Learning Bayesian Networks with Thousands of Variables

We present a method for learning Bayesian networks from data sets containing thousands of variables without the need for structure constraints. Our approach is made of two parts. The first is a novel algorithm that effectively explores the space of possible parent sets of a node. It guides the exploration towards the most promising parent sets on the basis of an approximated score function that is computed in constant time. The second part is an improvement of an existing ordering-based algorithm for structure optimization. The new algorithm provably achieves a higher score compared to its original formulation. Our novel approach consistently outperforms the state of the art on very large data sets.

Implications of aligning full registration of doctors with medical school graduation:a qualitative study of stakeholder perspectives

OBJECTIVES: The Shape of Training report recommended that full registration is aligned with medical school graduation. As part of a General Medical Council-funded study about the preparedness for practice of UK medical graduates, we explored UK stakeholders' views about this proposal using qualitative interviews (30 group and 87 individual interviews) and Framework Analysis.

SETTING: Four UK study sites, one in each country.Save

PARTICIPANTS: 185 individuals from eight stakeholder groups: (1) foundation year 1 (F1) doctors (n=34); (2) fully registered trainee doctors (n=33); (3) clinical educators (n=32); (4) undergraduate/postgraduate Deans, and Foundation Programme Directors (n=30); (5) other healthcare professionals (n=13); (6) employers (n=7); (7) policy and government (n=11); (8) patient and public representatives (n=25).

RESULTS: We identified four main themes: (1) The F1 year as a safety net: patients were protected by close trainee supervision and 'sign off' to prevent errors; trainees were provided with a safe environment for learning on the job; (2) Implications for undergraduate medical education: if the proposal was accepted, a 'radical review' of undergraduate curricula would be needed; undergraduate education might need to be longer; (3) Implications for F1 work practice: steps to protect healthcare team integration and ensure that F1 doctors stay within competency limits would be required; (4) Financial, structural and political implications: there would be cost implications for trainees; clarification of responsibilities between undergraduate and postgraduate medical education would be needed. Typically, each theme comprised arguments for and against the proposal.

CONCLUSIONS: A policy change to align the timing of full registration with graduation would require considerable planning and preliminary work. These findings will inform policymakers' decision-making. Regardless of the decision, medical students should take on greater responsibility for patient care as undergraduates, assessment methods in clinical practice and professionalism domains need development, and good practice in postgraduate supervision and support must be shared.

Improving E-Assessment Systems With A focus On Adaptive Feedback And Marking

Institutions involved in the provision of tertiary education across Europe are feeling the pinch. European universities, and other higher education (HE) institutions, must operate in a climate where the pressure of government spending cuts (Garben, 2012) is in stark juxtaposition to the EU’s strategy to drive forward and maintain a growth of student numbers in the sector (eurostat, 2015).

In order to remain competitive, universities and HE institutions are making ever-greater use of electronic assessment (E-Assessment) systems (Chatzigavriil et all, 2015; Ferrell, 2012). These systems are attractive primarily because they offer a cost-effect and scalable approach for assessment. In addition to scalability, they also offer reliability, consistency and impartiality; furthermore, from the perspective of a student they are most popular because they can offer instant feedback (Walet, 2012).

There are disadvantages, though.

First, feedback is often returned to a student immediately on competition of their assessment. While it is possible to disable the instant feedback option (this is often the case during an end of semester exam period when assessment scores must be can be ratified before release), however, this option tends to be a global ‘all on’ or ‘all off’ configuration option which is controlled centrally rather than configurable on a per-assessment basis.

If a formative in-term assessment is to be taken by multiple groups of
students, each at different times, this restriction means that answers to each question will be disclosed to the first group of students undertaking the assessment. As soon as the answers are released “into the wild” the academic integrity of the assessment is lost for subsequent student groups.

Second, the style of feedback provided to a student for each question is often limited to a simple ‘correct’ or ‘incorrect’ indicator. While this type of feedback has its place, it often does not provide a student with enough insight to improve their understanding of a topic that they did not answer correctly.

Most E-Assessment systems boast a wide range of question types including Multiple Choice, Multiple Response, Free Text Entry/Text Matching and Numerical questions. The design of these types of questions is often quite restrictive and formulaic, which has a knock-on effect on the quality of feedback that can be provided in each case.

Multiple Choice Questions (MCQs) are most prevalent as they are the most prescriptive and therefore most the straightforward to mark consistently. They are also the most amenable question types, which allow easy provision of meaningful, relevant feedback to each possible outcome chosen.
Text matching questions tend to be more problematic due to their free text entry nature. Common misspellings or case-sensitivity errors can often be accounted for by the software but they are by no means fool proof, as it is very difficult to predict in advance the range of possible variations on an answer that would be considered worthy of marks by a manual marker of a paper based equivalent of the same question.

Numerical questions are similarly restricted. An answer can be checked for accuracy or whether it is within a certain range of the correct answer, but unless it is a special purpose-built mathematical E-Assessment system the system is unlikely to have computational capability and so cannot, for example, account for “method marks” which are commonly awarded in paper-based marking.

From a pedagogical perspective, the importance of providing useful formative feedback to students at a point in their learning when they can benefit from the feedback and put it to use must not be understated (Grieve et all, 2015; Ferrell, 2012).

In this work, we propose a number of software-based solutions, which will overcome the limitations and inflexibilities of existing E-Assessment systems.

Dynamic State Estimation of Power Systems with Quantization Eects: A Recursive Filter Approach

In this paper, a recursive filter algorithm is developed to deal with the state estimation problem for power systems with quantized nonlinear measurements. The measurements from both the remote terminal units and the phasor measurement unit are subject to quantizations described by a logarithmic quantizer. Attention is focused on the design of a recursive filter such that, in the simultaneous presence of nonlinear measurements and quantization effects, an upper bound for the estimation error covariance is guaranteed and subsequently minimized. Instead of using the traditional approximation methods in nonlinear estimation that simply ignore the linearization errors, we treat both the linearization and quantization errors as norm-bounded uncertainties in the algorithm development so as to improve the performance of the estimator. For the power system with such kind of introduced uncertainties, a filter is designed in the framework of robust recursive estimation, and the developed filter algorithm is tested on the IEEE benchmark power system to demonstrate its effectiveness.

Learning Bayesian Networks with Bounded Tree-Width via Guided Search

Bounding the tree-width of a Bayesian network can reduce the chance of overfitting, and allows exact inference to be performed efficiently. Several existing algorithms tackle the problem of learning bounded tree-width Bayesian networks by learning from k-trees as super-structures, but they do not scale to large domains and/or large tree-width. We propose a guided search algorithm to find k-trees with maximum Informative scores, which is a measure of quality for the k-tree in yielding good Bayesian networks. The algorithm achieves close to optimal performance compared to exact solutions in small domains, and can discover better networks than existing approximate methods can in large domains. It also provides an optimal elimination order of variables that guarantees small complexity for later runs of exact inference. Comparisons with well-known approaches in terms of learning and inference accuracy illustrate its capabilities.

Applying the motivation-hygiene theory as a means of measuring learner satisfaction with blended learning courses in higher education

This study describes research on a postgraduate blended learning programme within the Department of Education at the University of Aveiro in Portugal. It is based on a multi-philosophical paradigm and examines students‟ satisfaction levels through the application of Herzberg‟s Motivation and Hygiene Theory. The main question being addressed in this research is: “Can the Motivation and Hygiene Theory be adopted as a means to measure student satisfaction with their blended learning environment?” Embedded within this research question are four fundamental questions which set the scene for the development of this research study and are explored in greater detail in Chapters 4 and 5 respectively: 1. What are the factors responsible for bringing about learning satisfaction with their b-Learning course? 2. What are the factors responsible for bringing about learning dissatisfaction with their b-Learning course? 3. Can these factors be represented as Motivation and Hygiene factors? 4. Will this method of measuring learning satisfaction lead to a set of guidelines that could be considered as a framework for the development of b-Learning courses? The results indicate that the Motivation and Hygiene Theory or an adapted version such as the Enricher and Enabler Theory proposed in this study could be considered as a plausible means of analysing an institution‟s b-Learning processes. The opportunity to carry out future research is evident and can be varied depending on the research objectives in mind. Examples where further exploration would be beneficial lay within the application of this theory to the wider sector; the use of larger samples, focusing on the teachers, as well as the learners and the application of Web 2.0 technologies as means of gathering information. The results of this research will be of great significance to those areas of education that are interested in locating quick and efficient means by which to evaluate their b-Learning and to no lesser extent e-Learning environments.

A Designer’s Approach: Exploring how Autistic Adults with Additional Learning Disabilities Experience their Home Environment

Autistic adults with limited speech and additional learning disabilities are people whose perceptions and interactions with their environment are unique, but whose experiences are under-explored in design research. This PhD by Practice investigates how people with autism experience their home environment through a collaboration with the autism charity Kingwood Trust, which gave the designer extensive access to a community of autistic adults that it supports. The PhD reflects upon a neurotypical designer’s approach to working with autistic adults to investigate their relationship with the environment. It identifies and develops collaborative design tools for autistic adults, their support staff and family members to be involved. The PhD presents three design studies that explore a person’s interaction with three environmental contexts of the home i.e. garden, everyday objects and interiors. A strengths-based rather than a deficit-based approach is adopted which draws upon an autistic person’s sensory preferences, special interests and action capabilities, to unravel what discomfort and delight might mean for an autistic person; this approach is translated into three design solutions to enhance their experience at home. By working beyond the boundaries of a neurotypical culture, the PhD bridges the autistic and neurotypical worlds of experience and draws upon what the mainstream design field can learn from designing with autistic people with additional learning disabilities. It also provides insights into the subjective experiences of people who have very different ways of seeing, doing and being in the environment

‘Running with the Fox and Hunting with the Hounds’: Social Work Tutors' Experiences of Managing Failing Social Work Students in Practice Learning Settings

The paper explores the issues raised by social work students failing in practice learning settings from the perspective of university tutors, by drawing on existing literature in this area from social work and nursing, as well as findings from a small‐scale empirical qualitative study. The qualitative study was influenced by practitioner‐researcher and practice‐near paradigms; and is based on interviews with twelve social work tutors in England. The findings reveal that tutors are able to articulate the important tasks and functions of their roles when issues of failing students in practice learning settings arise, although the process can be challenging. The challenges include: supporting practice educator and student, concerns about other tutors’ practices, the difficulties in promoting appropriate professional standards and values within higher education contexts and frustrations with practice educators and placements. Only a third of the respondents (four) however, articulated their gate keeping roles and responsibilities although this was not without its difficulties. Given the current reforms in social work education in England at this present time, with greater emphasis on threshold standards at entry level, and at key stages throughout the programme of study, the research is timely in terms of the critical consideration of the tutor role and challenges inherent in promoting appropriate standards.

‘It's been hell.’ Italian and British practice educators’ narratives of working with struggling or failing social work students in practice learning settings

Periods of assessed learning in practice settings are common requirements for social work students world wide. The ‘practice learning opportunity’ as it is known in the UK, and ‘tirocinio di servizio sociale’ as it is referred to in Italy, are important sites of gatekeeping in preventing unsuitable people from becoming social workers. The experience of assessing failing students in practice learning settings however, has been found to be particularly stressful and challenging for practice educators. This article documents findings from two qualitative studies that explored field educators’ experiences of working with struggling or failing social work students in Italy and England. The study finds both similarities and differences in the narratives of the assessors from the two countries Similarities include, unpleasant emotional experience of working with a failing student, internalisation of the students failing as the practice educators’ own failing, perceptions that the universities may hide negative information about students and lack of acknowledgement of the gatekeeping function inherent in the practice educator role. Differences include the level of emotionality experienced by educators, the way students are spoken about and the perceived role and responses of the university. Further comparative European research which focuses on practice education is indicated.

Contextual intelligent load management with ANN adaptive learning module

With the current increase of energy resources prices and environmental concerns intelligent load management systems are gaining more and more importance. This paper concerns a SCADA House Intelligent Management (SHIM) system that includes an optimization module using deterministic and genetic algorithm approaches. SHIM undertakes contextual load management based on the characterization of each situation. SHIM considers available generation resources, load demand, supplier/market electricity price, and consumers’ constraints and preferences. The paper focus on the recently developed learning module which is based on artificial neural networks (ANN). The learning module allows the adjustment of users’ profiles along SHIM lifetime. A case study considering a system with fourteen discrete and four variable loads managed by a SHIM system during five consecutive similar weekends is presented.

Reverse problem-based learning - a case study with a Braille machine

Engineering Education includes not only teaching theoretical fundamental concepts but also its verification during practical lessons in laboratories. The usual strategies to carry out this action are frequently based on Problem Based Learning, starting from a given state and proceeding forward to a target state. The possibility or the effectiveness of this procedure depends on previous states and if the present state was caused or resulted from earlier ones. This often happens in engineering education when the achieved results do not match the desired ones, e.g. when programming code is being developed or when the cause of the wrong behavior of an electronic circuit is being identified. It is thus important to also prepare students to proceed in the reverse way, i.e. given a start state generate the explanation or even the principles that underlie it. Later on, this sort of skills will be important. For instance, to a doctor making a patient?s story or to an engineer discovering the source of a malfunction. This learning methodology presents pedagogical advantages besides the enhanced preparation of students to their future work. The work presented on his document describes an automation project developed by a group of students in an engineering polytechnic school laboratory. The main objective was to improve the performance of a Braille machine. However, in a scenario of Reverse Problem-Based learning, students had first to discover and characterize the entire machine's function before being allowed (and being able) to propose a solution for the existing problem.

Machine learning Vasicek model calibration with Gaussian processes

In this article, we calibrate the Vasicek interest rate model under the risk neutral measure by learning the model parameters using Gaussian processes for machine learning regression. The calibration is done by maximizing the likelihood of zero coupon bond log prices, using mean and covariance functions computed analytically, as well as likelihood derivatives with respect to the parameters. The maximization method used is the conjugate gradients. The only prices needed for calibration are zero coupon bond prices and the parameters are directly obtained in the arbitrage free risk neutral measure.

Learning Activities with Semantic. Hypermedia in Higher Education

The increasing use of information and communication technologies (ICT) in diverse professional and personal contexts calls for new knowledge, and a set of abilities, competences and attitudes, for an active and participative citizenship. In this context it is acknowledged that universities have an important role innovating in the educational use of digital media to promote an inclusive digital literacy. The educational potential of digital technologies and resources has been recognized by both researchers and practitioners. Multiple pedagogical models and research approaches have already contributed to put in evidence the importance of adapting instructional and learning practices and processes to concrete contexts and educational goals. Still, academic and scientific communities believe further investments in ICT research is needed in higher education. This study focuses on educational models that may contribute to support digital technology uses, where these can have cognitive and educational relevance when compared to analogical technologies. A teaching and learning model, centered in the active role of the students in the exploration, production, presentation and discussion of interactive multimedia materials, was developed and applied using the internet and exploring emergent semantic hypermedia formats. The research approach focused on the definition of design principles for developing class activities that were applied in three different iterations in undergraduate courses from two institutions, namely the University of Texas at Austin, USA and the University of Lisbon, Portugal. The analysis of this study made possible to evaluate the potential and efficacy of the model proposed and the authoring tool chosen in the support of metacognitive skills and attitudes related to information structuring and management, storytelling and communication, using computers and the internet.

Learning domain-specific sentiment lexicons with applications to recommender systems

Search is now going beyond looking for factual information, and people wish to search for the opinions of others to help them in their own decision-making. Sentiment expressions or opinion expressions are used by users to express their opinion and embody important pieces of information, particularly in online commerce. The main problem that the present dissertation addresses is how to model text to find meaningful words that express a sentiment. In this context, I investigate the viability of automatically generating a sentiment lexicon for opinion retrieval and sentiment classification applications. For this research objective we propose to capture sentiment words that are derived from online users’ reviews. In this approach, we tackle a major challenge in sentiment analysis which is the detection of words that express subjective preference and domain-specific sentiment words such as jargon. To this aim we present a fully generative method that automatically learns a domain-specific lexicon and is fully independent of external sources. Sentiment lexicons can be applied in a broad set of applications, however popular recommendation algorithms have somehow been disconnected from sentiment analysis. Therefore, we present a study that explores the viability of applying sentiment analysis techniques to infer ratings in a recommendation algorithm. Furthermore, entities’ reputation is intrinsically associated with sentiment words that have a positive or negative relation with those entities. Hence, is provided a study that observes the viability of using a domain-specific lexicon to compute entities reputation. Finally, a recommendation system algorithm is improved with the use of sentiment-based ratings and entities reputation.