Browsing by Subject "online learning"

Now showing 1 - 11 of 11

A Collaborative Learning Process from a Distance
(2023) Jackson, Jeff; Reichenbach, Michael
An online collaborative learning process was designed to bridge the gap between research and practice, engaging researchers in what is meaningful to a community of interest. The process is based on a collaborative learning model that includes a project focus, participant experience, expert knowledge, dialogue, deliberation, and reflection. Participants helped us identify gaps in education, research, and policy, which then resulted in a plan for additional research projects and educational programming. The process is described so others can create similar opportunities connecting researcher and stakeholder viewpoints.
Convex Optimization and Online Learning: Their Applications in Discrete Choice Modeling and Pricing
(2018-05) Li, Xiaobo
The discrete choice model has been an important tool to model customers' demand when facing a set of substitutable choices. The random utility model, which is the most commonly used discrete choice framework, assumes that the utility of each alternative is random and follows a prescribed distribution. Due to the popularity of the random utility model, the probabilistic approach has been the major method to construct and analyze choice models. In recent years, several choice frameworks that are based on convex optimization are studied. Among them, the most widely used frameworks are the representative agent model and the semi-parametric choice model. In this dissertation, we first study a special class of the semi-parametric choice model - the cross moment model (CMM) - and reformulate it as a representative agent model. We also propose an efficient algorithm to calculate the choice probabilities in the CMM model. Then, motivated by the reformulation of the CMM model, we propose a new choice framework - the welfare-based choice model - and establish the equivalence between this framework and the other two choice frameworks: the representative agent model and the semi-parametric choice model. Lastly, motivated by the multi-product pricing problem, which is an important application of discrete choice models, we develop an online learning framework where the learning problem shares some similarities with the multi-product pricing problem. We propose efficient online learning algorithms and establish convergence rate results for these algorithms. The main techniques underlying our studies are continuous optimization and convex analysis.
Essays on Design, Operation, and Pricing of On-Demand Service Systems
(2024-04) Shen, Xiaobing
In this dissertation, we describe research on the design, operation, and pricing of on-demand service systems. We use the term "on-demand service systems" to refer to systems distinguished by two important features: (1) an "on-demand" feature, with requests arising randomly over time without prior booking or notice and (2) a "resource reusability" feature, with the fulfillment of demand involving the usage of resources for a limited (and random) amount of time. We model the systems as closed queueing systems and provide non-asymptotic performance guarantees.
I CARE 2.0: Facilitating Online Learning in Higher Education
(Association for the Advancement of Computing in Education (AACE), 2014) Kennedy, Jolie
Hoffman & Ritchie’s (1998) I CARE model helps facilitators think about the way they design their online courses and interact with learners. The five-step model affords organization to a dynamic online environment, which is helpful for novice and seasoned online instructors and learners. The I CARE model is customized with constructivist approaches and integrated with contemporary social learning technologies in a weekly class agenda format. The instructor (I) introduces the content with a video and sets the context for the lesson. Learners (C) connect to their prior knowledge and (A) apply new knowledge through personalized active learning and authentic assessments. Students (R) reflect on the lesson topic. Learning is (E) extended through an evaluation of the experience or through a guided micro-blog activity. Theories, examples, and options for customization are discussed.
Nontraditional Student Participation in Asynchronous Online Discussions
(2017-05) Hachey, Valera
Success in higher education is a concept that has been researched for years and is especially critical in relation to the shift to online higher education. Online learning is inevitably a part of the future landscape of higher education, but success rates in online courses are often lower than in traditional courses. To contribute to the existing literature, this research explores the nature of participation in asynchronous online discussions of nontraditional students in online courses. The research has an overarching constructivist framework in order to maintain the focus on the social nature of learning, in addition to framing it with the theory of capital and the principles of andragogy. The methodology used is quantitative, including ANOVA, linear regression, and chi squares, to analyze differences across course levels and post types. The categories used are based on an established framework for content analysis. Differences in types of presence were found across course level, predictive relationships were found among the types of presence, and differences in the more detailed categories of types of presence were found across course level and post type. Such findings point to the importance of discussion prompts and teaching behaviors within the curriculum in online courses that will best serve nontraditional students.
Online Convex Optimization in Changing Environments and its Application to Resource Allocation
(2019-12) Yuan, Jianjun
In the era of the big data, we create and collect lots of data from all different kinds of sources: the Internet, the sensors, the consumer market, and so on. Many of the data are coming sequentially, and would like to be processed and understood quickly. One classic way of analyzing data is based on batch processing, in which the data is stored and analyzed in an offline fashion. However, when the volume of the data is too large, it is much more difficult and time-consuming to do batch processing than sequential processing. What’s more, sequential data is usually changing dynamically, and needs to be understood on-the-fly in order to capture the changes. Online Convex Optimization (OCO) is a popular framework that matches the above sequential data processing requirement. Applications using OCO include online routing, online auctions, online classification and regression, as well as online resource allocation. Due to the general applicability of OCO to the sequential data and the rigorous theoretical guarantee, it has attracted lots of researchers to develop useful algorithms to fulfill different needs. In this thesis, we show our contributions to OCO’s development by designing algorithms to adapt to changing environments. In the first part of the thesis, we propose algorithms to have better adaptivity by examining the notion of dynamic regret, which compares the algorithm’s cumulative loss against that incurred by a comparison sequence. Dynamic regret extends a common performance measure known as static regret. Since it may not be known whether the environment is dynamic or not, it is desirable to take advantage of both regrets by having a trade-off between them. To achieve that, we discuss recursive least-squares algorithms and show how forgetting factors can be used to develop new OCO algorithms that have such a regret trade-off. More specifically, we rigorously characterize the effect of forgetting factors for a class of online Newton algorithms. For exp-concave or strongly convex objective, the improved dynamic regret of max{O(log T),O(\sqrt{TV })} is achieved, where V is a bound on the path length of the comparison sequence. In particular, we show how classic recursive least-squares with a forgetting factor achieves this dynamic regret bound. By varying V , we obtain the regret trade-off. In order to obtain more computationally efficient algorithm, we also propose a novel gradient descent step size rule for strongly convex functions, which recovers the dynamic regret bounds described above. For smooth problems, we can obtain static regret of O(T^{1−\beta}) and dynamic regret of O(T^{\beta}V^*), where \beta\in(0,1) and V^* is the path length of the sequence of minimizers. By varying \beta, we obtain the regret trade-off. The second part of the thesis describes how to design efficient algorithms to adapt to the changing environments. Previous literature runs a pool of algorithms in parallel to gain better adaptivity, which increases both the running time and the online implementation complexity. Instead, we propose a new algorithm requiring only one update per time step, while with the same adaptive regret performance guarantee as the current state-of-the-art result. We then apply the algorithm to online Principal Component Analysis (online PCA) and variance minimization under changing environments, since the previous literature on online PCA has focused on performance guarantee under stationary environment. We demonstrate both theoretically and experimentally that the proposed algorithms can adapt to the changing environments. The third part of the thesis starts from the observation that the projection operator used in constrained OCO algorithms cannot really achieve true online implementation due to the high time-consumption. To accelerate the OCO algorithms’ update, previous literature is proposed to approximate the true desired projection with a simpler closed-form one at the cost of constraint violation (g(\theta) >0) for some time steps. Nevertheless, it can guarantee sub-linearity for both the static regret and the long-term constraint, \sum_{t=1}^T g(\theta_t), having constraint satisfaction on average. However, the sub-linear long-term constraint does not enforce small constraint violation for every time step, because a strictly feasible solution can cancel out the effects of violated constraints. To resolve it, we propose algorithms to have the cumulative constraint of the form \sum_{t=1}^T(max{g(\theta_t), 0})^2 upper bounded sub-linearly. This new form heavily penalizes large constraint violations while the cancellation effects cannot occur. Furthermore, useful bounds on the single step constraint violation are derived. For convex objectives, our result generalizes existing bounds, and for strongly convex objectives we give improved regret bounds. In numerical experiments, we show that our algorithm closely follows the constraint boundary leading to low cumulative violation. Furthermore, we extend the proposed algorithms’ idea to the more general time-dependent online resource allocation problems with performance guarantee by a variant of dynamic regret.
Renewing the Investigation into Social Presence: The Impact of Psychological Closeness and Technology Modality on Satisfaction, Future Persistence and Final Grade
(2016-01) Norden, Amie
In 1975, Weinberg (2001) first posed the challenging question that we, as educators, are unfortunately still struggling to answer: Why are we so unable to anticipate the second order effects of the first order victories of science and technology? Forty years later, education is still struggling to identify and address the second order effects of the technological changes that exploded around 1995 with the advent of the worldwide web. Faculty still struggle with rapid technological developments. Students desire greater flexibility with online learning and seek learning that embeds the technological formats they use in their day-to-day lives. Institutions of higher education grapple to meet the demand for more online courses, as well as to resolve the challenges that online learning poses at the institutional level. The field of education is still wondering where online learning fits with more traditional pedagogical designs. One aspect of online learning that has come into the recent limelight is the topic of social presence. Currently, the term social presence is thrown around as a panacea for a variety of online learning design problems. Unfortunately, the topic of social presence is fraught with ambiguity and controversy. Educational researchers define and measure social presence in a multitude of ways, which makes its application all the more problematic. This debate is happening at the same time as instructors and instructional designers work to implement strategies to increase social presence in online courses, seeking to use it as way to address the challenges that online learning brings to current learning environments. This study examines the topic of social presence by harkening back to its original conception: a consilience of psychological closeness and technology modality. The study begins with a literature review, including aspects of immediacy and use of technology in online learning, and ends with a call to the field of education in creating online course design in the years ahead.
The Role of the Technical Communicator in the Corporate eLearning Industry
(2020-05-05) Arnquist, Marissa, D
Technical communicators bring myriad skills to professional roles, including textual and visual content creation, content organization, content strategy, technology proficiency, and user analysis. These skills are well-matched for roles in elearning, strategizing and creating instructional content for online consumption. Through literature review, informational interviews, and job posts analysis, the correlation between a technical communicator’s skills and the skills needed to be successful in elearning becomes clear. This research elucidates the opportunities for technical communicators in the corporate elearning industry and offers an introductory guide into exploring a career in the industry.
Scalable Learning Adaptive to Unknown Dynamics and Graphs
(2019-06) Shen, Yanning
With the scale of information growing every day, the key challenges in machine learning include the high-dimensionality and sheer volume of feature vectors that may consist of real and categorical data, as well as the speed and the typically streaming format of data acquisition that may also entail outliers and misses. The latter may be present, either unintentionally or intentionally, in order to cope with scalability, privacy, and adversarial behavior. These challenges provide ample opportunities for algorithmic and analytical innovations in online and nonlinear subspace learning approaches. Among the available nonlinear learning tools, those based on kernels have merits that are well documented. However, most rely on a preselected kernel, whose prudent choice presumes task-specific prior information that is generally not available. It is also known that kernel-based methods do not scale well with the size or dimensionality of the data at hand. Besides data science, the urgent need for scalable tools is a core issue also in network science that has recently emerged as a means of collectively understanding the behavior of complex interconnected entities. The rich spectrum of application domains comprises communication, social, financial, gene-regulatory, brain, and power networks, to name a few. Prominent tasks in all network science applications are those of topology identification and inference of nodal processes evolving over graphs. Most contemporary graph-driven inference approaches rely on linear and static models that are simple and tractable, but also presume that the nodal processes are directly observable. To cope with these challenges, the present thesis first introduces a novel online categorical subspace learning approach to track the latent structure of categorical data `on the fly.' Leveraging the random feature approximation, it then develops an adaptive online multi-kernel learning approach (termed AdaRaker), which accounts not only for data-driven learning of the kernel combination, but also for the unknown dynamics. Performance analysis is provided in terms of both static and dynamic regrets to quantify the novel learning function approximation. In addition, the thesis introduces a kernel-based topology identification approach that can even account for nonlinear dependencies among nodes and across time. To cope with nodal processes that may not be directly observable in certain applications, tensor-based algorithms that leverage piecewise stationary statistics of nodal processes are developed, and pertinent identifiability conditions are established. To facilitate real-time operation and inference of time-varying networks, an adaptive tensor decomposition based scheme is put forth to track the topologies of time-varying networks. Last but not least, the present thesis offers a unifying framework to deal with various learning tasks over possibly dynamic networks. These tasks include dimensionality reduction, classification, and clustering. Tests on both synthetic and real datasets from the aforementioned application domains are carried out to showcase the effectiveness of the novel algorithms throughout.
Seven Traits of Personal Learning Environments for Designing Quality Online Learning Programs: A Systems View of Connectedness
(IGI Global, 2019) Kennedy, Jolie
This chapter reports on research findings that illustrate a system view of connectedness across personal, professional, and academic contexts with implications for designing quality online learning programs. Connected learners organically blur the line between formal and informal learning when they call on their social networks and engage in online learning systems towards goals in their personal, professional, and academic lives. The phenomenological study referenced in this chapter is framed by complexity theory and grounded in research on complex adaptive systems applied to educational contexts. Examples of lived experiences illustrate how being connected in a personal learning environment is experienced as immersion in a complex adaptive system. Implications and recommendations for quality online learning programs are discussed.
Usability Testing
(Neal-Schuman Publishers, 2003) Veldof, Jerilyn
Design a library e-learning interface that gets out of the way of your learners; ensure that the learner's full attention is on the learning, not on navigating the interface; learn how to conduct usability testing that works.

University Digital Conservancy

Browse by Subject

Browsing by Subject "online learning"