Section: New Results

Machine Learnig applied to Smart Farming

Participants : Jamal Ammouri ( Internship Cnam ), Malika Boudiaf ( Ummto, Tizi-Ouzou, Algeria ), Samia Bouzefrane ( Cnam ), Pascale Minet, Meziane Yacoub ( Cnam ).

Intelligent Farming System (IFS) is made possible by the use of 4 elements: sensors and actuators, the Internet of Things (IoT), edge/cloud processing, and machine learning.

Soil degradation and a hot climate explain the poor yield of olive groves in North Algeria. Edaphic, climatic and geographical data were collected from 10 olive groves over several years and analyzed by means of Self-Organizing Maps (SOMs). SOM is a non-supervised neural network that projects high-dimensional data onto a low-dimension discrete space, called a topological map, such that close data are mapped onto nearby locations on the map. In the paper [28] presented at the PEMWN 2019 conference, we have shown how to use self-organizing maps to determine olive grove clusters with similar features, characterize each cluster and show the temporal evolution of each olive grove. With the SOM, it becomes possible to alert the farmer when some specific action needs to be done in the case of hydric stress, NPK stress, pest/disease attack. As a result, the nutritional quality of the oil produced is improved. SOM can be integrated in the Intelligent Farming System (IFS) to boost conservation agriculture.

This work requires a strong collaboration with agronomists. Malika Boudiaf (Laboratoire Ressources Naturelles, UMMTO, Tizi-Ouzou, Algeria) provided the data set and gave us many explanations about soil conservation. Meziane Yacoub (Cnam) is an expert in SOMs. Jamal Ammouri (Cnam) was co-advised by Samia Bouzefrane, Pascale Minet and Meziane Yacoub.