Search for: All records

Solar flares not only pose risks to outer space technologies and astronauts’ well being, but also cause disruptions on earth to our high-tech, interconnected infrastructure our lives highly depend on. While a number of machine-learning methods have been proposed to improve flare prediction, none of them, to the best of our knowledge, have investigated the impact of outliers on the reliability and robustness of those models’ performance. In this study, we investigate the impact of outliers in a multivariate time series benchmark dataset, namely SWAN-SF, on flare prediction models, and test our hypothesis. That is, there exist outliers in SWAN-SF, removal of which enhances the performance of the prediction models on unseen datasets. We employ Isolation Forest to detect the outliers among the weaker flare instances. Several experiments are carried out using a large range of contamination rates which determine the percentage of present outliers. We assess the quality of each dataset in terms of its actual contamination using TimeSeriesSVC. In our best findings, we achieve a 279% increase in True Skill Statistic and 68% increase in Heidke Skill Score. The results show that overall a significant improvement can be achieved for flare prediction if outliers are detected and removed properly. 

Fiber-based materials are prevalent around us. While microscopically these systems resemble a discrete assembly of randomly interconnected fibers, the network architecture varies from one system to another. To identify the role of the network architecture, we study here cellular and fibrous random networks in tension and compression, and in the context of large strain elasticity. We observe that, compared to cellular networks of same global parameter set, fibrous networks exhibit in tension reduced strain stiffening, reduced fiber alignment, and reduced Poisson’s contraction in uniaxial tension. These effects are due to the larger number of kinematic constraints in the form of cross-links per fiber in the fibrous case. The dependence of the small strain modulus on network density is cubic in the fibrous case and quadratic in the cellular case. This difference persists when the number of cross-links per fiber in the fibrous case is rendered equal to that of the cellular case, which indicates that the different scaling is due to the higher structural disorder of the fibrous networks. The behavior of the two network types in compression is similar, although softening induced by fiber buckling and strain localization is less pronounced in the fibrous case. The contribution of transient interfiber contacts is weak in tension and important in compression