This report investigates the issue of selecting and scaling ground motions as input excitations for response history analyses of buildings in performance-based earthquake engineering. Many ground motion selection and modification (GMSM) procedures have been developed to select ground motions for a wide variety of objectives. This report focuses on the selection and scaling of single, horizontal...

Entity resolution (ER) presents unique challenges for evaluation methodology. While crowd sourcing provides a platform to acquire ground truth, sound approaches to sampling must drive labelling efforts. In ER, extreme class imbalance between matching and non-matching records can lead to enormous labelling requirements when seeking statistically consistent estimates of population parameters. Thi...

||||||||||||| R esum e. Une m ethode de simulation fond ee sur des techniques d'echantil-lonage d'importance, de Gibbs et Metropolis-Hastings permet d'approcher le rapport de la vraisemblance en deux valeurs du param etre, et donc l'estimateur du maximum de vraisemblance, dans le cadre g en eral des mod-eles dynamiques avec variables latentes. Cette classe contient comme cas particuliers les mo...

Training deep directed graphical models with many hidden variables and performing inference remains a major challenge. Helmholtz machines and deep belief networks are such models, and the wake-sleep algorithm has been proposed to train them. The wake-sleep algorithm relies on training not just the directed generative model but also a conditional generative model (the inference network) that run...

Hypothesis testing of variable importance measures (VIMPs) is still the subject ongoing research. This particularly applies to random forests (RF), for which VIMPs are a popular feature. Among recent developments, heuristic approaches parametric have been proposed whose distributional assumptions based on empirical evidence. Other formal tests under regularity conditions were derived analytical...