As discussed in Section 7.2.4, both expert-based and deep learning (DL) approaches have respective strengths and shortcomings in realistic trajectory modeling, and the combination of both methods has the penitential for improved performance. Thus, in collaboration with Hao Cheng, we conduct a preliminary work to combine the expert-based GSFM model with a deep learning model, LSTM-DBSAN, by considering the possibility of conflict as the basis of choosing between GSFM and LSTM-DBSCAN for predicting trajectories in any specific situation. We name the combined model as GSFM-w-LSTM; it is discussed in same detail in the following section.
Towards Combining Expert- and DL-based Models
Modeling Interactions among Pedestrians and Cars in Shared Spaces ; Kapitel : 8 ; 135-140
2022-06-26
6 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Towards an expert system on ionospheric informatics
Elsevier | 1990
|A simulation engine - combining an expert system with a simulation language
Tema Archiv | 1988
|An Expert System and Associated Models
British Library Online Contents | 1998
|