Building the Rice Blast Disease Prediction Model based on Machine Learning and Neural Networks

EasyChair Preprint no. 1197

Abstract

Rice blast disease (RBD) is one of the most damaging crop disease for the rice in Taiwan. RBD may be widespread and cause severe losses if it is not controlled in the early stage. The goal of this research is to build an early warning mechanism for the RBD using the machine learning model under current climatic condition. Five years of climatic data (ranging from 2014 to 2018) are used as candidate features in our model, which are collected by the Taiwan government. The RBD conditions are labeled via the field observation during these years. With the climate data, we conduct the recursive feature elimination algorithm to select the key features that have impacts on the RBD. To derive the RBD prediction model, we applied the Auto-Sklearn and neural network algorithms to train the classification model. The experiment results show that the proposed model can classify the RBD conditions (whether exacerbated or relieved) with an accuracy of 72% in average. In particular, our model can achieve an accuracy of 89% in the exacerbation case, which demonstrates the effectiveness of the proposed classification model.

Keyphrases: Auto-Sklearn, neural network, Recursive Feature Elimination, rice blast disease

@Booklet{EasyChair:1197,
  author = {Jia-You Hsieh and Wei Huang and Hsin-Tieh Yang and Chia-Chieh Lin and Yo-Chung Fan and Huan Chen},
  title = {Building the Rice Blast Disease Prediction Model based on Machine Learning and Neural Networks},
  howpublished = {EasyChair Preprint no. 1197},

  year = {EasyChair, 2019}}