Pavithira, L (2022) Deep Maximum Entropy with Memetic Reinforced Learning for Agriculture Drought Prediction. Deep Maximum Entropy with Memetic Reinforced Learning for Agriculture Drought Prediction, 20 (15). pp. 3304-3316. ISSN 1303-5150

[thumbnail of Dr.L.Pavithira November 2022.pdf] Text
Dr.L.Pavithira November 2022.pdf - Published Version
Download (1MB)

Abstract

The most obvious impact of agricultural drought is a decrease in crop yield as a result of irregular
and inadequate rainfall. Drought forecasting has a massive effect on vegetation. As a result,
predicting the occurrence of drought early in the season could help to mitigate the detrimental
effects of drought. Drought prediction requires information on rainfall, temperature, and pressure
for its decision-making model in order to enact proper agriculture planning. Though there is much
literature related to agriculture drought prediction, the class imbalance is the most influential factor
which affects the learning rate of the classification model. This paper focuses on overcoming class
imbalance to optimize agriculture drought prediction model by devising a deep learning and
memetic nature inspired model. In this proposed model, deep maximum entropy is used for
identifying decision features in the dataset that provides maximum information about the
conditional feature to improve the accuracy of drought prediction. In conventional RL, a policy that
gets high reward alone is considered for training the model randomly. Deep Maximum Entropy
Reinforcement Learning aims to maximize entropy of policy to handle the class imbalance issue. The
parameters of the RL are optimized by inducing shuffled frog leaping algorithms with its food
searching strategy for assigning optimal values to the parameters of the RL involved in prediction of
Agriculture Drought. The simulation results proved that the proposed DME-MRL produces the
highest rate of accuracy for agriculture drought prediction compared to other existing models.
Keywords: Entropy, Reinforcement Learning, Fuzzy Subtractive clustering

Item Type: Article
Uncontrolled Keywords: Entropy, Reinforcement Learning, Fuzzy Subtractive clustering
Divisions: PSG College of Arts and Science > Department of Computer Science
Depositing User: Dr. B Sivakumar
Date Deposited: 16 Apr 2026 07:36
Last Modified: 16 Apr 2026 07:36
URI: https://ir.psgcas.ac.in/id/eprint/2792

Actions (login required)

View Item
View Item