SYSTEM AND METHOD FOR PREDICTING THE SPECTROSCOPIC PROPERTIES OF MOLECULES BASED ON MACHINE LEARNING

摘要

A machine learning-based molecular spectral property prediction system and method, wherein the spectral property prediction system according to an embodiment receives molecular information corresponding to at least one target molecule and surrounding molecular system information, A preprocessing unit for vectorizing system information, a feature extraction unit for extracting molecular characteristics corresponding to the vectorized molecular information and surrounding molecular system characteristics corresponding to the vectorized neighboring molecular system information, and molecular characteristics and neighboring molecular system characteristics as inputs and an integrated characteristic extracting unit for extracting integrated characteristics between molecular characteristics and surrounding molecular properties using the received integrated characteristic extraction algorithm, and a spectral characteristic output unit for outputting a predicted value of spectral characteristics of a target molecule corresponding to the integrated characteristics.