Predictive Ensemble Pruning by Expectation Propagation

An ensemble is a group of learners that work together as a committee to solve a problem. The existing ensemble learning algorithms often generate unnecessarily large ensembles, which consume extra computational resource and may degrade the generalization performance. Ensemble pruning algorithms aim to find a good subset of ensemble members to constitute a small ensemble, which saves the computational resource and performs as well as, or better than, the unpruned ensemble. This paper introduces a probabilistic ensemble pruning algorithm by choosing a set of “sparse” combination weights, most of which are zeros, to prune the ensemble. In order to obtain the set of sparse combination weights and satisfy the nonnegative constraint of the combination weights, a left-truncated, nonnegative, Gaussian prior is adopted over every combination weight. Expectation propagation (EP) algorithm is employed to approximate the posterior estimation of the weight vector. The leave-one-out (LOO) error can be obtained as a by-product in the training of EP without extra computation and is a good indication for the generalization error. Therefore, the LOO error is used together with the Bayesian evidence for model selection in this algorithm. An empirical study on several regression and classification benchmark data sets shows that our algorithm utilizes far less component learners but performs as well as, or better than, the unpruned ensemble. Our results are very competitive compared with other ensemble pruning algorithms.     

Interactive survival analysis with the OCDM system: From development to application

Medical data mining is currently actively pursued in computer science and statistical research but not in medical practice. The reasons therefore lie in the difficulties of handling and statistically analyzing medical data. We have developed a system that allows practitioners in the field to interactively analyze their data without assistance of statisticians or data mining experts. In the course of this paper we will introduce data mining of medical data and show how this can be achieved for survival data. We will demonstrate how to solve common problems of interactive survival analysis by presenting the Online Clinical Data Mining (OCDM) system. Thereby the main focus is on similarity based queries, a new method to select similar cases based on their covariables and the influence of these on their survival.     

Finding Your True IT Transformation

If you want to transform your business and cause a dramatic improvement for the enterprise, you'll need to start by truly transforming your IT life cycle. It's easy to believe that IT is truly reinventing itself by hiring new people, installing new tools, or preaching a new methodology. However, providing transformational leadership starts with the leadership itself along with the development of the three core competencies of strategy, solutions, and implementation. Your job in IT leadership is to develop all three competencies in your organization by cultivating the necessary skills and techniques, and then to cause each of them to happen continuously and concurrently. When these changes become the organization's new habits, you're on the path to true transformation.     

An algebra of hybrid systems

Hybrid systems are heterogeneous systems characterised by the interaction of discrete and continuous dynamics. We present a trajectory-based algebraic model for describing hybrid systems; the trajectories used are closely related to streams. The algebra is based on left quantales and left semirings and provides a new application for these algebraic structures. We show that hybrid automata, which are probably the standard tool for describing hybrid systems, can conveniently be embedded into our algebra. Moreover we point out some important advantages of the algebraic approach. In particular, we show how to handle Zeno effects, which are excluded by most other authors. The development of the theory is illustrated by a running example and a larger case study.     

Classification of semantic relations between nominals

The NLP community has shown a renewed interest in deeper semantic analyses, among them automatic recognition of semantic relations in text. We present the development and evaluation of a semantic analysis task: automatic recognition of relations between pairs of nominals in a sentence. The task was part of SemEval-2007, the fourth edition of the semantic evaluation event previously known as SensEval. Apart from the observations we have made, the long-lasting effect of this task may be a framework for comparing approaches to the task. We introduce the problem of recognizing relations between nominals, and in particular the process of drafting and refining the definitions of the semantic relations. We show how we created the training and test data, list and briefly describe the 15 participating systems, discuss the results, and conclude with the lessons learned in the course of this exercise.     

Capillary-driven pumping for passive degassing and fuel supply in direct methanol fuel cells

In this paper we present a new concept of creating and using capillary pressure gradients for passive degassing and passive methanol supply in direct methanol fuel cells (DMFCs). An anode flow field consisting of parallel tapered channels structures is applied to achieve the passive supply mechanism. The flow is propelled by the surface forces of deformed CO₂ bubbles, generated as a reaction product during DMFC operation. This work focuses on studying the influence of channel geometry and surface properties on the capillary-induced liquid flow rates at various bubbly gas flow rates. Besides the aspect ratios and opening angles of the tapered channels, the static contact angle as well as the effect of contact angle hysteresis has been identified to significantly influence the liquid flow rates induced by capillary forces at the bubble menisci. Applying the novel concept, we show that the liquid flow rates are up to thirteen times higher than the methanol oxidation reaction on the anode requires. Experimental results are presented that demonstrate the continuous passive operation of a DMFC for more than 15 h.     

Synaptic proteome changes in the superior frontal gyrus and occipital cortex of the alcoholic brain

Cognitive deficits and behavioral changes that result from chronic alcohol abuse are a consequence of neuropathological changes that alter signal transmission through the neural network. To focus on the changes that occur at the point of connection between the neural network cells, synaptosomal preparations from post‐mortem human brain of six chronic alcoholics and six non‐alcoholic controls were compared using 2‐D differential in‐gel electrophoresis (DIGE). Functionally affected and spared regions (superior frontal gyrus, SFG, and occipital cortex, OC, respectively) were analyzed from both groups to further investigate the specific pathological response that alcoholism has on the brain. Forty‐nine proteins were differentially regulated between the SFG of alcoholics and the SFG of controls and 94 proteins were regulated in the OC with an overlap of 23 proteins. Additionally, the SFG was compared to the OC within each group (alcoholics or controls) to identify region‐specific differences. A selection was identified by MALDI‐TOF mass spectrometry revealing proteins involved in vesicle transport, metabolism, folding and trafficking, and signal transduction, all of which have the potential to influence synaptic activity. A number of proteins identified in this study have been previously related to alcoholism; however, the focus on synaptic proteins has also uncovered novel alcoholism‐affected proteins. Further exploration of these proteins will illuminate the mechanisms altering synaptic plasticity, and thus neuronal signaling and response, in the alcoholic brain.     

The development of the Quick Exposure Check (QEC) for assessing exposure to risk factors for work-related musculoskeletal disorders

This paper describes the development and evaluation of the Quick Exposure Check (QEC), which is an observational tool developed for Occupational Safety and Health (OSH) practitioners to assess exposure to risks for work-related musculoskeletal disorders and provide a basis for ergonomic interventions. The tool is based on epidemiological evidence and investigations of OSH practitioners' aptitudes for undertaking assessments. It has been tested, modified and validated using simulated and workplace tasks, in two phases of development, with participation of 206 practitioners. The QEC allows the four main body areas to be assessed and involves practitioners and workers in the assessment. Trials have determined its usability, intra- and inter-observer reliability, and validity which show it is applicable to a wide range of working activities. The tool focuses primarily on physical workplace factors, but also includes the evaluation of psychosocial factors. Tasks can normally be assessed within 10min. It has a scoring system, and exposure levels have been proposed to guide priorities for intervention. Subsequently it should be used to evaluate the effectiveness of any interventions made. The QEC can contribute to a holistic assessment of all the elements of a work system.