Publica
Hier finden Sie wissenschaftliche Publikationen aus den FraunhoferInstituten. Modeling continuoustime stochastic processes using NCurve mixtures
 Online im WWW, 2019, arXiV:1908.04030v4, 25 S. 

 Englisch 
 Bericht, Elektronische Publikation 
 Fraunhofer IOSB () 
 multistep sequence prediction; Stochastic Process Modeling; Bezier Curves; Mixture Density Networks 
Abstract
Representations of sequential data are commonly based on the assumption that observed sequences are realizations of an unknown underlying stochastic process, where the learning problem includes determination of the model parameters. In this context the model must be able to capture the multimodal nature of the data, without blurring between modes. This property is essential for applications like trajectory prediction or human motion modeling. Towards this end, a neural network model for continuoustime stochastic processes usable for sequence prediction is proposed. The model is based on Mixture Density Networks using Bézier curves with Gaussian random variables as control points (abbrev.: NCurves). Key advantages of the model include the ability of generating smooth multimode predictions in a single inference step which reduces the need for Monte Carlo simulation, as required in many multistep prediction models, based on stateoftheart neural networks. Essential properties of the proposed approach are illustrated by several toy examples and the task of multistep sequence prediction. Further, the model performance is evaluated on two real world usecases, i.e. human trajectory prediction and human motion modeling, outperforming different stateoftheart models.