Emulating Simulation Models with Neural Networks

Singh, Aryan (2023) Emulating Simulation Models with Neural Networks. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration manual) Download (2MB) | Preview

Abstract

Deep Learning (DL) algorithms have innovated and repurposed a myriad of fields and even improved methodologies previously thought to be optimal. Modeling, especially Simulation Models (SM) is one such field. In this paper, a novel methodology is proposed to improve the performance and computation time of SMs through the application of DL models as surrogate models. An SM of traffic behavior in an inter-connected road network was created as the data source for this project. This model was run iteratively with varying parameters to generate data points. The I/O data of the SM was stored and curated to be used as the training data for the DL models. The relationships between the features in the data points generated by the SM were studied to model a Graph. Two NNs were trained on the raw simulation data and the transformed graph data, respectively. The NNs were evaluated on their accuracy and R-squared values against the output of the SM. The NNs provided high accuracy and were able to map the relationships between the features of the SM. The primary goal of this research was to identify and quantify the performance benefits of using an NN to emulate an SM. The processing speeds of both the models were compared and the Neural Networks proved to be exponentially faster than the Simulation model.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Rifai, Hicham UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QA Mathematics > Algebra > Algorithms > Computer algorithms T Technology > TA Engineering (General). Civil engineering (General) > Systems engineering > Simulation methods > Mathematical models > Computer simulation Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Tamara Malone
Date Deposited:	03 Jan 2025 12:28
Last Modified:	03 Jan 2025 12:28
URI:	https://norma.ncirl.ie/id/eprint/7273

Actions (login required)

View Item