Learning analytics tasks as services in smart classrooms

A smart classroom integrates the different components in a traditional classroom, by using different technologies as artificial intelligence, ubiquitous, and cloud paradigms, among others, in order to improve the learning process. On the other hand, the learning analytics tasks are a set of tools that can be used to collect and analyze the data accumulated in a smart classroom. In this paper, we propose the definition of the learning analytics tasks as services, which can be invoked by the components of a smart classroom. We describe how to combine the cloud and multi-agent paradigms in a smart classroom, in order to provide academic services to the intelligent and non-intelligent agents in the smart classroom, to adapt and respond to the teaching and learning requirements of students. Additionally, we define a set of learning analytics tasks as services, which defines a knowledge feedback loop for the smart classroom, in order to improve the learning process in it, and we explain how they can be invoked and consumed by the agents in a smart classroom.     

THE GRID-E-CARD: ARCHITECTURE TO SHARE COLLECTIVE INTELLIGENCE ON THE GRID

Using the Grid for electronic learning can be considered a major innovation that increases the potential of the Internet for collaborative learning and transforms it into a gigantic high speed network of knowledge and services. Furthermore, with the refinement of peer-to-peer communication, the emerging network models will allow thousands of learners to access these solutions, regardless of their different computer systems. These solutions can be compared to storage devices that share learning resources according to the user' s needs. With the substantial increase of information nodes and the multiplicity of computers making use of this complex network, cognitive overload or transactional distance inherent to the technology can quickly become an obstacle to the learning process. Therefore, the Grid requires the development of appropriate learning tools and services. This paper presents an approach to reduce cognitive overload and transactional distance for a virtual learning community (VLC) on the Grid. This paper proposes a computer-Grid communication device called “Grid-e-Card.” Its goal is to bring users together according to their signature to share collective intelligence in a social context: the knowledge they have acquired, the objectives they wish to meet, or the learning services that correspond to their needs. This methodology is based on a set of P2P-agents who handle users' electronic portfolios (e-Portfolios) as knowledge prosthesis and exploit e-Learning qualification (e-Qualification) processes as aggregation methods to dynamically organize people in relevant virtual learning communities.     

Adversarial attacks on text classification models using layer-wise relevance propagation

Due to the nested nonlinear structure inside neural networks, most existing deep learning models are treated as black boxes, and they are highly vulnerable to adversarial attacks. On the one hand, adversarial examples shed light on the decision-making process of these opaque models to interrogate the interpretability. On the other hand, interpretability can be used as a powerful tool to assist in the generation of adversarial examples by affording transparency on the relative contribution of each input feature to the final prediction. Recently, a post-hoc explanatory method, layer-wise relevance propagation (LRP), shows significant value in instance-wise explanations. In this paper, we attempt to optimize the recently proposed explanation-based attack algorithms (EAAs) on text classification models with LRP. We empirically show that LRP provides good explanations and benefits existing EAAs notably. Apart from that, we propose a LRP-based simple but effective EAA, LRPTricker. LRPTricker uses LRP to identify important words and subsequently performs typo-based perturbations on these words to generate the adversarial texts. The extensive experiments show that LRPTricker is able to reduce the performance of text classification models significantly with infinitesimal perturbations as well as lead to high scalability.     

整合全局—局部度量学习的人体目标再识别

目的 人体目标再识别的任务是匹配不同摄像机在不同时间、地点拍摄的人体目标。受光照条件、背景、遮挡、视角和姿态等因素影响,不同摄相机下的同一目标表观差异较大。目前研究主要集中在特征表示和度量学习两方面。很多度量学习方法在人体目标再识别问题上了取得了较好的效果,但对于多样化的数据集,单一的全局度量很难适应差异化的特征。对此,有研究者提出了局部度量学习,但这些方法通常需要求解复杂的凸优化问题,计算繁琐。方法 利用局部度量学习思想,结合近几年提出的XQDA（cross-view quadratic discriminant analysis）和MLAPG（metric learning by accelerated proximal gradient）等全局度量学习方法,提出了一种整合全局和局部度量学习框架。利用高斯混合模型对训练样本进行聚类,在每个聚类内分别进行局部度量学习;同时在全部训练样本集上进行全局度量学习。对于测试样本,根据样本在高斯混合模型各个成分下的后验概率将局部和全局度量矩阵加权结合,作为衡量相似性的依据。特别地,对于MLAPG算法,利用样本在各个高斯成分下的后验概率,改进目标损失函数中不同样本的损失权重,进一步提高该方法的性能。结果 在VIPeR、PRID 450S和QMUL GRID数据集上的实验结果验证了提出的整合全局—局部度量学习方法的有效性。相比于XQDA和MLAPG等全局方法,在VIPeR数据集上的匹配准确率提高2.0%左右,在其他数据集上的性能也有不同程度的提高。另外,利用不同的特征表示对提出的方法进行实验验证,相比于全局方法,匹配准确率提高1.3%~3.4%左右。结论 有效地整合了全局和局部度量学习方法,既能对多种全局度量学习算法的性能做出改进,又能避免局部度量学习算法复杂的计算过程。实验结果表明,对于使用不同的特征表示,提出的整合全局—局部度量学习框架均可对全局度量学习方法做出改进。     

Improving device-aware Web services and their mobile clients through an aspect-oriented,model-driven approach

ContextMobile devices have become an essential element in our daily lives, even for connecting to the Internet. Consequently, Web services have become extremely important when offering services through the Internet. However, current Web services are very inflexible as regards their invocation from different types of device, especially if we consider the need for them to be adaptable when being invoked from mobile devices.ObjectiveIn this paper, we provide an approach for the creation of flexible Web services which can be invoked transparently from different device types and which return subsequent responses, as well as providing the client’s adaptation as a result of the particular device characteristics and end-user preferences in a completely decoupled way.MethodAspect-Oriented Programming and model-driven development have been used to reduce both the impact of service and client code adaptation for multiple devices as well as to facilitate the developer’s task.ResultsA model-driven methodology can be followed from system models to code, providing the Web service developer with the option of marking which services should be adapted to mobile devices in the UML models, and obtaining the decoupled adaptation code automatically from the models.ConclusionWe can conclude that the approach presented in this paper provides us with the possibility of following the development of mobile-aware Web services in an integrated platform, benefiting from the use of aspect-oriented techniques not only for maintaining device-related code completely decoupled from the main functionality one, but also allowing a modularized non-intrusive adaptation of mobile clients to the specific device characteristics as well as to final user preferences.     

A survey on the use of machine learning methods in context-aware middlewares for human activity recognition

Human activity recognition (HAR) essentially uses (past) sensor data or complex context information for inferring the activities a user performs in his daily tasks. HAR has been extensively studied using different paradigms, such as different reasoning mechanisms, including probabilistic, rule-based, statistical, logical reasoning, or the machine learning (ML) paradigm, to construct inference models to recognize or predict user activities. ML for HAR allows that activities can be recognized and even anticipated through the analysis of collected data from different sensors, with greater accuracy than the other paradigms. On the other hand, context-aware middlewares (CAMs) can efficiently integrate a large number of different devices and sensors. Moreover, they provide a programmable and auto-configurable infrastructure for streamline the design and construction of software solutions in scenarios where lots of sensors and data are their bases, such as ambient intelligence, smart cities, and e-health domains. In this way, the full integration of ML capabilities as services in CAMs can advance the development of software solutions in these domains when ML is necessary, specially for HAR, which is the basis for many scenarios in these domains. In this work, we present a survey for identifying the state-of-the-art in using ML for HAR in CAMs through a systematic literature review (SLR). In our SLR, we worked to answer four research questions: (i) what are the different types of context reasoners available in CAMs; (ii) what are the ML algorithms and methods used for generating models for context reasoning; (iii) which CAMs support data processing in real time; and (iv) what are the HAR scenarios usually tackled by the research works. In our analysis, we observed that, although ML offers viable approaches to construct inference models for HAR using different ML approaches, including batch learning, adaptive learning and data stream learning, there are yet some gaps and research challenges to be tackled, specially on the use of data stream learning considering concept drift on data, mechanisms for adapting the inference models, and further considering all of this as services in CAMs, specially for HAR.

     

分形曲线生成的频域方法

目的 分形几何学的理论研究与应用实践方兴未艾,在分形的计算机生成领域,传统方法是在空间域中,通过对生成元的迭代操作而形成。为了扩展分形的生成方法,本文将频谱分析引入到分形几何中。方法 正交函数系是频谱分析的核心问题之一。考虑到分形曲线是一类连续而不光滑的折线型信号,通常的三角函数（Fourier变换）、连续小波变换仅适用于光滑的对象,否则会出现所谓“Gibbs现象”;另一方面,以V-系统为代表的正交分段多项式函数系适用于表达包含间断性的对象,否则会出现信息冗余。因此,通常的正交函数系均不适合分形的频谱表达与分析。针对分形曲线的特点,本文将其视为一次样条函数,通过引入一类正交样条函数系-Franklin函数系,实现了对分形曲线的有限项精确正交表达,得到Franklin频谱,从而完成分形的时频变换。然后,对Franklin频谱系数在不同尺度上进行修改。最后,通过正交重构得到新的分形。结果 对比实验验证了Franklin函数系在分形曲线频域表达方面的优越之处,它既能通过最小项数实现分形的正交表达,而且不会出现Gibbs现象。本文以von Koch曲线、Sierpinski square曲线和Hilbert曲线这3个经典分形为例,通过对Franklin谱在不同尺度上的自由调节,能够方便地生成大量形态各异的新的分形曲线。结论 Franklin谱不仅能够实现对分形曲线的有限精确重构,而且还能在不同尺度上刻画分形的形态特征。基于Franklin频谱调节实现的分形生成方法,只要修改频谱就可以得到大量的新型分形曲线,而且这些分形的样式千变万化,几乎不可预测,这种分形生成方式为分形设计带来了巨大的自由空间,为分形的生成提供了新的思路与方案。     

混合生成式和判别式模型的图像自动标注

目的 由于图像检索中存在着低层特征和高层语义之间的“语义鸿沟”,图像自动标注成为当前的关键性问题.为缩减语义鸿沟,提出了一种混合生成式和判别式模型的图像自动标注方法.方法 在生成式学习阶段,采用连续的概率潜在语义分析模型对图像进行建模,可得到相应的模型参数和每幅图像的主题分布.将这个主题分布作为每幅图像的中间表示向量,那么图像自动标注的问题就转化为一个基于多标记学习的分类问题.在判别式学习阶段,使用构造集群分类器链的方法对图像的中间表示向量进行学习,在建立分类器链的同时也集成了标注关键词之间的上下文信息,因而能够取得更高的标注精度和更好的检索效果.结果 在两个基准数据集上进行的实验表明,本文方法在Corel5k数据集上的平均精度、平均召回率分别达到0.28和0.32,在IAPR-TC12数据集上则达到0.29和0.18,其性能优于大多数当前先进的图像自动标注方法.此外,从精度—召回率曲线上看,本文方法也优于几种典型的具有代表性的标注方法.结论 提出了一种基于混合学习策略的图像自动标注方法,集成了生成式模型和判别式模型各自的优点,并在图像语义检索的任务中表现出良好的有效性和鲁棒性.本文方法和技术不仅能应用于图像检索和识别的领域,经过适当的改进之后也能在跨媒体检索和数据挖掘领域发挥重要作用.     

A comprehensive study of age-related macular degeneration detection

Age-related macular degeneration (AMD) is an illness involving the degeneration of the macula of the retina. Fundus photography is the most affordable and convenient way to monitor individuals, in which AMD symptoms segmentation is necessary to assist clinical diagnosis. This study conducted a large number of experimental discussions on the annotation quality and symptoms categories to find a reliable learning strategy, and then applied it to early detection of AMD. Specifically, we discuss the inference of the representational power of the deep neural network, loss function selection, the preprocessing scheme of annotation augmentation, and the annotation quality of the dataset on prediction performance. This paper verified that different learning strategies need to be selected for the AMD symptoms segmentation tasks with varying characteristics of database, which can be used as a reference for developing the related research in the future. On the other hand, we demonstrated that current medical datasets suffer from annotation quality uncertainty, leading to limited learning capabilities. In the future, it is necessary to develop methods to overcome the impact of datasets with poor annotation quality.

     

Relational networks of conditional preferences

Much like relational probabilistic models, the need for relational preference models naturally arises in real-world applications involving multiple, heterogeneous, and richly interconnected objects. On the one hand, relational preferences should be represented into statements which are natural for human users to express. On the other hand, relational preference models should be endowed with a structure that supports tractable forms of reasoning and learning. Based on these criteria, this paper introduces the framework of relational conditional preference networks (RCP-nets), that maintains the spirit of the popular ??CP-nets?? by expressing relational preferences in a natural way using the ceteris paribus semantics. We show that acyclic RCP-nets support tractable inference for optimization and ranking tasks. In addition, we show that in the online learning model, tree-structured RCP-nets (with bipartite orderings) are efficiently learnable from both optimization tasks and ranking tasks, using linear loss functions. Our results are corroborated by experiments on a large-scale movie recommendation dataset.     

A Review of “What They Don’t Teach You in Library School”

SUMMARY

Distance learning students may not think of the “campus” library as the first place to fulfill their information needs and may not even be aware of the services available to them. One way to reach these students is to adopt and adapt marketing techniques from the business world. This article examines the findings of a survey conducted at Emporia State University concerning the awareness of distance learning services. It will also examine marketing techniques and illustrate how they can be applied to increase awareness of reference support services for distance learners.     

The role of abduction in chance discovery

Recently, researches on discovery science and knowledge discovery have been carried out in various fields. Basically they are types of learning that learn tendencies from the sets of data of the same or similar categories. In this sense, discovery is to discover the tendencies. As a result, they cannot predict the events that are different from the trend. On the other hand, abduction is thought of as an explanatory reasoning. Indeed, abduction is a reasoning to generate hypotheses to explain an observation. However, the original meaning of abduction was to discover new things that cannot be known in a simple way. In this paper, abduction is defined using the original definition that discovers something that cannot be easily predicted. Then, this paper shows a role of abduction that can suggest or foresee the events that are different from the trend. In fact, Abductive Analogical Reasoning that can generate new hypotheses is adopted to solve the problem. Akinori Abe, Ph.D.: He obtained his Doctor of Engineering (Ph.D) from the University of Tokyo in 1991, with a thesis entitledA Fast Hypothetical Reasoning System using Analogical Case. His main research interests are abduction (hypothetical reasoning), analogical reasoning, chance dicovery and language sense processing. He is a member of the Planning Committee of the New Generation Computing. He worked in NTT MSC (Malaysia) from 2000 to 2002. Currently, he works in ATR.     

字典重建和空间分布关系约束下的特征选择与图像拼接

目的 针对大型图像检索领域中,复杂图像中SIFT特征描述子的冗余和高维问题,提出了一种基于字典重建和空间分布关系约束的特征选择的方法,来消除冗余特征并保留最具表现力的、保留原始空间结构性的SIFT特征描述子。方法 首先,实验发现了特征选择和字典学习方法在稀疏表示方面的内在联系,将特征选择问题转化为字典重构任务;其次,在SIFT特征选择问题中,为了保证特征空间中特征的鲁棒性,设计了新型的字典学习模型,并采用模拟退火算法进行迭代求解;最后,在字典学习的过程中,加入熵理论来约束特征的空间分布,使学习到的特征描述子能最大限度保持原始SIFT特征空间的空间拓扑关系。结果 在公开数据集Holiday大型场景图片检索数据库上,通过与国际公认的特征选择方法进行实验对比,本文提出的特征选择方法在节省内存空间和提高时间效率（30%~ 50%）的同时,还能保证所筛选的特征描述子的检索准确率比同类特征提高8%~ 14.1%;在国际通用的大型场景图片拼接数据库IPM上,验证本文方法在图像拼接应用中特征提取和特征匹配上的有效性,实验表明本文方法能节省（50% ~70%）图像拼接时间。结论 与已有的方法比较,本文的特征选择方法既不依赖训练数据集,也不丢失重要的空间结构和纹理信息,在大型图像检索、图像拼接领域和3D检索领域中,能够精简特征,提高特征匹配效率和准确率。     

TOWARD LEARNING GRID INFRASTRUCTURES: AN OVERVIEW OF RESEARCH ON GRID LEARNING SERVICES

The Learning Grid refers to the promise of projects that pool together instructional materials on distant computers. The Grid provides a wide range of available and potential learning services and resources and does not simply refer to taking advantage of the multiplying effects of connectivity. It supports the personalized use of the collective intelligence provided by networked computers and supports the exchange, negotiation, and dialogue within and among virtual, evolutionary, and pervasive learning communities. This article provides an overview of papers from the first workshop on Grid Learning Services, which brought together researchers discussing their views of infrastructure, services, and resources. It also addresses several research questions, including: What are the relevant resources and services and how can they be identified or built? How do they rely on the basic open Grid service architecture? How can intelligent tutoring systems be built on the Grid? How do the performance, efficiency, usability, and the global ability of those services meet individual and collective users' expectations?     

融合语义—表观特征的无监督前景分割

目的 前景分割是图像理解领域中的重要任务,在无监督条件下,由于不同图像、不同实例往往具有多变的表达形式,这使得基于固定规则、单一类型特征的方法很难保证稳定的分割性能。针对这一问题,本文提出了一种基于语义-表观特征融合的无监督前景分割方法（semantic apparent feature fusion,SAFF）。方法 基于语义特征能够对前景物体关键区域产生精准的响应,但往往产生的前景分割结果只关注于关键区域,缺乏物体的完整表达;而以显著性、边缘为代表的表观特征则提供了更丰富的细节表达信息,但基于表观规则无法应对不同的实例和图像成像模式。为了融合表观特征和语义特征优势,研究建立了融合语义、表观信息的一元区域特征和二元上下文特征编码的方法,实现了对两种特征表达的全面描述。接着,设计了一种图内自适应参数学习的方法,用于计算最适合的特征权重,并生成前景置信分数图。进一步地,使用分割网络来学习不同实例间前景的共性特征。结果 通过融合语义和表观特征并采用图像间共性语义学习的方法,本文方法在PASCAL VOC（pattern analysis,statistical modelling and computational learning visual object classes）2012训练集和验证集上取得了显著超过类别激活映射（class activation mapping,CAM）和判别性区域特征融合方法（discriminative regional feature integration,DRFI）的前景分割性能,在F测度指标上分别提升了3.5%和3.4%。结论 本文方法可以将任意一种语义特征和表观特征前景计算模块作为基础单元,实现对两种策略的融合优化,取得了更优的前景分割性能。     

Learning second and subsequent programming languages: A problem of transfer

Learning second and subsequent programming languages is easier than learning a first programming language because many concepts and constructs are shared. However, it is still a hard task. In this protocol analysis of moderately experienced programmers transferring to a new programming language, we classified episodes by whether they involved the syntactic, semantic, or planning level of programming knowledge. We discovered that most episodes involve planning and that in solving a given subproblem there are typically many cycles of language‐independent tactical planning followed by language‐dependent implementation planning. On the other hand, programmers have relatively minor problems with the syntax and semantics of a new language. Our subjects’ protocols and their final programs revealed that the plans they develop are strongly influenced by their knowledge of what would be convenient and appropriate in other languages they know. This prevents them from taking full advantage of the capabilities of the new language.     

Adoption of open source software in software-intensive organizations – A systematic literature review

ContextOpen source software (OSS) is changing the way organizations develop, acquire, use, and commercialize software.ObjectiveThis paper seeks to identify how organizations adopt OSS, classify the literature according to these ways of adopting OSS, and with a focus on software development evaluate the research on adoption of OSS in organizations.MethodBased on the systematic literature review method we reviewed publications from 24 journals and seven conference and workshop proceedings, published between 1998 and 2008. From a population of 24,289 papers, we identified 112 papers that provide empirical evidence on how organizations actually adopt OSS.ResultsWe show that adopting OSS involves more than simply using OSS products. We moreover provide a classification framework consisting of six distinctly different ways in which organizations adopt OSS. This framework is used to illustrate some of the opportunities and challenges organizations meet when approaching OSS, to show that OSS can be adopted successfully in different ways, and to organize and review existing research. We find that existing research on OSS adoption does not sufficiently describe the context of the organizations studied, and it fails to benefit fully from related research fields. While existing research covers a large number of topics, it contains very few closely related studies. To aid this situation, we offer directions for future research.ConclusionThe implications of our findings are twofold. On the one hand, practitioners should embrace the many opportunities OSS offers, but consciously evaluate the consequences of adopting it in their own context. They may use our framework and the success stories provided by the literature in their own evaluations. On the other hand, researchers should align their work, and perform more empirical research on topics that are important to organizations. Our framework may be used to position this research and to describe the context of the organization they are studying.     

THE STROBE MODEL: DYNAMIC SERVICE GENERATION ON THE GRID

This article presents the STROBE model: both an agent representation and an agent communication, model based on a social approach, which means interaction centered. This model represents how agents may realize the interactive, dynamic generation of services on the Grid. Dynamically generated services embody a new concept of service implying a collaborative creation of knowledge, i.e., learning; services are constructed interactively between agents depending on a conversation. The approach consists of integrating selected features from multi-agent systems and agent communication, language interpretation in applicative/functional programming and e-learning/human-learning into a unique, original, and simple view that privileges interactions, including control. The main characteristic of STROBE agents is that they develop a language (environment + interpreter) for each of their interlocutors. The model is inscribed within a global approach, defending a shift from the classical algorithmic (control based) view to problem solving in computing to an interaction-based view of social informatics, where artificial as well as human agents operate by communicating as well as by computing. The paper shows how the model may not only account for the classical communicating agent approaches, but also represent a fundamental advance in modeling societies of agents in particular in dynamic service generation scenarios such as those necessary today on the Web and proposed tomorrow for the Grid. Preliminary concrete experimentations illustrate the potential of the model; they are significant examples for a very wide class of computational and learning situations.     

Personalization for unobtrusive service interaction

Increasingly, mobile devices play a key role in the communication between users and the services embedded in their environment. With ever greater number of services added to our surroundings, there is a need to personalize services according to the user needs and environmental context avoiding service behavior from becoming overwhelming. In order to prevent this information overload, we present a method for the development of mobile services that can be personalized in terms of obtrusiveness (the degree in which each service intrudes the user’s mind) according to the user needs and preferences. That is, services can be developed to provide their functionality at different obtrusiveness levels depending on the user by minimizing the duplication of efforts. On the one hand, we provide mechanisms for describing the obtrusiveness degree required for a service. On the other hand, we make use of Feature Modeling techniques in order to define the obtrusiveness level adaptation in a declarative manner. An experiment was conducted in order to put in practice the proposal and evaluate the user acceptance for the personalization capabilities provided by our approach.     

MASHED: Security and privacy-aware mutual authentication scheme for heterogeneous and distributed mobile cloud computing services

ABSTRACT

Rapid development in mobile devices and cloud computing technologies has increased the number of mobile services from different vendors on the cloud platform. However, users of these services are facing different security and access control challenges due to the nonexistence of security solutions capable of providing secure access to these services, which are from different vendors, using a single key. An effective security solution for heterogeneous Mobile Cloud Computing (MCC) services should be able to guarantee confidentiality and integrity through single key-based authentication scheme. Meanwhile, a few of the existing authentication schemes for MCC services require different keys to access different services from different vendors on a cloud platform, thus increases complexity and overhead incurred through generation and storage of different keys for different services.

In this paper, an efficient mutual authentication scheme for accessing heterogeneous MCC services is proposed. The proposed scheme combines the user’s voice signature with cryptography operations to evolve efficient mutual authentication scheme devoid of key escrow problem and allows authorized users to use single key to access the heterogeneous MCC services at a reduced cost.