Search

Rear-end collision prediction has gained an increasing attention for safety improvement in smart cities. It is urgent to design efficient warning strategies for rear-end collisions which is one of the main causes of traffic accidents. The existing researches have been conducted to predict the collisions. Learning-based methods are proposed to solve this complicated issue, which the traditional methods are difficult to solve. However, due to some limitations in terms of the feature extraction and prediction performance, back-propagation learning methods are facing challenge. In this paper, we proposed a novel Rear-end Collision Prediction Mechanism with deep learning method (RCPM), in which a convolutional neural network model is established. In RCPM, the dataset is smoothed and expanded based on genetic theory to alleviate the class imbalance problem. The preprocessed dataset is divided into training and testing sets as the input to train our convolutional neural network model. The experimental results show that compared with the Honda, Berkeley and multi-layer perception neural-network-based algorithms, RCPM effectively improves performance to predict rear-end collisions. [ABSTRACT FROM AUTHOR]

This paper proposes a multiclassification algorithm using multilayer perceptron neural network models. It tries to boost two conflicting main objectives of multiclassifiers: a high correct classification rate level and a high classification rate for each class. This last objective is not usually optimized in classification, but is considered here given the need to obtain high precision in each class in real problems. To solve this machine learning problem, we use a Pareto-based multiobjective optimization methodology based on a memetic evolutionary algorithm. We consider a memetic Pareto evolutionary approach based on the NSGA2 evolutionary algorithm (MPENSGA2). Once the Pareto front is built, two strategies or automatic individual selection are used: the best model in accuracy and the best model in sensitivity (extremes in the Pareto front). These methodologies are applied to solve 17 classification benchmark problems obtained from the University of California at Irvine (UCI) repository and one complex real classification problem. The models obtained show high accuracy and a high classification rate for each class. [ABSTRACT FROM AUTHOR]

The Heavy Ion Fusion (HIF) program is progressing through a series of physics and technology demonstrations leading to an inertial fusion power plant. The High Current Experiment (HCX) at Lawrence Berkeley National Laboratory is exploring the physics of intense beams with high line-charge density. Superconducting focusing quadrupoles were developed for magnetic transport studies at the HCX. A baseline design was selected following several pre-series models. Optimization of the baseline design led to the development of a first prototype that achieved a conductor-limited gradient of 132 T/m in a 70 mm bore, without training, with measured field errors at the 0.1% level. Based on these results, the magnet geometry and fabrication procedures were adjusted to improve the field quality. These modifications were implemented in a second prototype. In this paper, the optimized design is presented and comparisons between the design harmonics and magnetic measurements performed on the new prototype are discussed. [ABSTRACT FROM AUTHOR]

A calendar of events is presented for Institute of Electrical & Electronics Engineers(IEEE) Computer Society events around the world in 2007. From April 16-20, 2007, the 23rd IEEE International Conference on Data Engineering will take place in Istanbul, Turkey. From May 6-10, the IEEE VLSI Test Symposium will take place in Berkeley, California. On May 7 until the 9, the 10th IEEE International Symposium on Object, Component, & Service-Oriented Real-Time Distributed Computing will take place in Santorini, Greece. From May 7-11, the 8th International Conference on Mobile Data Management will take place in Mannheim, Germany.

A new deep learning (DL)-based parameter extraction method is presented in this brief; 50k training cases are generated by Monte Carlo simulations of these preselected parameters in Berkeley short-channel IGFET model (BSIM)-common multigate (CMG). DL models are trained using backward propagation with ${C} _{\text {gg}} - {V} _{g}$ and ${I} _{d} - {V} _{g}$ as the input and selected BSIM-CMG parameters as the output. A TCAD simulated FinFET device, calibrated to Intel 10-nm node, is used to test the DL models. The DL-based parameters extraction results show an excellent fit to capacitance and drain current data, with 0.16% rms error in ${C} _{\text {gg}} - {V} _{g}$ and 6.1% rms error in ${I} _{d} - {V} _{g}$ (0.69% rms error in above-threshold-voltage ${I} _{d} - {V} _{g}$), respectively. In addition, devices with a 10% variation in gate length and oxide thickness are successfully modeled with the trained DL model. The results show tremendous promise in using the DL-based models for parameter extraction. [ABSTRACT FROM AUTHOR]

A new instrument for the characterization of superconducting materials as a function of Magnetic Field, Temperature and Strain, was designed, constructed and tested at Lawrence Berkeley National Laboratory (LBNL). A U-shaped bending spring was selected, since that design has proven to enable accurate characterizations of a multitude of superconducting materials for more than a decade. The new device is validated though measurements on very high current Rod Restack Processed (RRP) Internal-Tin (IT) wires, for which we will present initial results, including parameterizations of the superconducting phase boundaries and comparisons with other wire types. Accurate parametrization of modern high magnetic field conductors is important for the analysis of the performance of magnet systems. [ABSTRACT FROM AUTHOR]

Details on the XVIIth International Symposium on Discharges and Electrical Insulation in Vacuum (ISDEIV) hosted by Lawrence Berkeley National Laboratory (LBNL) in Berkeley, California on July 20-25, 1996. Participants of the Symposium; Highlights of the Symposium; Two awards granted at the Symposium; Sponsors of the Symposium.

Extreme ultraviolet lithography is a leading candidate for volume production of nanoelectronics at the 32-nm node and beyond. In order to ensure adequate maturity of the technology by the start date for the 32-nm node, advanced development tools are required today with numerical apertures of 0.25 or larger. In order to meet these development needs, a microexposure tool based on SEMATECH's 0.3-numerical aperture microfield optic has been developed and implemented at Lawrence Berkeley National Laboratory, Berkeley, CA. Here we describe the Berkeley exposure tool in detail, discuss its characterization, and summarize printing results obtained over the past year. Limited by the availability of ultrahigh resolution chemically amplified resists, present resolving capabilities limits are approximately 32 nm for equal lines and spaces and 28 nm for semi-isolated lines. [ABSTRACT FROM AUTHOR]

Describes upgrades on the Advanced Light Source (ALS) of the Lawrence Berkeley National Laboratories in California with three superconducting dipoles. Incorporation of conduction-cooled NbTi windings and high temperature superconducting current leads, epoxy-glass suspension straps and a Gifford-McMahon cryocooler.

Describes the use of the scenario assignment tool in an upper division Mechanical Engineering course at the University of California-Berkeley. Challenge regarding finding effective qualitative tools to assess students' skills and knowledge of engineering practices; Learning outcomes associated with the scenario tool; Identification of students' strengths and weaknesses.

The article describes the research work of Erik B. Sudderth, a postdoctoral scholar at the University of California's Department of Electrical Engineering and Computer Science in Berkeley, California. Sudderth's research focuses on probabilistic graphical models, nonparametric Bayesian method as well as applications of statistical machine learning in several areas, such as image processing, tracking, object recognition and visual scene analysis. Sudderth was able to develop hierarchical generative models for objects, their parts as well as the scenes surrounding these objects through application of nonparametric Bayesian methods.