Relaxed Lasso

The Lasso is an attractive regularisation method for high-dimensional regression. It combines variable selection with an efficient computational procedure. However, the rate of convergence of the Lasso is slow for some sparse high-dimensional data, where the number of predictor variables is growing fast with the number of observations. Moreover, many noise variables are selected if the estimator is chosen by cross-validation. It is shown that the contradicting demands of an efficient computational procedure and fast convergence rates of the ?₂-loss can be overcome by a two-stage procedure, termed the relaxed Lasso. For orthogonal designs, the relaxed Lasso provides a continuum of solutions that include both soft- and hard-thresholding of estimators. The relaxed Lasso solutions include all regular Lasso solutions and computation of all relaxed Lasso solutions is often identically expensive as computing all regular Lasso solutions. Theoretical and numerical results demonstrate that the relaxed Lasso produces sparser models with equal or lower prediction loss than the regular Lasso estimator for high-dimensional data.     

Inhibition-augmented trainable COSFIRE filters for keypoint detection and object recognition

The shape and meaning of an object can radically change with the addition of one or more contour parts. For instance, a T-junction can become a crossover. We extend the COSFIRE trainable filter approach which uses a positive prototype pattern for configuration by adding a set of negative prototype patterns. The configured filter responds to patterns that are similar to the positive prototype but not to any of the negative prototypes. The configuration of such a filter comprises selecting given channels of a bank of Gabor filters that provide excitatory or inhibitory input and determining certain blur and shift parameters. We compute the response of such a filter as the excitatory input minus a fraction of the maximum of inhibitory inputs. We use three applications to demonstrate the effectiveness of inhibition: the exclusive detection of vascular bifurcations (i.e., without crossovers) in retinal fundus images (DRIVE data set), the recognition of architectural and electrical symbols (GREC’11 data set) and the recognition of handwritten digits (MNIST data set).     

德国机械设备制造业联合会纺织机械协会

<正>ITMA ASIA+CITME 2010开展在即,德国纺机制造业终于看到了曙光。与2009年相比,其2010年第一季度的新订单同比增加了130%。VDMA纺织机械协会预测,2010年     

Rheology of thermo-reversible fish protein isolate gels

Fish protein isolates (FPI) from cod muscle were studied at pH 9 and 11. Thermo-reversible gels, stable at and below room temperature could be produced at both pH values. Application of low shear stress to the gels led to an initial elastic response followed by a power-law deformation (creep) at both pH. Above a critical shear stress gels fractured and flowed regularly with a viscosity of about 0.05 Pa s.Strain–stress curves were recorded using both flow and oscillatory measurements. For all systems strain-hardening was observed followed by fracture. Fracture occurred at approximately the same deformation, but the amplitude of strain-hardening and the yield stress decreased with decreasing rate of the stress ramp. Results obtained from oscillatory shear at 1 or 0.1 Hz were close to those obtained with continuous shear during the fastest stress ramps.The structure of the gels was investigated using confocal laser scanning microscopy and turbidity measurements.     

Application of dimensionality reduction to visualisation of high-throughput data and building of a classification model in formulated consumer product design

Several dimensionality reduction techniques were applied to two data sets of consumer products formulations in order to infer their intrinsic structure and specific product design rules. High throughput experiments were used to generate the data sets of sufficient size. Supervised isometric feature mapping (S-Isomap) was combined with a k-nearest neighbours (k-NN) classifier and k-means clustering algorithm to perform categorization of viscosity of new formulations, not used to train the model. We compared prediction results of this approach with several well-established classification models. The results show the accuracy of the S-Isomap based approach to be superior and with a potential for further improvement. Compared with other dimensionality reduction techniques, applying S-Isomap has allowed for a superior visualization of category separation within the formulations, for the data sets used.     

Charge Transport and Recombination in Polyspirobifluorene Blue Light‐Emitting Diodes

The charge transport in blue light‐emitting polyspirobifluorene is investigated by both steady‐state current‐voltage measurements and transient electroluminescence. Both measurement techniques yield consistent results and show that the hole transport is space‐charge limited. The electron current is found to be governed by a high intrinsic mobility in combination with electron traps. Numerical simulations on light‐emitting diodes reveal a shift in the recombination zone from the cathode to the anode with increasing bias.     

Adaptive Matrices and Filters for Color Texture Classification

In this paper we introduce an integrative approach towards color texture classification and recognition using a supervised learning framework. Our approach is based on Generalized Learning Vector Quantization (GLVQ), extended by an adaptive distance measure, which is defined in the Fourier domain, and adaptive filter kernels based on Gabor filters. We evaluate the proposed technique on two sets of color texture images and compare results with those other methods achieve. The features and filter kernels learned by GLVQ improve classification accuracy and they are able to generalize much better for data previously unknown to the system.     

Device Physics of White Polymer Light‐Emitting Diodes

The charge transport and recombination in white‐emitting polymer light‐ emitting diodes (PLEDs) are studied. The PLED investigated has a single emissive layer consisting of a copolymer in which a green and red dye are incorporated in a blue backbone. From single‐carrier devices the effect of the green‐ and red‐emitting dyes on the hole and electron transport is determined. The red dye acts as a deep electron trap thereby strongly reducing the electron transport. By incorporating trap‐assisted recombination for the red emission and bimolecular Langevin recombination for the blue emission, the current and light output of the white PLED can be consistently described. The color shift of single‐layer white‐emitting PLEDs can be explained by the different voltage dependencies of trap‐assisted and bimolecular recombination.