教学课程资源库 更多>>

关键词： Remote sensing   Computer science   Artificial intelligence   Civil engineering  

摘要： In recent decades, due to global warming and climate change, we have witnessed ever-growing occurrences of natural disasters such as flooding, tornadoes, earthquakes, and wildfires. As such, it is more important than ever to provide emergency response personnel with accurate and timely information for their effective responses to crises. Among the variety of information which is needed for emergency responses and managements, it is vital that response personnel get informed in a timely manner as to where and how severely a building is damaged so that rescue efforts can be the most effective. However, challenges remain despite a great deal of effort in the field of image classification for disaster response. In this work, a promising approach based on deep learning is proposed to detect damaged buildings on high-resolution satellite imagery. By utilizing generic data augmentation, the proposed approach overcame the problem of limited training data popular in many remote sensing applications. A fine-tuning strategy is proposed to use transfer learning with a pretrained model for the task of interest. The experiments with imagery of Port-au-Prince, Haiti demonstrated the proposed method is effective when training data is limited. The Convolutional Neural Network (CNN) model with augmented training data can achieve 83% accuracy in detecting damaged buildings, greatly improved from 53% with the original training data. Future work would be focused on exploring automated methods to acquire larger training datasets and model generalization by investigating more robust data augmentation techniques.

关键词： Software   Embedded systems   Artificial intelligence   Decision making   Taxonomy   Algorithms   Validation studies   Computer science  

摘要： Context:Companies are increasingly collecting data from all possible sources to extract insights that help in data-driven decision-making. Increased data volume, variety, and velocity and the impact of poor quality data on the development of data products are leading companies to look for an improved data management approach that can accelerate the development of high-quality data products. Further, AI is being applied in a growing number of fields, and thus it is evolving as a horizontal technology. Consequently, AI components are increasingly been integrated into embedded systems along with electronics and software. We refer to these systems as AI-enhanced embedded systems. Given the strong dependence of AI on data, this expansion also creates a new space for applying data management techniques. Objective:The overall goal of this thesis is to empirically identify the data management challenges encountered during the development and maintenance of AI-enhanced embedded systems, propose an improved data management approach and empirically validate the proposed approach. Method:To achieve the goal, we conducted this research in close collaboration with Software Center companies using a combination of different empirical research methods: case studies, literature reviews, and action research. Results and conclusions:This research provides five main results. First, it identifies key data management challenges specific to Deep Learning models developed at embedded system companies. Second, it examines the practices such as DataOps and data pipelines that help to address data management challenges. We observed that DataOps is the best data management practice that improves the data quality and reduces the time to develop data products. The data pipeline is the critical component of DataOps that manages the data life cycle activities. The study also provides the potential faults at each step of the data pipeline and the corresponding mitigation strategies. Finally, the dat

关键词： Alternative Energy   Atmospheric sciences   Climate Change   Computer science   Energy   Web Studies  

关键词： Mechanical engineering   Design   Computer engineering   Computer science   Electrical engineering  

摘要： Vibrotactile feedback is a key feature of many small touchscreen devices, but is often absent or incomplete in large touchscreen displays due to a lack of suitable actuators for such applications. Thus, a growing need exists for haptic actuators capable of producing meaningful feedback in large touch displays. This study proposes and evaluates a dual-electrode electrostatic resonant actuator (ERA) as a means to fulfill such a need. The dual-electrode ERA was compared to a similar single electrode ERA to study the effect of electrode configuration. It produced a maximum vibration 73% higher than the single-electrode actuator, showing promising potential for its use in large touchscreen applications. To study the ERA in a large display application, a prototype touch bar system with spring boundaries was designed, fabricated, and evaluated. By varying the number of actuators excited in the system, the actuators’ magnitude, excitation frequency, and signal duration, a maximum vibration of 4 g-forces could be achieved throughout the majority of the display in both sustained and pulse sensations. This demonstrates a promising potential for generating a freely positionable and fully controllable point of vibrotactile stimulation at any point of a touch bar display. These results show the feasibility of the actuator spring boundary implementation and the dual-electrode ERA for large touchscreen display applications.

关键词： Mechanics   Biology   Biophysics   Cellular biology   Computer science   Medical imaging   Physics  

关键词： Motivation   Deep learning   Artificial intelligence   Signal processing   Neural networks   Algorithms   Computer science   Electrical engineering   Engineering  

关键词： Computer science   Transportation   Transportation planning   Internet of Things   Travel   Older people   Traffic congestion  

关键词： Engineering   Computer science  

关键词： Computer engineering   Artificial intelligence   Computer science  

摘要： Outdoor positioning systems based on the Global Navigation Satellite System have several shortcomings that have deemed their use for indoor positioning impractical. Location fingerprinting, which utilizes machine learning, has emerged as a viable method and solution for indoor positioning due to its simple concept and accurate performance. In the past, shallow learning algorithms were traditionally used in location fingerprinting. Recently, the research community started utilizing deep learning methods for fingerprinting after witnessing the great success and superiority these methods have over traditional/shallow machine learning algorithms. The contribution of this dissertation is fourfold: First, a Convolutional Neural Network (CNN)-based method for localizing a smartwatch indoors using geomagnetic field measurements is presented. The proposed method was tested on real world data in an indoor environment composed of three corridors of different lengths and three rooms of different sizes. Experimental results show a promising location classification accuracy of 97.77% with a mean localization error of 0.14 meter (m). Second, a method that makes use of cellular signals emitting from a serving eNodeB to provide symbolic indoor positioning is presented. The proposed method utilizes Denoising Autoencoders(DAEs) to mitigate the effects of cellular signal loss. The proposed method was evaluated using real-world data collected from two different smartphones inside a representative apartment of eight symbolic spaces. Experimental results verify that the proposed method outperforms conventional symbolic indoor positioning techniques in various performance metrics. Third, an investigation is conducted to determine whether Variational Autoencoders (VAEs) and Conditional Variational Autoencoders (CVAEs) are able to learn the distribution of the minority symbolic spaces, for a highly imbalanced fingerprinting dataset, so as to generate synthetic fingerprints that promote enhan

关键词： Quantum physics   Optics   Computer science  

摘要： Color centers in diamond are attractive candidates for implementing single-atom quantum memories in a quantum network. This thesis describes an approach to build quantum networks nodes based on color centers in diamond. We propose to use a novel single-atom quantum memory, the neutral charge state of silicon vacancy (SiV0), as the building block for future quantum network. The unique combination of long spin coherence times and efficient optical transitions makes SiV0 a promising candidate for such application. Leveraging the excellent spin and optical properties of SiV0, we design a hybrid III-V diamond photonic platform that can both enhance the photon emission of SiV0 and perform on-chip frequency conversion to the telecommunication C-band. As a first step towards building quantum network nodes based on SiV0, we design, fabricate and characterize nanophotonic cavities on the GaAs-on-diamond platform. Preliminary results show that a quality factor of 1,328 can be achieved despite the challenges in fabrication. Coupling the cavity to a single SiV0 center in diamond could in principle enable a Purcell enhancement of the SiV0 emission with a factor of 64. This could potentially enhance spin readout and spin-photon entanglement fidelity for SiV0. Eventually, the results and techniques described in this thesis could contribute to the developments of multi-node quantum networks that span across the globe.