Search

•Crowdsourcing enables independent and unbiased evaluations of computational methods.•Markers predictive of smoking status were identified in blood gene expression data.•Discriminating former and never smokers was challenging using blood transcriptomics.•The development of human and cross-species inductive predictive models was a success.•The crowd wisdom enabled to find consensus blood smoking exposure gene signatures.

Alzheimer’s disease (AD) is the most common form of dementia with physical, psychological, social, and economic impacts on patients, their carers, and society. Its early diagnosis allows clinicians to initiate the treatment as early as possible to arrest or slow down the disease progression more effectively. We consider the problem of classifying AD patients through a machine learning approach using different data modalities acquired by non-invasive techniques. We perform an extensive evaluation of a machine learning classification procedure using omics, imaging, and clinical features, extracted by the ANMerge dataset, taken alone or combined together. Experimental results suggest that integrating omics and imaging features leads to better performance than any of them taken separately. Moreover, clinical features consisting of just two cognitive test scores always lead to better performance than any of the other types of data or their combinations. Since these features are usually involved in the clinician diagnosis process, our results show how their adoption as classification features positively biases the results.

Abstract: Supervised classification of data affected by noise or error, with unknown probability distribution, is a challenging task. To this extend, we propose the Fuzzy Regularized Eigenvalue Classifier, based on a recent technique to classify data in two or more classes. We compare the execution time and accuracy of the classifier with other de facto standard methods. With the adoption of a novel membership function, the classifier is capable to produce more accurate models that well compare with results obtained by other methods, and fuzzy weighting functions, on benchmark datasets. [Copyright &y& Elsevier]

In this paper we propose a block Lanczos algorithm suitable for MIMD distributed memory message passing architectures. It is based on an efficient parallelizaiion of basic linear algebra operations, such as matrix-matrix, sparse matrix-matrix, and dense QR factorization. We assume an unidirectional ring as connection topology and a block column wrap-around matrices distribution. We have chosen this approach to improve load-balancing, to eliminate the intersection of messages and to decrease communicationThe parallel Lanczos algorithm has been tested on a Convex Meta Series, a cluster of HP Series 9000 workstations running the PVM communication system. Results of the performance evaluation based on some classical parameters are shown.