Morisita-Horn’s Correlated Gaussian Regressive Deep Boltzmann Machine Neural Network Classification for Soil Quality Prediction

Abstract

Soil is theimportant carbon and comprised carbon-rich naturalelement. The quality of soil is
used for particular soil kind using normal limits, for maintaining the efficiency and improves
the water level. In recent days, many researches carried out their research on soil quality
prediction methods. But, the soil quality prediction accuracy was not enhancedas well as time
was not reduced by existing methods. In order to handle these limitations, Morisita-Horn’s
Correlated Gaussian Regressive Deep Boltzmann Machine Neural Network Classification
(MHCGRDBNNC) Method is introduced. The aim of proposed MHCGRDBMNNC Method
is to perform an efficient feature selection process and classification process for improving the
soil quality prediction performance. Initially in MHCGRDBMNNC Method, the number of
soil data points is gathered with help of IoT devices positioned at different locations. Then,
Gaussian Process Regression is used in MHCGRDBMNNC Method to choose the relevant
features for minimizing the time consumption and space consumption during soil quality
prediction. After that, the Morisita-Horn Correlated Index is carried out in MHCGRDBMNNC
Method to classify the soil data points into particular class for efficient soil quality prediction.
This in turn helps to improve the accuracy and time consumption. Simulation is performedwith
various metrics namely accuracy, space complexity, as well as soil quality prediction time
using number of soil data points. Qualitative and quantitative outcomesindicate
MHCGRDBMNNC Method is more effective to achieve higher soil quality prediction
accuracy and lesser space complexity as well as time when compared with existing approaches.