Learning from time series:Supervised Aggregative Feature Extraction

Many modeling problems require to estimate a scalar output from one or more time series. Such problems are usually tackled by extracting a fixed number of features from the time series (like their statistical moments), with a consequent loss in information that leads to suboptimal predictive models. Moreover, feature extraction techniques usually make assumptions that are not met by real world settings (e.g. uniformly sampled time series of constant length), and fail to deliver a thorough methodology to deal with noisy data. In this paper a methodology based on functional learning is proposed to overcome the aforementioned problems; the proposed Supervised Aggregative Feature Extraction (SAFE) approach allows to derive continuous, smooth estimates of time series data (yielding aggregate local information), while simultaneously estimating a continuous shape function yielding optimal predictions. The SAFE paradigm enjoys several properties like closed form solution, incorporation of first and second order derivative information into the regressor matrix, interpretability of the generated functional predictor and the possibility to exploit Reproducing Kernel Hilbert Spaces setting to yield nonlinear predictive models. Simulation studies are provided to highlight the strengths of the new methodology w.r.t. standard unsupervised feature selection approaches. © 2012 IEEE.

MSC-clustering and forward stepwise regression for virtual metrology in highly correlated input spaces

Increasingly semiconductor manufacturers are exploring opportunities for virtual metrology (VM) enabled process monitoring and control as a means of reducing non-value added metrology and achieving ever more demanding wafer fabrication tolerances. However, developing robust, reliable and interpretable VM models can be very challenging due to the highly correlated input space often associated with the underpinning data sets. A particularly pertinent example is etch rate prediction of plasma etch processes from multichannel optical emission spectroscopy data. This paper proposes a novel input-clustering based forward stepwise regression methodology for VM model building in such highly correlated input spaces. Max Separation Clustering (MSC) is employed as a pre-processing step to identify a reduced srt of well-conditioned, representative variables that can then be used as inputs to state-of-the-art model building techniques such as Forward Selection Regression (FSR), Ridge regression, LASSO and Forward Selection Ridge Regression (FCRR). The methodology is validated on a benchmark semiconductor plasma etch dataset and the results obtained are compared with those achieved when the state-of-art approaches are applied directly to the data without the MSC pre-processing step. Significant performance improvements are observed when MSC is combined with FSR (13%) and FSRR (8.5%), but not with Ridge Regression (-1%) or LASSO (-32%). The optimal VM results are obtained using the MSC-FSR and MSC-FSRR generated models. © 2012 IEEE.

Machine Learning for Predictive Maintenance: A Multiple Classifiers Approach

In this paper a multiple classifier machine learning methodology for Predictive Maintenance (PdM) is presented. PdM is a prominent strategy for dealing with maintenance issues given the increasing need to minimize downtime and associated costs. One of the challenges with PdM is generating so called ’health factors’ or quantitative indicators of the status of a system associated with a given maintenance issue, and determining their relationship to operating costs and failure risk. The proposed PdM methodology allows dynamical decision rules to be adopted for maintenance management and can be used with high-dimensional and censored data problems. This is achieved by training multiple classification modules with different prediction horizons to provide different performance trade-offs in terms of frequency of unexpected breaks and unexploited lifetime and then employing this information in an operating cost based maintenance decision system to minimise expected costs. The effectiveness of the methodology is demonstrated using a simulated example and a benchmark semiconductor manufacturing maintenance problem.

Efficient Marginal Likelihood Computation for Gaussian Process Regression

In a Bayesian learning setting, the posterior distribution of a predictive model arises from a trade-off between its prior distribution and the conditional likelihood of observed data. Such distribution functions usually rely on additional hyperparameters which need to be tuned in order to achieve optimum predictive performance; this operation can be efficiently performed in an Empirical Bayes fashion by maximizing the posterior marginal likelihood of the observed data. Since the score function of this optimization problem is in general characterized by the presence of local optima, it is necessary to resort to global optimization strategies, which require a large number of function evaluations. Given that the evaluation is usually computationally intensive and badly scaled with respect to the dataset size, the maximum number of observations that can be treated simultaneously is quite limited. In this paper, we consider the case of hyperparameter tuning in Gaussian process regression. A straightforward implementation of the posterior log-likelihood for this model requires O(N^3) operations for every iteration of the optimization procedure, where N is the number of examples in the input dataset. We derive a novel set of identities that allow, after an initial overhead of O(N^3), the evaluation of the score function, as well as the Jacobian and Hessian matrices, in O(N) operations. We prove how the proposed identities, that follow from the eigendecomposition of the kernel matrix, yield a reduction of several orders of magnitude in the computation time for the hyperparameter optimization problem. Notably, the proposed solution provides computational advantages even with respect to state of the art approximations that rely on sparse kernel matrices.

Personalising Crime and Crime-fighting in Factual Television:an Analysis of Social Actors and Transitivity in Language and Images

This article addresses the lack of work on media and crime in Critical Discourse Analysis (CDA), using an example of a factual television crime report. The existing research in media studies and criminology points to the way that the media misrepresents crime by distorting public understandings and backgrounding structural issues, such as poverty, which are related to crime thereby legitimising a criminal justice system that serves the interests of the powerful in society. Using social actor and transitivity analysis, this article shows how multimodal CDA can make an important contribution as it reveals the more subtle linguistic strategies and visual representations by which this process is accomplished, showing how each plays a part in the recontextualisation of social practice. This programme backgrounds which crimes are committed but foregrounds mental states and the neutrality of policing.

An efficacy and mechanism evaluation study of Levosimendan for the Prevention of Acute oRgan Dysfunction in Sepsis (LeoPARDS): Protocol for a randomized controlled trial

Background

Organ dysfunction consequent to infection (‘severe sepsis’) is the leading cause of admission to an intensive care unit (ICU). In both animal models and early clinical studies the calcium channel sensitizer levosimendan has been demonstrated to have potentially beneficial effects on organ function. The aims of the Levosimendan for the Prevention of Acute oRgan Dysfunction in Sepsis (LeoPARDS) trial are to identify whether a 24-hour infusion of levosimendan will improve organ dysfunction in adults who have septic shock and to establish the safety profile of levosimendan in this group of patients.

This is a multicenter, randomized, double-blind, parallel group, placebo-controlled trial. Adults fulfilling the criteria for systemic inflammatory response syndrome due to infection, and requiring vasopressor therapy, will be eligible for inclusion in the trial. Within 24 hours of meeting these inclusion criteria, patients will be randomized in a 1:1 ratio stratified by the ICU to receive either levosimendan (0.05 to 0.2 μg.kg^-1.min^-1 or placebo for 24 hours in addition to standard care. The primary outcome measure is the mean Sequential Organ Failure Assessment (SOFA) score while in the ICU. Secondary outcomes include: central venous oxygen saturations and cardiac output; incidence and severity of renal failure using the Acute Kidney Injury Network criteria; duration of renal replacement therapy; serum bilirubin; time to liberation from mechanical ventilation; 28-day, hospital, 3 and 6 month survival; ICU and hospital length-of-stay; and days free from catecholamine therapy. Blood and urine samples will be collected on the day of inclusion, at 24 hours, and on days 4 and 6 post-inclusion for investigation of the mechanisms by which levosimendan might improve organ function. Eighty patients will have additional blood samples taken to measure levels of levosimendan and its active metabolites OR-1896 and OR-1855. A total of 516 patients will be recruited from approximately 25 ICUs in the United Kingdom.

This trial will test the efficacy of levosimendan to reduce acute organ dysfunction in adult patients who have septic shock and evaluate its biological mechanisms of action.

Looking for free riding: Energy efficiency incentives and Italian homeowners

We examine the effect of energy efficiency incentives on household energy efficiency home improvements. Starting in February 2007, Italian homeowners have been able to avail themselves of tax credits on the purchase and installation costs of certain types of energy efficiency renovations. We examine two such renovations—door/window replacements and heating system replacements—using multi-year cross-section data from the Italian Consumer Expenditure Survey and focusing on a narrow period around the introduction of the tax credits. Our regressions control for dwelling and household characteristics and economy-wide factors likely to influence the replacement rates. The effects of the policy are different for the two types of renovations. With window replacements, the policy is generally associated with a 30 % or stronger increase in the renovation rates and number of renovations. In the simplest econometric models, the effect is not statistically significant, but the results get stronger when we allow for heterogeneous effects across the country. With heating system replacements, simpler models suggest that the tax credits policy had no effect whatsoever or that free riding was rampant, i.e., people are now accepting subsidies for replacements that they would have done anyway. Further examination suggests a strong degree of heterogeneity in the effects across warmer and colder parts of the country, and effects in the colder areas that are even more pronounced than those for window replacements. These results should, however, be interpreted with caution due to the low rates of renovations, which imply that the effects are estimated relatively imprecisely.

Saccharomyces uvarum, a distinct group within Saccharomyces sensu stricto

A natural subgroup (that we refer to as Saccharomyces uvarum) was identified, within the heterogeneous species Saccharomyces bayanus. The typical electrophoretic karyotype, interfertility of hybrids between strains, distinctive sugar fermentation pattern, and uniform fermentation characteristics in must, indicated that this subgroup was not only highly homogeneous, but also clearly distinguishable from other species within the Saccharomyces sensu stricto group. Investigation of the S. bayanus type strain and other strains that have been classified as S. bayanus, confirmed the apparent lack of homogeneity and, in some cases, supported the hypothesis that they are natural hybrids. Copyright (C) 1999 Federation of European Microbiological Societies.

Radioprotection of targeted and bystander cells by methylproamine

INTRODUCTION: Radioprotective agents are of interest for application in radiotherapy for cancer and in public health medicine in the context of accidental radiation exposure. Methylproamine is the lead compound of a class of radioprotectors which act as DNA binding anti-oxidants, enabling the repair of transient radiation-induced oxidative DNA lesions. This study tested methylproamine for the radioprotection of both directly targeted and bystander cells.

METHODS: T98G glioma cells were treated with 15 μM methylproamine and exposed to (137)Cs γ-ray/X-ray irradiation and He(2+) microbeam irradiation. Radioprotection of directly targeted cells and bystander cells was measured by clonogenic survival or γH2AX assay.

RESULTS: Radioprotection of directly targeted T98G cells by methylproamine was observed for (137)Cs γ-rays and X-rays but not for He(2+) charged particle irradiation. The effect of methylproamine on the bystander cell population was tested for both X-ray irradiation and He(2+) ion microbeam irradiation. The X-ray bystander experiments were carried out by medium transfer from irradiated to non-irradiated cultures and three experimental designs were tested. Radioprotection was only observed when recipient cells were pretreated with the drug prior to exposure to the conditioned medium. In microbeam bystander experiments targeted and nontargeted cells were co-cultured with continuous methylproamine treatment during irradiation and postradiation incubation; radioprotection of bystander cells was observed.

DISCUSSION AND CONCLUSION: Methylproamine protected targeted cells from DNA damage caused by γ-ray or X-ray radiation but not He(2+) ion radiation. Protection of bystander cells was independent of the type of radiation which the donor population received.

Analytical model of interwoven spiral arrays

Interweaving planar spiral conductors in doubly periodic arrays enable substantially sub-wavelength resonant response along with broadening fractional bandwidth. A self-contained analytical model is proposed to accurately predict the characteristics of the intertwined quadrifilar spiral array near the fundamental resonance. The model, based upon a multiconductor transmission line (MTL) approach, provides physical insight into the unique properties of the distributed interactions between the interleaved counter-wound spiral arms extended beyond a single unit cell and elucidates the mechanisms underlying the array performance at normal and oblique incidence of TE and TM polarised waves. The developed MTL model is instrumental in the design of the artificial surfaces with the specified response.