Deep learning has been successfully showing promising results in plant disease detection, fruit counting, yield estimation, and gaining an increasing interest in agriculture. Deep learning models are generally based on several millions of parameters that generate exceptionally large weight matrices. The latter requires large memory and computational power for training, testing, and deploying. Unfortunately, these requirements make it difficult to deploy on low-cost devices with limited resources that are present at the fieldwork. In addition, the lack or the bad quality of connectivity in farms does not allow remote computation. An approach that has been used to save memory and speed up the processing is to compress the models. In this work, we tackle the challenges related to the resource limitation by compressing some state-of-the-art models very often used in image classification. For this we apply model pruning and quantization to LeNet5, VGG16, and AlexNet. Original and compressed models were applied to the benchmark of plant seedling classification (V2 Plant Seedlings Dataset) and Flavia database. Results reveal that it is possible to compress the size of these models by a factor of 38 and to reduce the FLOPs of VGG16 by a factor of 99 without considerable loss of accuracy.

The accurate prediction of subscriber traffic is a critical challenge in telecommunication networks, as it is essential for making informed decisions to maintain consistent service quality. Various artificial intelligence (AI) methods have been explored for this purpose, with recurrent neural networks (RNNs) being particularly popular. However, RNNs often suffer from the gradient vanishing problem, which reduces their effectiveness in processing long sequences of data. Long short–term memory (LSTM) networks partially address this issue by controlling the propagation of learning errors, but they still face challenges related to convergence speed and residual learning errors. To address these limitations, this study proposes a new hybrid RNN architecture that combines features of Jordan and Frasconi networks for improved subscriber traffic prediction. The proposed architecture integrates two sublayers, a sigmoid and a Gaussian function, within a hidden layer that forms a loop. This configuration enables the network to leverage incremental learning through restricted Coulomb energy (RCE) at the local level, while also benefiting from global learning through backpropagation. The architecture was tested on a dataset consisting of cellular traffic data collected from a telecommunication operator in Cameroon. The results demonstrated that the hybrid RNN model outperforms traditional Jordan and Frasconi networks, achieving a root mean square error (RMSE) of 14.18 and a mean absolute error (MAE) of 8.61 using the Adam optimizer with a batch size of 64. In addition, the hybrid model showed superior performance in terms of RMSE, MAE, and convergence speed compared to the existing models. These findings suggest that the proposed model could support telecom operators in proactive congestion management and resource optimization.