A regular diet for diabetics is a diet with a low glycemic load, which reduces the risk of certain chronic diseases and thus increases life expectancy. In this paper, we present the development of an optimal diet model using multi-objective programming to improve nutritional balance by minimizing the cost and total glycemic load of foods and satisfying the daily nutritional requirements consistent with the living standards of the Moroccan population.
Path planning of intelligent agents in an emergency context is one of the most popular issues within nowadays context. This work proposes an environment acquisition and a path optimization solution based on reinforcement learning. The proposed solution implements Q-Learning algorithm and enables the agent to choose the path that maximizes the reward and minimizes the penalty. When tested in an experiment grid and compared to other solutions the proposed solution proved to be more stable and more efficient.
Date palms cultivation is a crucial factor in oasis agriculture. The latter plays an essential role in the socioeconomics of many countries, including Morocco. However, many diseases threaten this valuable tree causing economic and environmental damages. White Scale is one of the most harmful pests that negatively affect the quality of date fruits and reduce their yield. Therefore, early detection of infected plants is essential in any Pest Management System. In this study, we propose a framework based on feature extraction and Machine Learning techniques to automatically classify the degree of infestation by Date Palm White Scale Disease. Gray-Level Co-occurrence Matrix features and HSV Color Moments were extracted and combined, then fitted to a K-Nearest Neighbors classifier that outperformed an average accuracy of 96.90%. In addition, other metrics such as Precision, Recall, F1-score, and confusion matrix are calculated to obtain more details about the stage-wise classification performance of the proposed framework. Finally, we conducted a comparative analysis between the proposed framework and some state-of-the-art-based methods. As a result, the proposed approach surpasses the state-of-the-art models in terms of various performance metrics. It is anticipated that this application will help farmers and stakeholders across the country and worldwide on both a macro and micro scale.
In this paper, we propose a novel architecture called quaternion discrete Racah moments convolutional neural network (QRMCNN) for color face recognition to improve the recognition accuracy and to reduce the computational complexity. Quaternion Racah moments (QRM) can extract effective features from color images. The use of QRM as first layer aims to reduce the input matrix size of the proposed architecture in the earliest orders and hence enormously decrease the number of parameters and the computational time. Experiments are carried out on faces96 and FEI face datasets to demonstrate the efficiency of the proposed architecture in the training process. The obtained results indicate clearly that the QRMCNN outperforms previous methods in terms of recognition rate and GPU elapsed time.
Metaheuristics have been commonly used in several engineering optimizations, they prerequisite reduced time to converge and produce a better-improved solution. The most applied are Evolutionary Algorithms. Many complex problems are no longer considered difficult due to swarm intelligent optimization algorithms that supply rapid and reliable methods for solutions. and this returns to its features such as robustness, flexibility, self-organization, parallel, and distributive. In this paper, a comparative study among five metaheuristics algorithms in terms of convergence, robustness, and computing time is accomplished, and three benchmark functions are applied to perform simulation experiments with Genetic Algorithm (GA), Firefly Algorithm (FA), Particle Swarm Algorithm (PSO), Invasive Weed optimization (IWO), and Grey Wolf optimizer (GWO). The experimental results show that GE provides more accuracy for complex optimization problems, while the GWO and PSO are better in terms of convergence speed.
