Ivo Hrib, Jan Šturm, Oleksandra Topal and Maja Škrjanc

Abstract
With the rapid expansion of the computing industry, efficient
energy utilization and reduction of CO2 emissions are critically
important. This research develops analytical tools to predict CO2
emissions from various machine learning processes. We present a
novel methodology for data acquisition and analysis of CO2 emis-
sions during model training and testing. Our results demonstrate
the environmental impact of different algorithms and provide
insights into optimizing energy consumption in artificial intelli-
gence applications.