Federated Learning for Crop Yield Estimation

Abstract

Data is a precious asset in Artificial Intelligence, as it is basic source of training algo rithms in AI. And data related to agriculture sector is also an important sector of any country. As, food is basic need of every living organism. Every living organism needs food for his existence. And increasing word population also increased food demand in every country.The timely prediction of crops can help in many ways it can help farmers and government. Crop yield estimation helps at National and regional level.Crop yield estimation is itself a very complex problem because we need genotype data, management data and other data like soil data and weather data. And the major barrier in crop yield estimation is the availability of data. Due to competitive environment data owners don’t feel safe in sharing their information. And collection of data to a central machine is also tedious. In this study we are going to use federated learning for crop yield estimation. federated learning is everywhere like it helps in detection of bank fraud, self-driving cars, digital health care, industry and etc. In recent two to three years it can be seen that federated learning also used on crop related data and providing outstanding results. We want to get advantage of machine learning models without using centralized data and for this advantage we will be using a machine learning based approach that uses decentralized data for estimation.