教学课程资源库 更多>>

关键词： Estimation   Social networking (online)   Blogs   Licenses   Computer science   Urban areas   Cultural differences   Cultural differences   home location estimation   online social networks   Twitter   user profile  

摘要： Social media is used all over the world. Although the home location attribute is available anywhere, few studies have been conducted on the differences in the performance of home location estimation between countries. Because social media is used differently in different countries, the difference between countries in terms of users' relationships needs to be investigated. We consider the performance of graph-based home location estimation in capturing users' relationships, as it is affected by users' social media usage. We analyze user profiles in 10 countries by graph-based estimation and find that some profile attributes, in particular, contribute to the estimation performance for all the countries examined.

关键词： Energy   Computer science   Design   Materials science  

摘要： Metal laser powder bed fusion (L-PBF) allows for production of complex components using the energy from a laser to locally melt micron-sized powder following a layer-wise approach. Considerable scientific efforts are focused on addressing the influence of the process parameters on the melting stability and the control of material properties, while developing necessary monitoring and characterization tools. The importance of the process atmosphere has largely been undermined in favour of first order parameters connected to the laser scanning. The role of the atmosphere has been limited to the reduction of the operating residual oxygen level down to typically 1000 ppm. This thesis focuses on providing knowledge on the influence of the process atmosphere on the laser metal powder interaction during L-PBF and the resulting properties of the built material in terms of generated defects, microstructure and mechanical properties. Different purities and compositions of generated atmospheres have been investigated to manufacture the most used materials in the field, namely 316L stainless steel, Alloy 718 and Ti-6Al-4V. The scope of process gases was extended from the traditionally employed argon to also include nitrogen, helium and mixtures of argon and helium. Purities from the typical 1000 ppm O2 threshold down to a few ppm were achieved using external monitoring of the atmosphere on both industrial- and laboratory-scale production machines. The investigated materials displayed different sensitivities to the atmosphere composition. 316L stainless steel had limited differences in terms of composition and strength when processed with high purity argon or nitrogen. Only processing with a built-in nitrogen generator, with which the process starts as soon as 10000 ppm residual O2 is reached, led to the increased oxidation of spatter particles and the appearance of large lack-of-fusion defects. A reduction in residual oxygen down to few ppm allowed to significantly hinder the de

关键词： Data processing   Telescopes   Software   Interferometry   Satellites   Parameter estimation   Aerospace engineering   Astronomy   Computer science   Engineering   Optics  

关键词： Design   Mechanics   Physics   Computer science   Materials science   Mathematics   Textile Research  

摘要： Unreinforced woven fabrics are used in a variety of high performance applications, including body armor, deployable structures, and as reinforcements in composites, which are a widely implemented material in various industries. These textile fabrics contain individual fiber filaments at the microscale which are bundled together to produce tows or yarns at the mesoscale. Tows are then interwoven to produce the macroscale fabric. The analysis of parachute fabrics is heavily dependent on legacy data due to the complex nature of the multiscaled fabric structure, identifying the need for accurate models to aid in the design of new materials at a low computational cost. To efficiently simulate the mechanics that occur at the different length scales present, NASA’s Multiscale Analysis Tool (NASMAT), traditionally used for the analysis of reinforced composites, is amended to incorporate unreinforced textile behavior. Using the generalized method of cell micromechanics theory for homogenization of a triply-periodic repeating unit cell (RUC) and allowing the geometry of the tows to change with applied loading, fast and accurate modeling of the tensile behavior of a plain woven fabric is presented. The mesoscale geometry of the fabric is updated as a function of the applied loading, offering predictive capabilities of simulating macroscale fabric behavior that have been validated against experimental data. Tow crimp angles are allowed to vary as a function of the applied local strains by analyzing the repeating volume element (RVE) of the fabric at each load step. Tow rotations are applied to each subcell in the RUC using developed ideal pin-joint kinematic equations for any arbitrary loading direction. The presented formulation allows for highly efficient, predictive modeling of unreinforced fabric behavior for fabrics subject to uniaxial tension at an arbitrary loading angle. The capability of predicting the onset and propagation of failure in plain woven fabrics in NASMAT i

关键词： Computer science   Information technology   Artificial intelligence   Datasets   Ablation   Video recordings   Experiments   Learning   Semantics  

关键词： Computer science   Information Technology   Artificial intelligence   Industrial engineering  

摘要： Driven by the Fourth Industrial Revolution (or Industry 4.0), industrial Internet of Things (IoT) promises one of the largest potential economic effects of IoT. Industrial networks, the underlying support of industrial IoT, typically connect sensors, actuators, and controllers in industrial facilities, such as manufacturing plants, steel mills, and oil refineries. IEEE 802.15.4-based Wireless Sensor-Actuator Networks (WSANs) operate at low-power and can be manufactured inexpensively, which makes them ideal for industrial networks where energy consumption and costs are important. Battery-powered wireless modules easily and inexpensively retrofit existing sensors, actuators, and controllers in industrial facilities without the need to run cables for communication and power. A decade of real-world deployments of WSANs not only demonstrates the feasibility to achieve reliable, real-time wireless communication in industrial environments but also exposes several limitations including error-prone configuration, poor scalability, and insufficient support on mobile platforms. This dissertation presents my research that addresses such limitations. To better configure a WSAN, we develop the Parameter Selection and Adaptation FramEwork (P-SAFE) that leverages extensive wireless simulations to optimally select network parameters based on the application's Quality of Service (QoS) requirements and apply deep learning based domain adaptation to close the simulation-to-reality gap. To enhance network scalability, we develop the Distributed Graph routing and autonomous Scheduling (DiGS) solution that allows network devices to compute their own graph routes and transmission schedules as well as two novel Cross-Technology Communication (CTC) approaches, which enable the deliveries of control commands through a long distance link instead of the hop-by-hop transport. To better support mobile platforms, we develop a new approach that leverages the wireless link characteristics measured b

关键词： Computer science   Behavioral sciences   Artificial intelligence  

摘要： Human face is believed to be one of the most powerful and natural signals for human beings to convey their emotional states and intentions, and has shown great importance in a wide range of applications, such as medicine, marketing and entertainment. Among the many ways of human face behavior understanding, facial expression and facial action units are the most studied. Although significant progress has been made towards improving the performance of facial expression recognition and facial action units detection, the challenges that come from the high inter-subject variations introduced by age, gender, ethnic background and personality, still remain. The relatively small facial dataset, privacy issue for human subjects, and expensive operations for annotation lead to the problem even more challenging. In this thesis, we first target the identity-related challenge and propose two novel methods for facial expression recognition, that are identity-independent and achieve superior performance compared to the state-of-the-art methods. Next, two methods are innovatively designed to extract implicit information for facial action units detection, which advance the performance by either learning extra cues or more effective use of the existing cues. Finally, we focus on the multiple modalities that are available in current facial datasets, and propose to fuse information from different modalities for facial action units detection, achieving significant improvement for AU detection.

关键词： Agronomy   Biochemistry   Bioengineering   Bioinformatics   Botany   Cellular biology   Chemistry   Computer science   Engineering   Genetics  

摘要： In plants, genetic diversity is important for species adaptation in nature and for crop improvement through breeding. For decades, genetic diversity was measured at the nucleotide and gene level. With the abundance of whole genome sequences and chromosome level genome assemblies, the importance of diversity at the genome level - due to rearrangements within genomes - became more obvious given its effect on phenotypic diversity. In genetics research it has become crucial to understand the role of rearrangements in genome evolution and environmental adaptation. Likewise in plant breeding, it is essential to inventory diversity in a wide range of plant varieties for crop improvement as well as to understand how to create new varieties. In this thesis, we touch upon three different genomic rearrangements: recombination, structural variations (SVs) and transposon insertions. We approach them from a fundamental point of view and consider their application to plant breeding. As a first step, we studied genomic recombination investigating the position, distribution and genomic patterns of recombination (crossovers) in offspring from interspecific crosses of tomato (chapter 2). Interspecific recombination is used to introgress wild alleles into elite crops, but this is often problematic, among others due to linkage drag. We learned that recombination in tomato occurs preferentially at the distal ends of chromosomes in open chromatin or euchromatin. On a closer look, at the sequence level, the crossover breakpoints showed preferential occurrence near certain sequence motifs and transcription start sites. The next step was to develop a genome-wide predictor for regions prone to recombination. In chapter 3, we developed a classifier based on genomic patterns associated with recombination based on results presented in the second chapter as well as from studies on other plants (such as Arabidopsis, rice and maize). This model not only accurately predicted recombination probabilit

关键词： Videos   Visualization   Task analysis   Feature extraction   Monitoring   Kinetic theory   Computer science   Gaze tracking   visual field defect   perimetry   eye biomarkers  

摘要： Visual field defects (VFDs) can be caused by numerous diseases, some of which (e.g., glaucoma) are the main causes of blindness in humans. The present perimetry, which is defined as the measurement of visual field function, has high requirements for patients such as long-term fixation and reliable interaction with the system. In this pilot study, we combined the videos and eye-tracking techniques in a free-watching task and explored the different eye movement behaviors of people with several types of artificial VFDs, such as hemianopia, altitudinal VFDs, and tunnel vision. We carried out the task in a gaze-contingent modality where 38 participants with normal vision were recruited to watch a group of videos monocularly. The eye gaze data were recorded by the tracker during the task. We hypothesized that people with simulated VFDs will produce a more active eye movement coping mechanism to compensate for their visual field disadvantage. A new measurement called eye movement amount (MA) was proposed to describe the amount of eye movement toward a specific direction. Statistically significant differences caused by artificial VFDs were observed by comparing the MA values between the VFD groups and the control group. In addition, we found artificial hemianopia and inferior VFDs lead participants to produced increased MA in the horizontal and vertical directions respectively. Artificial tunnel vision, on the contrary, induces a decreased MAs in both horizontal and vertical directions. The proposed metrics can be used as potential biomarkers for distinguishing VFDs in a free-watching task.

关键词： Electrical engineering   Computer science   Mechanical engineering  

摘要： Demand for next-generation power electronic devices is driven by continually evolving requirements of power systems. Devices utilizing III-nitride materials (GaN) and vertical selective-area doped architectures are advantageous due to their wide- bandgap, thermal management, small form-factor, and current handling. Such devices incorporate junctions at multiple crystalline planes. Thus, effects of impurity contamination and etch damage are investigated on the m-plane (10-10) of GaN. Impurites (Si, O, and C) are shown to reduce blocking voltage (~ 102 ×) and increase forward leakage current (~ 104 ×) in regrown versus continuously-grown p-n diodes. Elevated deep level defects at Ec – 1.9, 2.9, and 3.3 eV are identified with increased reverse leakage (~ 103 ×) in etch-and-regrown versus continuously-grown Schottky diodes. Post-dry-etch methods are used to reduce defects and reverse current leakage (~ 10 – 103 ×). Additionally, leakage current mechanisms are investigated via an RF method to extract dynamic parameters of etched-and-regrown p-n diodes.