Enhancing Forest Cover Type Classification Through Deep Learning Neural Networks

Deforestation is a huge problem in its harm to natural ecosystems and contribution to climate change. Reforestation efforts should be empowered with the use of technology. Specifically, forest planning can be made more efficient with the practice of AI. Various logistic regression models and artific...

Full description

Saved in:
Bibliographic Details
Published in2024 9th International Conference on Mechatronics Engineering (ICOM) pp. 277 - 280
Main Authors Baldovino, Renann G., Tolentino, Aldrin Joshua C.
Format Conference Proceeding
LanguageEnglish
Published IEEE 13.08.2024
Subjects
Accuracy
batch normalization
Biological system modeling
Climate change
deep learning
Deforestation
dropouts
Logistic regression
Mechatronics
Reforestation
Online AccessGet full text

Cover

More Information
Summary:Deforestation is a huge problem in its harm to natural ecosystems and contribution to climate change. Reforestation efforts should be empowered with the use of technology. Specifically, forest planning can be made more efficient with the practice of AI. Various logistic regression models and artificial neural networks (ANN) have been applied to forest cover type classification with good to great accuracy. However, this study aims to apply a deep learning approach and compare its advantages to these kinds of problem. Models proposed had varying results with the convolutional neural network (CNN) performing badly with low accuracy, while the use of batch normalization and dropouts resulted into a balanced fit and accurate model. The study suggests to apply the simple neural network approach to other forest management activities to improve and hasten their processes.
DOI:10.1109/ICOM61675.2024.10652531