Abstract

Abstract
We present a method for improving the prediction accuracy using multiple predictive algorithms. Several techniques have been developed to tackle this issue such as bagging, boosting and stacking. In contrary to the first two that, usually, generate homogeneous ensembles of classifiers, stacking techniques have demonstrated success using heterogeneous ensembles. In our method, we adopt the stacking mechanism. Several models are generated using different learning algorithms. Forward stepwise selection is implemented to link each instance to its appropriate learning model. Experiments with three datasets benchmarked with four learning schemes show that this novel method improves prediction accuracy and can serve as a bridge to transfer knowledge between tasks given the same feature space but different data distributions. © 2018, Springer Nature Switzerland AG.

Abstract
Considering the avalanche of evolving data and the memory constraints, streaming networks’ sampling has gained much attention in the recent decade. However, samples choosing data uniformly from the beginning to the end of a temporal stream are not very relevant for temporally evolving networks where recent activities are more important than the old events. Moreover, the relationships also change overtime. Recent interactions are evident to show the current status of relationships, nevertheless some old stronger relations are also substantially significant. Considering the above issues we propose a fast memory less dynamic sampling mechanism for weighted or multi-graph high-speed streams. For this purpose, we use a forgetting function with two parameters that help introduce biases on the network based on time and relationship strengths. Our experiments on real-world data sets show that our samples not only preserve the basic properties like degree distributions but also maintain the temporal distribution correlations. We also observe that our method generates samples with increased efficiency. It also outperforms current sampling algorithms in the area.

Abstract
This special issue collects selected articles from the Third International Workshop on Linearity (LINEARITY 2014), which was held in Vienna, on July 13th, 2014. The workshop was a one-day satellite event of FLoC 2014, the sixth Federated Logic Conference, which was held as part of the 2014 Vienna Summer of Logic. Copyright © Cambridge University Press 2016

Abstract

Abstract
Scalability is one of the most relevant features of today’s data management systems. In order to achieve high scalability and availability, recent distributed key-value stores refrain from costly replica coordination when processing requests. However, these systems typically do not perform well under churn. In this paper, we propose DataFlagons, a large-scale key-value store that integrates epidemic dissemination with a probabilistic total order broadcast algorithm. By ensuring that all replicas process requests in the same order, DataFlagons provides probabilistic strong data consistency while achieving high scalability and robustness under churn.