An Analysis of the Impact of Machine Learning Model Parameters on the Effectiveness of Classifying Disturbances in Forest Ecosystems

The goal of the project is to conduct a quantitative analysis of the impact of selected machine learning model parameters on the effectiveness of classifying disturbances in forest ecosystems based on image data. The work will include preparing and standardizing the dataset, defining classes and an evaluation protocol, and then implementing and comparing models in controlled experiments in which key model parameters and training procedures will be systematically varied.

As part of the project, comparative tests will be conducted to identify the parameters that most strongly influence prediction quality and the most effective configurations. An important element of the project will be ensuring the quality of the input data: data for the project will be obtained through collaboration with the Machine Learning Team, which will provide access to relevant data resources and support in their preliminary preparation.

The expected outcome will be a set of conclusions and recommendations regarding the selection of model parameters for the task of classifying forest disturbances, as well as research material serving as a basis for further work.

Machine-translated
Przetłumaczone maszynowo