A connection between score matching and denoising autoencoders

Denoising autoencoders have been previously shown to be competitive alternatives to restricted Boltzmann machines for unsupervised pretraining of each layer of a deep architecture. We show that a simple denoising autoencoder training criterion is equivalent to matching the score (with respect to the data) of a specific energy-based model to that of a nonparametric Parzen density estimator of the data. This yields several useful insights. It defines a proper probabilistic model for the denoising autoencoder technique, which makes it in principle possible to sample from them or rank examples by their energy. It suggests a different way to apply score matching that is related to learning to denoise and does not require computing second derivatives. It justifies the use of tied weights between the encoder and decoder and suggests ways to extend the success of denoising autoencoders to a larger family of energy-based models.     

An automated microfluidic colourimetric sensor applied in situ to determine nitrite concentration

A stand-alone sensor system with integrated sub-systems is demonstrated. The system is portable and capable of in situ reagent-based nutrient analysis. The system is based on a low cost optical detection method, together with an automated microfluidic delivery system that is able to detect nitrite with a limit of detection (LOD) of 15 nM. The sensor was operated in situ at Southampton Dockhead for 57 h (December 2010) and 375 measurements were taken.     

Lip contour segmentation and tracking compliant with lip-reading application constraints

We propose to use both active contours and parametric models for lip contour extraction and tracking. In the first image, jumping snakes are used to detect outer and inner contour key points. These points initialize a lip parametric model composed of several cubic curves that are appropriate to the mouth deformations. According to a combined luminance and chrominance gradient, the initial model is optimized and precisely locked onto the lip contours. On subsequent images, the segmentation is based on the mouth bounding box and key point tracking. Quantitative and qualitative evaluations show the effectiveness of the algorithm for lip-reading applications.     

The ManyEars open framework

ManyEars is an open framework for microphone array-based audio processing. It consists of a sound source localization, tracking and separation system that can provide an enhanced speaker signal for improved speech and sound recognition in real-world settings. ManyEars software framework is composed of a portable and modular C library, along with a graphical user interface for tuning the parameters and for real-time monitoring. This paper presents the integration of the ManyEars Library with Willow Garage’s Robot Operating System. To facilitate the use of ManyEars on various robotic platforms, the paper also introduces the customized microphone board and sound card distributed as an open hardware solution for implementation of robotic audition systems.     

Mapping the 3D-surface strain field of patterned tensile stainless steels using atomic force microscopy

The quantification of microstructural strains at the surface of materials is of major importance for understanding the reactivity of solids. The present paper aims at demonstrating the potentialities of the atomic force microscopy (AFM) for mapping the three-dimensional surface strain field on patterned tensile specimens. Electron beam (e-beam) lithography has been used to deposit 16 x 16 arrays of gold-squared pads. Monitoring the evolution of such a pattern under applied strain allows to quantify the triaxial strains both at the micro-(plastic) domain and nanoscale (elastic) domain vs. applied strain. The proposed method was applied to stainless steels after 4.5% plastic strain.     

Precision instrumentation for rolling element bearing characterization

This article describes an instrument to measure the error motion of rolling element bearings. This challenge is met by simultaneously satisfying four requirements. First, an axial preload must be applied to seat the rolling elements in the bearing races. Second, one of the races must spin under the influence of an applied torque. Third, rotation of the remaining race must be prevented in a way that leaves the radial, axial/face, and tilt displacements free to move. Finally, the bearing must be fixtured and measured without introducing off-axis loading or other distorting influences. In the design presented here, an air bearing reference spindle with error motion of less than 10 nm rotates the inner race of the bearing under test. Noninfluencing couplings are used to prevent rotation of the bearing outer race and apply an axial preload without distorting the bearing or influencing the measurement. Capacitive displacement sensors with 2 nm resolution target the nonrotating outer race. The error motion measurement repeatability is shown to be less than 25 nm. The article closes with a discussion of how the instrument may be used to gather data with sufficient resolution to accurately estimate the contact angle of deep groove ball bearings.     

Dynamic simulation as a tool to integrate process design and process control

Dynamic simulation provides the process design and process control engineer with a highly effective tool to analyze process transients. When applied in the conceptual phase of a new process design, dynamic simualtion can be used to determine the optimum design of a proces by simultaneously considering steady-state economics and plant-wide controllability. A methodology is shown how rigorous dynamic process simulation can be used as a tool to effectively integrate the process design and process control disciplines. Illustration on extractive distillation systems demonstrate that this analysis leads to an optimum economic process design with a robust simple single-input single-output regulatory control scheme that meets process operability requirements.     

用于旋转对称三角传感器的一种鲁棒的处理方法

旋转对称三角传感器由于其光学系统的旋转对称性而比其他的传统激光三角传感器具有更好的柔性，将旋转对称三角传感器应用于工业领域，相应的处理方法的性能是至关重要的。为了使传感器在复杂物体表面情况下仍具有较好的性能，我们设计了一种用于旋转对称三角传感器的鲁棒的处理方法。首先，分析旋转对称三角传感器图像的特性，这些特性取决于被测物体的表面特性，而这些图像特性将有助于传感器图像的处理方法的设计。其次，用一种由粗到精检测方法实现对传感器图像的处理。提出在精检测中，一种结合了自适应平滑和峰值法的方法，以实现检测的鲁棒性。在实际传感器图像上的实验结果证明，这种处理方法对于来自复杂表面的传感器图像具有很好的鲁棒性，从而提高了传感器的柔性。     

Sliding wear transition for the CW614 brass alloy

Dry sliding wear tests were performed on a CW614 brass alloy using a pin-on-ring configuration. Wear kinetics were measured within a load range of 20–80 N and sliding velocity ranging from 1 to 7 m/s. Chemical compositions, morphologies and microstructures of worn surfaces and wear debris were characterised by scanning electron microscope (SEM) and energy dispersive X-ray spectrometer (EDS). Two main wear regimes have been observed: severe wear and mild wear. The results of wear tests and metallographic investigations on worn surfaces have been summarised in a wear mechanism map. It was found that the wear transition is controlled by a critical temperature at the contact surface.     

重振城市景观的的“乌托邦” 阿尔斯特图书馆和表演艺术学院

<正>阿尔斯特图书馆和表演艺术学院坐落在比利时阿尔斯特,由一栋8000m~2的砖结构建筑和一栋引人注目的 19世纪后期的历史建筑组成。项目的名称为"乌托邦",与托马斯莫尔的著作《乌托邦》同名,这本书是由阿尔斯特市民德克马滕斯首先出版。