Sabir Zulqurnain, Simbawa Eman
Department of Computer Science and Mathematics, Lebanese American University, Beirut, Lebanon.
Department of Mathematics, Faculty of Science, King Abdulaziz University, Jeddah, Saudi Arabia.
PLoS One. 2025 Apr 1;20(4):e0320327. doi: 10.1371/journal.pone.0320327. eCollection 2025.
The current research investigations present the numerical solutions of the anthrax disease system in animals by designing a machine learning stochastic procedure. The mathematical anthrax disease system in animals is classified into susceptible, infected, recovered and vaccinated.
A Runge-Kutta solver is applied to collect the dataset, which decreases the mean square error by dividing into training as 78%, testing 12% and verification 10%. The proposed stochastic computing technique is performed through the logistic sigmoid activation function, and a single hidden layer construction, twenty-seven numbers of neurons, and optimization through the Bayesian regularization for the mathematical anthrax disease system in animals.
The designed procedure's correctness is authenticated through the results overlapping and reducible absolute error, which are calculated around 10-05 to 10-08 for each case of the model. The best training performances are performed as 10-10 to 10-12 of the model. Moreover, the statistical performances in terms of regression coefficient, error histogram, and state transition values enhance the reliability of the proposed stochastic machine learning approach.
The designed scheme is not applied before to get the numerical results of the anthrax disease system in animals.
当前的研究通过设计一种机器学习随机程序来呈现动物炭疽病系统的数值解。动物体内的数学炭疽病系统分为易感、感染、康复和接种疫苗四类。
应用龙格 - 库塔求解器收集数据集，通过将其按78%用于训练、12%用于测试和10%用于验证来降低均方误差。所提出的随机计算技术通过逻辑 sigmoid 激活函数、单隐藏层结构、二十七个神经元以及针对动物体内数学炭疽病系统的贝叶斯正则化优化来执行。
通过结果重叠和可降低的绝对误差验证了所设计程序的正确性，对于模型的每种情况，计算得出的误差在10^-05到10^-08左右。模型的最佳训练性能为10^-10到10^-12。此外，回归系数、误差直方图和状态转换值方面的统计性能提高了所提出的随机机器学习方法的可靠性。
所设计的方案以前未被用于获取动物炭疽病系统的数值结果。
