a reduced set of the states. Shaping filters are added that improve performance and yield guaranteed robustness and speed of response. This gives direct control over the design procedure and performance. Accurate identification of the System parameters is a challenging task for rotorcraft control, addition of loop shaping facilitates implementation engineers to counteract unmodeled high frequency dynamics. The net result yields control structures that have been historically accepted in the flight control community.     

SMARTINT: using mined attribute dependencies to integrate fragmented web databases

Many web databases can be seen as providing partial and overlapping information about entities in the world. To answer queries effectively, we need to integrate the information about the individual entities that are fragmented over multiple sources. At first blush this is just the inverse of traditional database normalization problem—rather than go from a universal relation to normalized tables, we want to reconstruct the universal relation given the tables (sources). The standard way of reconstructing the entities will involve joining the tables. Unfortunately, because of the autonomous and decentralized way in which the sources are populated, they often do not have Primary Key–Foreign Key relations. While tables may share attributes, naive joins over these shared attributes can result in reconstruction of many spurious entities thus seriously compromising precision. Our system, SmartInt is aimed at addressing the problem of data integration in such scenarios. Given a query, our system uses the Approximate Functional Dependencies (AFDs) to piece together a tree of relevant tables to answer it. The result tuples produced by our system are able to strike a favorable balance between precision and recall.     

Direct three dimensional tomography of flames using maximization of entropy technique

Presently, there are numerous applications for non-destructive techniques like emission tomography, laser based methods and particle image velocimetry that are used to study flame characteristics. Reconstruction of the flame intensity field using emission tomography has the advantage over other technologies that it gives accurate results but at the same time requires relatively inexpensive equipment, and therefore, has numerous industrial applications. In the present paper, a new algorithm performing Direct-3D reconstruction using the maximization of entropy (MENT) methodology has been introduced. Through detailed studies using a mathematical object, it has been shown that the Direct-3D algorithm shows significantly reduced errors as compared to 2D slice-by-slice reconstruction algorithms. Secondly, the major features of the proposed algorithm, for e.g., effect of orientation, effect of number of views, and robustness have been discussed. Finally, a few qualitative results from actual flames have been presented using a candle and a gas fired burner, and the results match well with the actual flame geometry and intensity distribution.