machine learning

Predicting student performance has become an urgent demand in most educational and higher education institutions. This is essential to help at-risk students and ensure their retention, provide excellent learning resources and experiences, and improve the ranking and reputation of institutions. However, this may be difficult to achieve for start-up organizations with small records to analyze. The purpose of the current research was to provide approaches based on traditional machine learning and regression on predicting students' performance.

The current research was of the qualitative research type and applied in terms of purpose and experimental analytical research in terms of method. Linear regression, decision tree, random forest and support vector machine methods were used in this research. In this section, after introducing the implementation environment, simulation parameters were introduced. In the following, by introducing the efficiency evaluation criteria of the proposed method, based on the described evaluation criteria, the findings were compared with other similar methods. For these comparisons, deep learning approach based on deep convolutional network and other deep learning approaches are used. In this research, the data collection of Dr. Hasht Roudi boys' school, which is among the top 10 institutions in Tehran, was used. The data of this institution are publicly available and can be downloaded through the GitHub site. Further investigation has been done on these data. The figure below shows the frequency of features in the dataset. that these features are considered as and on models.

Recommender systems try to help users and customers in choosing options and products. This article is part of the results of a research project in which a recommender system has been developed that can offer suitable courses to the student by receiving his preferences and attributes. The required basic concepts, applicable methods and evaluation criteria are introduced and the efficiency of the proposed method is verified with the help of test and evaluation. The obtained results show that the course recommender system can guide trainees to suitable training courses with proper accuracy.

Diabetes is a group of metabolic disorders that are the result of untreated high blood glucose greatly. Early diagnosis and continued control of this disease can reduce its effectsTherefore, providing a method for timely diagnosis of this disease is of great importance. Until now, researchers have made many efforts to provide machine learning methods to diagnose diabetes. But most of these models are either based on simple machine learning methods or based on the assumption that the available diabetes data are balanced. Both cases are factors of their complete failure. Therefore, considering the existing challenges as well as the importance of timely diagnosis of diabetes, in this research, a diabetes diagnosis model based on deep recurrent neural networks and SMOTE oversampling algorithm is presented. In this model, several pre-processing steps including quantification of missing values, removal of outlier data and then oversampling have been performed. Three deep recurrent neural networks with three recurrent hidden units including LSTM, GRU and BiLSTM have been used to diagnose diabetes. The results of the model presented on the Pima database received from the UCI repository indicate that the average accuracy in 10 different runs in LSTM, GRU and BiLSTM is 91.21%, 89.61% and 90.99%, respectively. The recurrent network with GRU unit has achieved the highest accuracy of 93.74% on average in 10 different executions. The results of the proposed model show that deep neural networks have a much more successful performance in diabetes diagnosis compared to traditional machine learning methods.

This article provides a comprehensive review of the impact of evolutionary optimization algorithms on improving learning processes. Evolutionary algorithms, inspired by natural evolution, offer powerful tools for solving complex optimization problems in machine learning. The review covers the theoretical foundations of these algorithms, including Genetic Algorithms, Particle Swarm Optimization, and Differential Evolution, and explores their applications in enhancing learning models. The study compares the performance and effectiveness of these algorithms in various machine learning contexts, highlighting their strengths in optimizing parameters, feature selection, and model structure. Additionally, the review identifies challenges such as parameter tuning, computational complexity, and convergence to local optima, which can limit the effectiveness of these algorithms. The findings suggest that evolutionary optimization algorithms have significant potential to improve learning processes, but also underscore the need for further research to address existing gaps and refine these methods. This review concludes with recommendations for future research directions, emphasizing the importance of developing hybrid algorithms and improving parameter tuning methods to achieve better performance in complex and high-dimensional learning tasks.