We consider the problem of efficiently learning optimal control policies and value functions over large state spaces in an online setting in which estimates must be available after each interaction with the world. This paper develops an explicitly model-based approach extending the Dyna architecture to linear function approximation. Dynastyle planning proceeds by generating imaginary experience...

After discussing general properties of implicit Finite Element analysis using ANSYS and explicit analysis using LS-DYNA it is shown when and how quasi-static limit load analyses can be performed by a transient analysis using explicit time integration. Then we focus on the remaining benefits of implicit analysis and how a proper combination of ANSYS and LS-DYNA can be used to prepare the transie...

Modern statistical AI systems are quite large and complex; this interferes with research, development, and education. We observe that most of the computation involves solving systems of equations of a certain kind. Specifically, recursive queries look up and aggregate relevant or potentially relevant values. If the results of these queries are memoized for reuse, the memos may need to be update...

PURPOSE Dyna-ATC is a unilateral external fixator with angulator, lengthener, and translator, which allows for angular correction and compensation of the secondary displacement during angular correction. The purpose of this study is to introduce surgical technique and calculation methods and to evaluate the clinical outcome of angular deformity correction using Dyna-ATC. MATERIALS AND METHODS...

A dynamic aquatic model (DynA model) was previously developed to predict the fate of a chemical in aquatic scenarios characterized by daily or periodic changes in several input parameters. DynA model is here calibrated with data obtained from the literature in specific unsteady state scenarios, such as those of rice fields. The results obtained for two herbicides (cinosulfuron and pretilachlor)...

This thesis presents a novel form of learning by reinforcement. Existing reinforcement learning algorithms rely on the provision of external reward signals to drive the learning algorithm. This new algorithm relies on reinforcing signals generated internally within the algorithm. The algorithm, SRS/E, described here generates expectancies ( -hypotheses), each of which gives rise to a specific p...