Tackling new machine learning problems with neural networks always means optimizing numerous hyperparameters that define their structure and strongly impact performances. In this work, we study the use of goal-oriented sensitivity analysis, based on Hilbert–Schmidt independence criterion (HSIC), for hyperparameter analysis optimization. Hyperparameters live in spaces are often complex awkward. ...

We introduce a goal-oriented strategy for multiscale computations performed using the Multiscale Finite Element Method (MsFEM). In previous work, we have shown how to use, in MsFEM framework, concept of Constitutive Relation Error (CRE) obtain guaranteed and fully computable posteriori error estimate energy norm (as well as indicators on various sources). Here, CRE is coupled with solution an a...

Human conversations are guided by short-term and long-term goals. We study how to plan goal sequences as coherently humans do naturally direct them an assigned in open-domain conversations. Goal a series of knowledge graph (KG) entity-relation connections generated KG walkers that traverse through the KG. The existing recurrent attention based either insufficiently utilize conversation states o...