ISE Seminar “Machine Learning and Computer Simulation (Digital Twin) for Stochastic Decision Making”

Details

ISE Seminar “Machine Learning and Computer Simulation (Digital Twin) for Stochastic Decision Making”

Chun-Hung Chen
Dept. of Systems Engineering & Operations Research
George Mason University, USA

ABSTRACT

While machine learning enjoys widespread popularity, its efficacy relies on the quality of accessible data and the depth of training, thereby constraining its ability to swiftly respond to unforeseen events. Transfer learning, by retaining knowledge within pre-trained models, accelerates adaptation to new tasks using smaller datasets and fewer resources. Nevertheless, its success is still contingent upon the data and training. In contrast, computer simulation (or digital twin) offers a remedy for scenarios where data are sparse or events remain unobserved, though it entails computational expenses, especially in the pursuit of optimal decision-making. In our presentation, we will introduce an innovative integration of computer simulation and machine learning designed to efficiently identify optimal decisions, while effectively tackling challenges related to data availability, training constraints, and unpredictability of future events. A key component of our methodologies is a popular technique called Optimal Computing Budget Allocation (OCBA) originally pioneered by the speaker, which aims to maximize the efficiency of simulation optimization.

Zoom link for the seminar: https://ufl.zoom.us/j/94548030552