教学课程资源库 更多>>

关键词： Innovations   Smart grid technology   Software   Ethnography   Alternative energy   Emotions   Households   Empowerment   Sustainability   Computer science   Energy   European studies  

摘要： Smart grids are heralded as promising socio-technical innovation and a comprehensive solution for making the energy grid more flexible and green. Smart grids innovations presume the involvement of householders, who, by conserving, monitoring, and timing their consumption with the help of various smart technologies and information flows, contribute to the sustainability and stability of the electricity grid. In this context, grid operator, energy producing companies, governments and other actors are redefining their relationships to each other and to citizens. These actors have to face the desires, values and actions of citizens. Meeting this challenge requires a closer investigation of the character and trajectory of householder engagement with smart grids. Further research is required into the needs and desires which (de)motivate householder engagement with smart grids. An analysis of the human scale of smart grids must account for the collectively shared meanings, aspirations and emotional attachments of householders. For these reasons, this thesis dives more deeply into the normative and motivational dimensions of householder participation and agency in smart grids. The thesis aims to enhance our understanding of emerging and transforming domestic energy practices enabled by smart grid innovations, and their consequences for sustainable grid management and the relationship between system actors, energy collectives, and households in Europe. The character of householder participation in smart grids is assessed critically in terms of empowerment to self-organize or co-construct democratic, sustainable energy systems and sustainable lifestyles. The term ‘human scale’ is introduced as means of broadening our view of householder involvement in energy transitions far beyond formal processes of participation. Moreover, the thesis aims to generate insights and recommendations into how participation of householders in smart grids could be supported by governance strategie

关键词： Computer science   Artificial intelligence  

摘要： In recent years, there has been significant progress in deep learning and computer vision, with many models proposed that have achieved state-of-art results on various image recognition tasks. However, to explore the full potential of the advances in this field, there is an urgent need to push the processing of deep networks from the cloud to edge devices. Unfortunately, many deep learning models cannot be efficiently implemented on edge devices as these devices are severely resource-constrained. In this thesis, I present QU-Net, a lightweight binary segmentation model based on the U-Net architecture. Traditionally, neural networks consider the entire image to be significant. However, in real-world scenarios, many regions in an image do not contain any objects of significance. These regions can be removed from the original input allowing a network to focus on the relevant regions and thus reduce computational costs. QU-Net proposes the salient regions (binary mask) that the deeper models can use as the input. Experiments show that QU-Net helped achieve a computational reduction of 25% on the Microsoft Common Objects in Context (MS COCO) dataset and 57% on the Cityscapes dataset. Moreover, QU-Net is a generalizable model that outperforms other similar works, such as Dynamic Convolutions.

关键词： Computer science   Artificial intelligence  

关键词： Computer science   Electrical engineering   Artificial intelligence  

摘要： With the growing complexity of big data workloads that require abundant data and computation, data centers consume a tremendous amount of power daily. In an effort to minimize data center power consumption, works in the literature developed power models that can be used for load balancing algorithm decision making. There are several difficulties that make power modeling a nontrivial task. For one, the inter-dependencies between components must be captured in addition to the direct effect components have on energy consumption. Additionally, a power model should persist for a wide variety of workloads and generalize to different platforms. In the past, analytical power models mainly focused on single-variable CPU utilization functions to predict power consumption. However, with the increase of system memory and disk/cache accesses in emerging workloads like machine learning, these models face a significant accuracy reduction because they do not account for power hungry memory related events. As a result, machine-learning-based power models attempt to consider relevant parameters that capture these memory relevant events in emerging workloads. However, these machine-learning-based power models exhibit high latency and accuracy improvement mostly on non-compute-intensive workloads. Additionally, machine learning power models are not easily generalized to different platforms because they are trained with profiling data of certain server nodes, given the increasing hardware heterogeneity in data centers. Additionally, as more datacenters are migrating to run containerized applications to leverage the isolation, security and scalability of containerized technology, there is a lack of study in the literature regarding container power modeling. Thus, load migration algorithms are unable to make accurate migrating decisions regarding the appropriate containers. This thesis tackles these issues by proposing several ideas. First, a hybrid power model is proposed that selects th

关键词： Internet of Things   Decomposition   Cloud computing   Performance evaluation   Optimization algorithms   Electric vehicles   Genetic algorithms   Decision making   Artificial intelligence   Computer science   Transportation  

摘要： Many real-world problems involve multiple and conflicting objectives to be optimized simultaneously, which are formulated as multi-objective optimization problems (MOPs). In general, problems with more than three objectives or a large number of decision variables are referred to as many-objective or large-scale optimization problems (MaOPs or LSMOPs), respectively. Recently, a variety of multi-objective evolutionary algorithms (MOEAs) inspired by nature have been developed to solve MOPs. However, the performance of MOEAs often deteriorates when faced with many-objective or large-scale optimization. In addition, these MOEAs using the classical Pareto dominance principle face no strong selection pressure towards the Pareto front with regard to many-objective optimization. Furthermore, it is difficult for a limited population size to explore the high-dimensional search space of large-scale optimization. With the development of the Internet of things (IoT) and mobile networks, more and more computation-intensive and latency-critical applications are deployed to mobile devices. Due to some limitations inherent to mobile devices including limited computing capability, storage space, and battery lifetime, these applications cannot run efficiently on mobile devices. To address this issue, computation offloading in mobile cloud and edge computing (MCC and MEC) provides a promising paradigm to migrate computation-heavy parts of mobile applications to the cloud and edge servers. Hence, MCC and MEC computation offloading may involve different objectives such as reducing time and saving energy. Offloading decision making can be described as multi-criteria optimization and we develop efficient MOEAs to solve different computation offloading models. This work covers evolutionary multi-objective optimization for computation offloading in collaborative MCC and MEC. Its main content includes two parts: (i) evolutionary multi-objective optimization and (ii) optimization in offloading.

关键词： Artificial intelligence   Computer science  

摘要： The World Wide Web contains countless semi-structured websites, which present information via text embedded in rich layout and visual features. These websites can be a source of information for populating knowledge bases if the facts they present can be extracted and transformed into a structured form, a goal that researchers have pursued for over twenty years. A fundamental opportunity and challenge of extracting from these sources is the variety of signals that can be harnessed to learn an extraction model, from textual semantics to layout semantics to page-to-page consistency of formatting. Extraction from semi-structured sources has been explored by researchers from the natural language processing, data mining, and database communities, but most of this work uses only a subset of the signals available, limiting their ability to scale solutions to extract from the large number and variety of such sites on the Web. In this thesis, we address this problem with a line of research that advances the state of semi-structured extraction by taking advantage of existing knowledge bases, as well as using modern machine learning methods to build rich representations of the textual, layout, and visual semantics of webpages. We present a suite of methods that will enable information extraction from semi-structured sources, addressing scenarios that include both closed and open domain information extraction and varying levels of prior knowledge about a subject domain.

关键词： Computer science   Computer engineering   Artificial intelligence   Neural networks   Accuracy   Methods   Algorithms   Success   Experiments  

摘要： Machine learning models, including neural networks, have gained great popularity in recent years. Deep neural networks are able to directly learn from raw data and can outperform traditional machine learning models. As a result, they have been increasingly used in a variety of application domains such as image classification, natural language processing, and malware detection. However, deep neural networks are demonstrated to be vulnerable to adversarial examples at the test time. Adversarial examples are malicious inputs generated from the legitimate inputs by adding small perturbations in order to fool machine learning models to misclassify. We mainly aim to answer two research questions in this thesis: How are machine learning models vulnerable to adversarial examples? How can we better defend against the adversarial examples? We first improve the effectiveness of adversarial training by designing an experimental framework to study Method-Based Ensemble Adversarial Training (MBEAT) and Round Gap Of Adversarial Training (RGOAT). We then demonstrate the strong distinguishability of adversarial examples and design a simple yet effective approach called defensive distinction under the formulation of multi-label classification to protect against adversarial examples. We also propose fuzzing-based hard-label black-box attacks against machine learning models. We design an AdvFuzzer to explore multiple paths between a source image and a guidance image, and design a LocalFuzzer to explore the nearby space around a given input for identifying potential adversarial examples. Lastly, we propose a key-based input transformation defense to defend against adversarial examples.

摘要： In the post-9/11 world, Canada has struggled with developing a sentencing regime that effectively punishes and deters defined terrorist activity such as the attack on the Canadian Parliament, the Danforth shootings, the rise of Khalistani and Islamic terrorism, and the Toronto Van Attack. Broadly speaking, the Canadian public still supports capital punishment, but it is unclear whether Canadian prosecutors perceive and view the issue in the same light in their professional and legal capacity. Canadian prosecutors are tasked with seeking sentences that meet criminal justice principles, including the principle of deterrence. Their views on what punishments are just and effective are important, as they have significant and broad discretion in asking the courts to impose sentences. Thus, while Clarence Darrow argued that “We have heard talk of justice. Is there anybody who knows what justice is? No one on earth can measure out justice.”, the criminal justice system functions within the legal fiction that prosecutors can make decisions on what justice can and should be. The present study has used a grounded theory design by collecting data from Canadian prosecutors on their views about capital punishment as it relates to defined terrorist activity and their roles concerning administering justice on behalf of the public. A theme emerging from the interviews was that the general deterrence principle of sentencing was either false or too hard to quantify to make it a meaningful factor in sentencing. This study may contribute to a positive social change by helping develop a theory about how Canadian prosecutors view the role of punishment, their role as prosecutors in the criminal justice system, and how they view the effectiveness of deterrence as it relates to defined terrorist activity and capital punishment.

关键词： Computer science   Constructivism   Education   Math   Modulo   Modulus  

摘要： The modulo operator is a simple mathematical concept that is almost universally taught in early computer science. While remainders are introduced in elementary mathematics curricula, the modulo operator is generally not discussed in K-12 mathematics and the concept of remainders receives little attention throughout middle and high school mathematics coursework. This thesis examines what prior knowledge high school computer science students utilize when challenged with a story problem whose ideal solution involves using the modulo operator in assigning objects to bins. In this study, students provided plain English solutions to isomorphic versions of an open-ended problem statement before and after formal instruction about the modulo operator. Solutions were qualitatively tagged for strategies and terminology used, and solution correctness was assessed using an adapted scale based on the ''Structure of the Observed Learning Outcome'' (SOLO) taxonomy. A mixed-methods analysis strategy was used to provide insight into how students approached these problems and how their answers changed after instruction. Students demonstrated a wide range of strategies in the pretest, but no pretest solutions contained direct indicators of students leveraging prior knowledge of the modulo operator or the concept of remainders. However, students did reveal a natural tendency to rely on concepts related to conditional logic and described methods of wrapping back to the first bin, a behavior implemented by using the modulo operator. Posttest results showed significantly greater usage of remainders and modulo operator concepts, and posttest answers were significantly more sophisticated relative to the SOLO taxonomy. Thus, computer science instructors should remember that the applicative use of the modulo operator may appear unfamiliar to their students when seen for the first time. However, instructors may find the context of continually wrapping or cycling back to a beginning more natural

关键词： Engineering   Computer engineering   Computer science  

摘要： Wide Bandgap (WBG) semiconductors like Silicon Carbide (SiC), Gallium Nitride (GaN) and Aluminum Gallium Nitride (AlGaN) have superior material properties as compared to Silicon (Si) like higher electrical breakdown voltages and bandgap energies as well as lower leakage currents as compared to Si which make them ideal to operate at higher voltage with lower thermal losses. These properties make WBG materials ideal for power devices like Vertical Double-diffused Metal Oxide Semiconductor Field Effect Transistors (VDMOSFETs). The use of digital prototyping through computer simulation increases the speed and flexibility of the design iterations while reducing the cost and time required for the design process. COMSOL Multiphysics is a Finite Element Method simulation software that has capabilities of combining different physics interfaces to simulate the effects of multiple interdependent physical phenomena. The use of these materials in switching devices like VDMOSFETs have been modelled in COMSOL Multiphysics in 2D and 3D for the purposes of this research dissertation. The electrical and thermal advantages of WBG materials, specifically SiC, GaN and AlGaN, as compared to Si as semiconductor materials for VDMOSFET structures for the exact same VDMOSFET structures are demonstrated and quantified from the results *** inverter is the most important component in a Photo Voltaic (PV) system that needs to be improved. Transformerless inverters have higher efficiencies at lower weight and size specifications as compared to ones with transformers. A modified topology of a single phase transformerless inverter with new current paths and improved efficiencies is proposed and its performance is analyzed in PSIM software with Si and WBG material power switching devices. The advantages of the WBG devices over Si in terms of power losses are also exhibited in this research. The power losses obtained from the models in PSIM are then used as inputs to COMSOL models for temper