The Erasmus-funded Lived experience of climate change: e-learning and virtual mobility (LECH-e) project is developing learning resources on the topic of climate change that will be openly accessible on the web for higher education institutions and the wider world public to use and adapt. Institutions are free to integrate the resources into their formal programs and non-formal courses. A description of the project is provided in a short prologue. In this paper we depict the open learning strategy implemented by the LECH-e partnership and how it can contribute to widening participation in environmental, and specifically climate change, education.

The rapid technological developments and its inherent changes cause the technologies to work as timeline paths, allowing a new generation to arise every ten years. The emergence of these new generations in a limited period of time has caused a revolution in society, because unlike what happened in the past, now there are several generations occupying the same work, schooling and leisure spaces. These generations are very different and that is why there is a kind of revolution in its own dynamic, because some of those crystallized habits and behaviors suffered, over time, recursive changes in very short periods, requiring individuals to adapt to a variety of people with very distinct tastes and habits and, more importantly, with different mental models [1]. In this article we will focus our attention identifying the characteristics of generations X, Y and Z, exposing succinctly their desires, aspirations and its relations with the world, as well as their relationship with school and, more specifically, with learning. As a corollary of this relationship, some strategies will be presented that we believe to be of great use in the teaching of generation Y and Z.; Universidade do Minho. Centro de Investigação em Educação (CIED)

We consider learning algorithms induced by regularization methods in the regression setting. We show that previously obtained error bounds for these algorithms using a-priori choices of the regularization parameter, can be attained using a suitable a-posteriori choice based on validation. In particular, these results prove adaptation of the rate of convergence of the estimators to the minimax rate induced by the "effective dimension" of the problem. We also show universal consistency for theses class methods.

In this paper we present a class of algorithms for similarity learning on spaces of images. The general framework that we introduce is motivated by some well-known hierarchical pre-processing architectures for object recognition which have been developed during the last decade, and which have been in some cases inspired by functional models of the ventral stream of the visual cortex. These architectures are characterized by the construction of a hierarchy of “local” feature representations of the visual stimulus. We show that our framework includes some well-known techniques, and that it is suitable for the analysis of dynamic visual stimuli, presenting a quantitative error analysis in this setting.

Within the framework of statistical learning theory we analyze in detail the so-called elastic-net regularization scheme proposed by Zou and Hastie ["Regularization and variable selection via the elastic net" J. R. Stat. Soc. Ser. B, 67(2):301-320, 2005] for the selection of groups of correlated variables. To investigate on the statistical properties of this scheme and in particular on its consistency properties, we set up a suitable mathematical framework. Our setting is random-design regression where we allow the response variable to be vector-valued and we consider prediction functions which are linear combination of elements (features) in an infinite-dimensional dictionary. Under the assumption that the regression function admits a sparse representation on the dictionary, we prove that there exists a particular "elastic-net representation" of the regression function such that, if the number of data increases, the elastic-net estimator is consistent not only for prediction but also for variable/feature selection. Our results include finite-sample bounds and an adaptive scheme to select the regularization parameter. Moreover, using convex analysis tools, we derive an iterative thresholding algorithm for computing the elastic-net solution which is different from the optimization procedure originally proposed in "Regularization and variable selection via the elastic net".

A large number of learning algorithms, for example, spectral clustering, kernel Principal Components Analysis and many manifold methods are based on estimating eigenvalues and eigenfunctions of operators defined by a similarity function or a kernel, given empirical data. Thus for the analysis of algorithms, it is an important problem to be able to assess the quality of such approximations. The contribution of our paper is two-fold: 1. We use a technique based on a concentration inequality for Hilbert spaces to provide new much simplified proofs for a number of results in spectral approximation. 2. Using these methods we provide several new results for estimating spectral properties of the graph Laplacian operator extending and strengthening results from [26].

peer-reviewed; Background: The benefits of Peer Assisted Learning (PAL) are well established with positive effects on examination scores, student satisfaction and personal and professional development reported. PAL is increasingly utilised as a resource within medical education where the restrictions on resources have forced teachers to look at creating new educational environments which can be delivered at a lower cost. This study sought to evaluate the processes at work as the emphasis of PAL research to date has largely been on the consideration of student outcomes. Methods: Fifth-year medical undergraduates, who had completed their communication skills modular training and attended a preparatory workshop, facilitated a role-play session for their second-year colleagues within an Early Patient Contact programme. Semi-structured interviews and focus groups were used to collect data at different time points in order to establish the views of peer learners and tutors towards this new method of teaching. The data was analysed according to the principles framework analysis using N-vivo software. Themes were shared and debated with the multidisciplinary team of authors and a concordance of views on common themes was reached after discussion and debate. Results: Analysis of the data resulted in the emergence of three thematic categories: Learning Environment...

peer-reviewed; In the late nineteenth century Christopher Columbus Langdell, Dean of Harvard Law School, introduced a new pedagogy in law that was designed around Socratic teaching. Prior to this, most common law jurists had emphasised the importance of artificial, natural law reasoning which was prone to be more subjective, speculative, value laden, and ultimately illogical (Davies, 1994, p. 110). This new pedagogy placed law cases at the centre of students’ learning, and demanded much more from the students in terms of analysis and defence of legal explanations. In particular, it involved a case-dialogue method, where students are called upon to recount facts, argue legal principles and explain their reasoning in the lecture hall before an authoritarian lecturer. All of this was designed to mirror the combative realities of adversarial proceedings. So, for example, in a lecture, the lecturer might demand of a student the facts of a case, the legal points at issue, the court’s reasoning by reference to other cases, the underlying legal doctrine or principle, and the effect that an altered fact pattern might have on the outcome.; PUBLISHED; Not peer reviewed

The problem we consider in this paper is reinforcement learning with value advice. In this setting, the agent is given limited access to an oracle that can tell it the expected return (value) of any state-action pair with respect to the optimal policy. The agent must use this value to learn an explicit policy that performs well in the environment. We provide an algorithm called RLAdvice, based on the imitation learning algorithm DAgger. We illustrate the effectiveness of this method in the Arcade Learning Environment on three different games, using value estimates from UCT as advice.

We extend the Q-learning algorithm from the Markov Decision Process setting to problems where observations are non-Markov and do not reveal the full state of the world i.e. to POMDPs. We do this in a natural manner by adding l0 regularisation to the pathwise squared Q-learning objective function and then optimise this over both a choice of map from history to states and the resulting MDP parameters. The optimisation procedure involves a stochastic search over the map class nested with classical Q-learning of the parameters. This algorithm fits perfectly into the feature reinforcement learning framework, which chooses maps based on a cost criteria. The cost criterion used so far for feature reinforcement learning has been model-based and aimed at predicting future states and rewards. Instead we directly predict the return, which is what is needed for choosing optimal actions. Our Q-learning criteria also lends itself immediately to a function approximation setting where features are chosen based on the history. This algorithm is somewhat similar to the recent line of work on lasso temporal difference learning which aims at finding a small feature set with which one can perform policy evaluation. The distinction is that we aim directly for learning the Q-function of the optimal policy and we use l0 instead of l1 regularisation. We perform an experimental evaluation on classical benchmark domains and find improvement in convergence speed as well as in economy of the state representation. We also compare against MC-AIXI on the large Pocman domain and achieve competitive performance in average reward. We use less than half the CPU time and 36 times less memory. Overall...

We present a new algorithm for general reinforcement learning where the true environment is known to belong to a finite class of N arbitrary models. The algorithm is shown to be near-optimal for all but O(N log2 N) timesteps with high probability. Infinite classes are also considered where we show that compactness is a key criterion for determining the existence of uniform sample-complexity bounds. A matching lower bound is given for the finite case.

Problem‐based learning (PBL) has been identified in the literature as one approach for enabling the development of generic employability skills. This approach differs from many traditional approaches to teaching and learning by integrating employability skills (such as problem solving, teamwork and self‐management) into the student learning process. The purpose of this paper is to explore problem‐based learning (PBL) as a vehicle for improving the quality of teaching, learning and assessment outcomes in the management field. Examples from a Masters course are provided as initial empirical data. The paper argues that PBL assists in the development of employability skills for students, supports a high level of interaction among students from diverse cultural backgrounds and facilitates the integration of multiple perspectives into group views as part of the problem‐solving process (Checkland, 1981).; http://www.eac2007.qut.edu.au/; Carmen Joham, Marilyn Clarke; Copyright © 2007 Joham and Clarke.

Learning leverage is characterized by three interactive dynamics: rigor, reward, and risk. Together, they create uncomfortable yet positive pressure that usually results in teacher learning. The model of learning leverage helps explain why most teachers consider National Board candidacy such a powerful learning experience and provides insight as to why some teachers learn much and others learn little through the experience. The model is applicable to learning experiences other than National Board candidacy, such as teacher professional development, mentoring and induction, and PK-16 classroom instruction.

Abstract: This paper examines the practice of participatory evaluation through an exploratory single case study of the Evaluation Team of Books & Beyond, a cocurricular service-learning program of the Global Village Living-Learning Center at Indiana University. The paper, which is authored by three undergraduate members of the evaluation team and their faculty advisor, juxtaposes the process of conducting the evaluation and reporting the results with reflections from the Evaluation Team participants on conducting youth participatory action research, which offers a means of improving youth-serving programs and developing a greater understanding of why youth choose to participate in these programs. In their review of the implementation of their evaluation project, the team noted that the difficulties of getting past lessons learned to methodological rigor in service-learning evaluation are compounded by the realities of engaging in a student-faculty partnership in a co-curricular service-learning context.

An introduction to the Queen's Learning Commons; Produced for the Queen's Learning Commons by Queen's University Video Services. 4:27 minutes

Measuring the quality of a b-learning environment is critical to determine the success of a b-learning course. Several initiatives have been recently conducted on benchmarking and quality in e-learning. Despite these efforts in defining and examining quality issues concerning online courses, a defining instrument to evaluate quality is one of the key challenges for blended learning, since it incorporates both traditional and online instruction methods. For this paper, six frameworks for quality assessment of technological enhanced learning were examined and compared regarding similarities and differences. These frameworks aim at the same global objective: the quality of e-learning environment/products. They present different perspectives but also many common issues. Some of them are more specific and related to the course and other are more global and related to institutional aspects. In this work we collected and arrange all the quality criteria identified in order to get a more complete framework and determine if it fits our b-learning environment. We also included elements related to our own b-learning research and experience, acquired during more than 10 years of experience. As a result we have create a new quality reference with a set of dimensions and criteria that should be taken into account when you are analyzing...

We compare the practical performance of several recently proposed algorithms for active learning in the online setting. We consider two algorithms (and their combined variants) that are strongly online, in that they do not store any previously labeled examples, and for which formal guarantees have recently been proven under various assumptions. We perform an empirical evaluation on optical character recognition (OCR) data, an application that we argue to be appropriately served by online active learning. We compare the performance between the algorithm variants and show significant reductions in label-complexity over random sampling.

Bloom’s Taxonomy of cognitive domains is a well-defined and broadly accepted tool for categorizing types of thinking into six different levels: knowledge, comprehension, application, analysis, synthesis and evaluation. It is commonly accepted that memorization and recall (knowledge-level skills) are lower order cognitive skills that require only a minimum level of understanding, whereas the application of knowledge and critical thinking (application, analysis, synthesis & evaluation-level skills) are higher order cognitive skills that require deep conceptual understanding. In our Biology department we have begun introducing students to Bloom’s taxonomy during the introductory series to help students recognize the different levels of thinking they will need to master to succeed in the curriculum. First, we indicate on our exams the Bloom’s level required to successfully answer each question (e.g. knowledge or application) and then provide students with the overall “Bloom’s distribution” for the exam i.e. what percentage of the questions require lower-order versus higher-order cognitive skills. To aid students in identifying the levels that are most challenging for them, we are piloting a program that provides students with their individual “Bloom’s score” after each exam. The Bloom’s score indicates how well they performed on questions requiring different levels of Bloom’s. Typically students receive Bloom’s scores that are high for knowledge and comprehension and lower for one or more higher-order cognitive skills. Once these challenging areas have been identified...

peer-reviewed; Over the past decade or so the open education movement has continued to gather momentum in higher education, spurred on by increasing demand for more flexible education options; by the potential of developments in technology and infrastructure; by advocacy at policy level; and by initiatives and developments at national and international levels. Open educational resources (OER1), one element of the open education movement, have seen exponential growth in this period. Navigating this OER landscape poses a number of important issues and questions for the practice of teaching and learning. From an educational development perspective, the focus rests on investigating how both students and teachers can use and share open educational resources in ways that optimally enhance teaching and learning.

El Observatorio Gallego de e-learning ha publicado un exhaustivo informe sobre los resultados del e-learning en la comunidad autónoma de Galicia en el ámbito universitario y empresarial. En cuanto al e-learning corporativo, se proporcionan datos del e-learning de la empresa consumidora, la empresa no consumidora y de la empresa ofertante. En cuanto al estado del e-learning en la universidad, se analiza el e-learning desde el punto de vista de los órganos directivos, los administradores de los campus virtuales, los grupos de investigación y el personal docente e investigador.