教学课程资源库 更多>>

关键词： Semantic segmentation   Semantics   Training   Annotations   Unsupervised learning   Convolutional neural networks   Computer science   Unsupervised semantic segmentation   self-supervised learning   semantic clustering   pseudo labeling   context-aware embedding  

摘要： Despite the great progress of semantic segmentation with supervised learning, annotating large amounts of pixel-wise labels is, however, very expensive and time-consuming. To this end, Unsupervised Semantic Segmentation(USS) has been proposed to learn semantic segmentation, without any form of annotations. This approach involves dense prediction of semantics which is however challenging due to the unreliable nature of local representations. To solve this problem, we propose a newly context-aware unsupervised semantic segmentation framework, which aims to enhance the unsupervised semantic segmentation by leveraging contextual knowledge within and across images. In particular, we introduce a training strategy based on our Pyramid Semantic Guidance (PSG), which utilizes holistic semantics on pyramid views to guide pixel clustering with a siamese network-based framework. Additionally, we introduce a Context-Aware Embedding (CAE) module to fuse global features with low-level geometrical and appearance representations. We evaluate our method on the COCO-Stuff dataset and achieved competitive results compared to both the convolutional and ViT-based USS methods. Specifically, we attain significant improvements of +4.5% and +5% mIoU for Stuff and all class segmentation respectively, compared to previous approaches that employ unsupervised convolutional backbones.

关键词： Computer science   artificial intelligence  

ISBN: (数字)9781394167562;9781394167579 ISBN: (纸本)9781394166398

摘要： Discover the groundbreaking potential of augmented reality in special education with this comprehensive book, which fills a significant research gap and explores the transformative impact of AR/VR on teaching and learning experiences for individuals with disabilities. Millions of young people across the world have impairments. Because of their apparent problems, these young people have typically been marginalized or excluded from schools. Studies on augmented reality applications in education for special children are still in their early stages and there is a dearth of research on the effects and implications of augmented reality in education for special children or individuals with special needs, such as intellectual disabilities, autism spectrum disorder, attention deficit hyperactivity disorder, and physical disabilities. Technological advancements have enabled the creation of whole new learning environments, vastly expanding the breadth and sophistication of teaching and learning activities. For example, several studies have shown that virtual reality and augmented reality can help autistic people understand facial emotions and improve their social skills. By compiling a collection of conceptual and research chapters investigating the infiltration of AR/VR into teaching and learning practices and experiences for disabled people, this book will fill a significant gap in current research literature.

关键词： Random matrices   Non-asymptotic random   Computer science application   Random matrix models  

摘要： Non-asymptotic random matrix theory focuses on obtaining quantitative high-probability estimates for the spectral properties of random matrices with large but fixed sizes. Such results are important in computer science applications. Also, in these applications, classical models of random matrices, e.g., the i.i.d. entries models, are not sufficient, and more complicated new models, ones that involve deterministic shifts, submatrices, and dependence between entries, are needed. This work is devoted to the study of non-asymptotic properties of these structured random matrix models and their implications in computer science *** first part of this thesis studies a shifted random matrix. More precisely, let R be a n by n random matrix with i.i.d. subgaussian entries. Let M be a n by n deterministic matrix with controlled norm. This dissertation gives a general estimate of the smallest singular value of the sum R + M, which improves upon an earlier result of Tao and *** second part of this thesis studies the existence of well conditioned n by n square submatrices in a large fat n by M matrix where M is much larger than n, and its implications on the existence of Riesz bases in random frames. We establish a phase transition for the probability that a random frame contains a Riesz basis. Namely, we first show that a random frame in Rn formed by M vectors with independent identically distributed coordinate having a non-degenerate symmetric distribution contains many Riesz bases with high probability provided that M ≥ exp(Cn). In the proof of this result, we use approximately Hadamard matrices as pattern matrices, where approximately Hadamard matrices are n by n matrices with ±1 entries which acts on Rn as a approximate scaled isometry. The existence of such matrices in any dimension n is proved by combining number-theoretic and probabilistic tools. On the other hand, we prove that if the entries are subgaussian, then a random frame fails to contain a Riesz ba

摘要： 【教学内容】人教版八年级上册第六单元“I’m going to study computer science”阅读课【案例分析】本单元属于人与自我、人与社会范畴，从主题群角度分析属于生活与学习、做人与做事的主题，包含自我认识、自我管理、自我提升、职业启蒙、公共服务等多方面内容。本单元的主要话题是“生活目标（Life goals）”，以谈论对未来的打算为主要内容，安排了四个对话以及两个阅读，整个单元的情境相对清晰，因此听、说、读、写活动的开展也要在整个情境中组织，

关键词： Culturo-techno-contextual approach   Difficult concepts education pedagogies   Spreadsheet education  

摘要： Purpose: The lecture method has been compared with teaching methods such as flip learning, cooperative learning and simulations to establish which holds the key to students' understanding of concepts. What is bereft in the education literature is its comparative efficiency with the culturo-techno contextual approach (CTCA) in the teaching of computer science education. Design/methodology/approach: This study adopted the quasi-experimental design to determine the efficacy of the CTCA in breaking difficulties related to the study of spreadsheets as a difficult concept in the Nigerian computer science education curriculum. Junior high school students studying computer science education participated in the study. The control group had 30 students, with 35 students in the experimental group. The experimental group was taught using CTCA, while the control group used the lecture method. The spread sheet achievement test, which had 40 items on spreadsheet, was used to collect data. Findings: The results showed that the experimental group significantly outperformed the control group [F (1,60) = 41.89;p < 0.05]. The findings showed the potential of CTCA in improving students' performance in spreadsheets in the computer science education curriculum. Originality/value: The originality of this study is hinged on its ground-breaking test of the CTCA to the study of the spreadsheet. The findings of this study indicate its efficacy in improving students' understanding of spreadsheet and computer science education. © 2023, Mariyam Abdulhadi, Fred Awaah, Deborah Agbanimu, Emmanuel Okyere Ekwam and Emmanuella Sefiamor Heloo.

关键词： 学习活动观   初中英语   阅读教学   思维能力  

摘要： 阅读教学是英语课堂教学中语言输入的重要环节，是学生通过文本阅读，深入理解文本内涵，内化与运用所学的语言知识和语言技能，并在主题意义的引领下解读文本内容的教学活动。本文结合目前阅读教学中存在的问题，并以人教版初中英语八年级上册Unit 6“I'm going to study computer science”Section B 2b为例，从文本解读、目标分析、活动设计三个方面提出了基于学习活动观视域下的初中英语阅读教学中培养学生思维能力的实施途径。通过学习理解、应用实践、迁移创新类活动设计，以读启思、导入激活，以读遐思、提炼文本，以读求思、超越文本，从而升华主题、锻炼思维，提升课堂阅读教学效果，激发学生创新思维，培养学生创造能力，促进学生核心素养的形成和发展。

关键词： Higher education   gender equality   Brazil   Higher education   gender equality   Brazil  

摘要： Challenges related to gender equality are being researched by different institutions, universities, and companies worldwide. In particular, there are still gaps concerning higher education in Science, Technology, Engineering, and Mathematics (STEM). Through a Rapid Review, this work aims to present the factors that affect women's journey in STEM education in Brazil, mainly in engineering and computer science. More specifically, we seek to describe the barriers, motivators, and strategies present in the researched academic literature. Barriers include stereotypes, oppression, patriarchal relations, prejudice, segregation, isolation, lack of support from colleagues and teachers, judgment, fear of mathematics, and mansplaining - the unnecessary explanation made by a man of something a woman says. Motivators include support from the school and teachers, exposure to the STEM area during high school, teamwork, and support from family, friends, and classmates. The strategies to overcome barriers women face in STEM education include exposure to STEM topics, the inclusion of teachers in encouraging and training, role models, participation in technological competitions, a friendly environment, and women participating in technical evaluators' work. This work presented data revealing women's challenges in STEM, highlighting existing barriers, motivators, and strategies tested to address the low female participation. This information significantly expands the discussion on this topic, especially in the labor market and education. In our view, quality and excellence depend on an inclusive policy where selection is based solely on merit, regardless of gender, ethnicity, or religion.

ISBN: (纸本)9781805432845;1805432842

摘要： Examines how current novels dealing with African migrationaddress social issues, immigrant subjectivity, and the politics ofmigration. African Migration and the...

关键词： Computer science   artificial intelligence   electrical engineering communications  

ISBN: (数字)9788770040785 ISBN: (纸本)9788770040792

摘要： For a decade, with the uptake of 4G, we have become accustomed to the relentless increase in data and services on the move. The deployment of 5G is advancing crucial key performance indicators (KPIs), along with quality of service (QoS). Setting the horizon to 2030 and later, 6G will take the KPIs to numbers 100–1000 times better than 5G. Yet, the actual disruption of 6G and future networks (FN) will take place following other unprecedented paths. Artificial intelligence (AI) will be exploited in a threadlike fashion, at any level of the network physical infrastructure. This will introduce, to date unknown features, like self-sustaining, self-evolution and high-resilience of small portions of the infrastructure, pioneering the concept of a network of networks. Each segment of the infrastructure will bear a high degree of independence, while working at the same time as a whole, in full orchestration with the rest of the network. Given such a scenario, this book claims that the established and currently in use paradigms for the design and development of hardware–software (HW–SW) systems, are not appropriate to address the challenges of 6G and, further ahead, of FN. In response, unprecedented design approaches are suggested, relying on a fresh reinterpretation of the standard concept of HW, with specific attention to the network edge and edge intelligence (EI). This work develops some conceptual tools that may help address the technical challenges resulting from the intricate scenario sketched above. Within the mentioned HW reconceptualization, a pivotal role is forecasted for microtechnologies and nanotechnologies, intended with a broad meaning, which embraces, among others, devices, systems (MEMS/NEMS) and materials.

关键词： Computer science  

ISBN: (数字)9781394262427 ISBN: (纸本)9781394262410

摘要： This book delivers a comprehensive overview of the Bot threat landscape and defense strategies in plain language to lift the mystery and make bots and fraud detection more accessible to web security professionals and website owners. It provides some insight into the evolution of the attack and defense strategy over time, the motivation of the attackers, how detection methods work, and how to analyze the traffic to assess the accuracy and decide on the most appropriate response strategy. The knowledge acquired from this book will help security teams regain their advantage over attackers. After explain how bots and the abuse they bring came to be possible because of the way the internet architecture works, the author dives into the most common types of attacks, detection methods, assessing bot activity, and your strategies for defense and response. The books also explores the conflicting challenges of the need for privacy versus security. By the end, web security professionals with all levels of prior experience will be able to defend against bots and their abuse and damage. Security executives and decision makers as well as technology managers and architects will better understand how to build systems that are reliant to bots by design and data scientists working in cyber security will be better able to comprehend and communicate the scope of bot problems.