Paper Details

PJB-2025-244

Application of NNE-CNN for non-numerical individual feature extraction in rice breeding optimization

Xiao Han, Qingrui Zhang, Ziting Gao, Xiaoliang He and Fenglou Ling
Abstract

---

Rice breeding plays a vital role in ensuring global food security; however, effective breeding programs require rapid, accurate, and objective identification of rice diseases. The optimization of rice breeding holds significant importance in achieving increased and high-yield production. Conventional visual inspection methods are labour-intensive, subjective, and prone to human error, particularly when disease symptoms are subtle or visually similar. To address these limitations, this study proposes a non-numerical encoding–based convolutional neural network (NNE-CNN) for rice disease recognition. It expounds upon the algorithm's implementation principles, encompassing aspects such as initialization methods, encoding techniques, and population updating procedures. The performance of the proposed model was evaluated using two public benchmark datasets (Convex and Rectangles) as well as a rice disease image dataset. Subsequently, it utilizes the differential operator to compute an updating operator for the individuals. This process involves the removal or sparsity of certain convolutional layers within the CNN, thereby reducing the model's complexity and computational overhead while preserving its generalization capabilities. By deleting or sparsifying part of the convolutional layers in the CNN, the complexity and computational overhead of the model are reduced, while maintaining the generalization ability of the model. Experimental results demonstrate that NNE-CNN consistently outperforms classical CNN architectures and other intelligent algorithm-optimized models, achieving accuracies of 96.94% and 96.17% on the public datasets and a test accuracy of 97.50% on rice disease images. These findings indicate that the proposed method provides a robust and computationally efficient solution for rice disease identification, with practical implications for automated crop monitoring and data-driven rice breeding optimization.

 

To Cite this article: Han, X., Q. Zhang, Z. Gao, X. He and F. Ling. 2026. Application of NNE-CNN for non-numerical individual feature extraction in rice breeding optimization. Pak. J. Bot., 58(4): DOI: http://dx.doi.org/10.30848/PJB2026-4(17)  
Download PDF