Increasing Performance of Supervised Machine Learning Methods by Analysis of Construction and Demolition Waste

Petr Kuritcyn, Katharina Anding, Gunther Notni

Abstract

Any recognition task, where the classes are given by quality rules or standards, needs the use of supervised machine learning. This paper discusses the ways of improvement the performance of methods of spectral analysis and supervised machine learning by classifying the construction and demolition waste (CDW). The first investigations in visible (VIS) and infrared (IR) spectrum have shown, that we can achieve a high recognition rate (98.3%). Therefore, investigations were done for analysing, which methods are useful for improvement classification performance of C&D aggregates.

Download

IMEKO-TC10-2016-067.pdf

DOI

-

IMEKO TC

TC10 - Measurement for Diagnostics, Optimization and Control

Back to the proceedings