ELECTROENCEPHALOGRAM-BASED DEPRESSION CLASSIFICATION USING DEEP NEURAL NETWORK

ABSTRACT

Plants like human are affected by numerous diseases which retard their growth. The result of this retarred growth is not only very detrimental to the plants but also to the human and the nation at large. A major challenge towards forestalling the effect of different diseases on plants is the inability of farmers to quickly identify by sight the type of disease affecting plants, and thereby necessitate taking prompt actions to avoid plant death or low yield of plants produce. Efforts have been made by researchers to deploy intelligent solutions, adopt data science and machine learning approaches to solve this challenge. In this paper, a review of deep learning techniques and intelligent machine learning models adopted for the prediction, classification and detection of plant diseases is conducted with the aim of providing insights towards the development of a more efficient, generalized model with improved accuracy for plant disease diagnosis. Result findings revealed that intelligent plant diseases diagnostic models developed by the combination of two or more (ensemble) techniques performed better than single-base models with an average accuracy rate ranging from 90% to 100%. Furthermore, the score ratings of reviewed publications showed that 14.29% of intelligent models are developed with limited and unbalanced classes of real dataset usually less than five thousand (5000) instances (images). This results in the high rate of misclassification and inefficient memory utilization due to data augmentation technique deployed by most plant disease diagnostic models.

Keywords: Artificial Intelligence, Machine learning, Plant Diseases, Diagnosis, Accuracy.