Random Forest and K-Nearest Neighbors Prediction of the Optical Properties of 2D Ultra-thin Quantum Dots under Hydrostatic Pressure

Abstract

This work analyzes the effect of hydrostatic pressure on the predicting excitonic optical response of a 2D Ultra-thin GaAs-system. The data are generated from a numerical model incorporating quantum confinement and excitonic effects, capturing the system's response to incident photon energy and applied pressure. To reduce computational effort and efficiently explore the parameter space, we employ two machine learning algorithms K-Nearest Neighbors (KNN) and Random Forest Regression (RFR) to predict the optical absorption coefficient based on the numerically generated dataset. Results reveal a strong agreement between predicted and numerical values across all pressure levels. Both KNN and RFR successfully replicate excitonic absorption peaks and their pressure-induced shifts, demonstrating that these machine learning approaches provide accurate, reliable, and computationally efficient alternatives to conventional numerical simulations for the design and optimization of optoelectronic devices.