Transforming Education with Deep Learning: A Systematic Review on Predicting Student Performance and Critical Challenges

M.; A.; M.; Mohammed; Ahmad

doi:https://doi.org/10.54216/FPA.180207

Journal of Artificial Intelligence and Metaheuristics | 2025

Comparative Analysis of Machine Learning Models and Greylag Goose Optimization for High-Accuracy Apple Quality Grading

A significant barrier to sustaining and achieving successful quality evaluation of apples has been the biases and inefficiencies of manual inspection, which cannot be effectively applied to large-scale industrial processing. This paper fills this gap by building an automated machine learning system for apple quality classification, which combines physical (size, weight, firmness) and sensory (sweetness, acidity, juiciness) features. Four baseline models, including K-Nearest Neighbors (KNN), Naive Bayes, Decision Tree, and Gradient Boosting, were trained on a preprocessed dataset. Imputation, categorical encoding, and feature scaling were used. To enhance predictive accuracy, hyperparameter tuning of three metaheuristic optimization algorithms — Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Greylag Goose Optimization (GGO) — was employed. The findings indicated that KNN had the highest baseline accuracy of 93.6% and that its accuracy improved to 98.6% with GGO optimization, which was better than that of GA and PSO. These results indicate that GGO-optimized KNN can provide a high-quality and scalable approach to assessing the quality of apples in the industry, offering enhanced accuracy, reduced variability, and a high degree of practical applicability in supply chain decision making, thereby increasing customer satisfaction and decreasing losses after harvesting.

groups

Marwa M. Eid mail -

Anis Ben Ghorbal mail

link https://doi.org/10.54216/JAIM.090106

Volume & Issue

Vol. Volume 9 / Iss. Issue 1

Details open_in_new

International Journal of Neutrosophic Science | 2025

Development of Neutrosophic Pareto Distribution for Survival Analysis

We provide a neutrosophic approach to the Pareto model, which is widely used to model survival data. In this paper, the neutrosophic Pareto model (NPM) is constructed under the framework of neutrosophic statistics, that can manage uncertain nature of data, commonly occur in many real word problems. This formulation generalizes the&ensp;classical model and is a useful method for dealing with fuzzy or uncertain data typically encountered in many applications in survival data. Using neutrosophic statistical framework, few key mathematic qualities of the proposed model such as its moments, survival function, and hazard rate are presented in the study. These properties are motivated and rigorously established to ensure theoretical soundness of the proposed model. Moreover, the maximum likelihood estimation (MLE) is used to estimate the neutrosophic parameters of the distribution. This approach is essential for deriving accurate parameter estimates from the data available, especially in cases where uncertainty or imprecision is present within the data as it is usually the case for any real-world situation. Based on the simulation experiment, we display the adequate performance of the suggested model. The simulations allow us to evaluate the performance of the routine as well as the stability of the model parameters across different settings. At the end, the real data analysis is conducted to show the applicability of proposed approach. The proposed model processes such a dataset filled with a range of uncertain values and presents its possibilities to be applied for information extraction from real world data sets that are abundant in uncertainty. Our results open a new avenue for neutrosophic statistical model approaches to the analysis&ensp;of survival data in subsequent studies.

groups

Ahmedia Musa M. Ibrahim mail

link https://doi.org/10.54216/IJNS.250426

Volume & Issue

Vol. Volume 25 / Iss. Issue 4

Details open_in_new

International Journal of BIM and Engineering Science | 2025

Sustainable Practices in the Design, Modeling, and Evaluation of Medical Centers: A Case Study of Basilia City

This study concludes that integrating sustainable practices into the design and modeling of medical centers significantly contributes to enhancing resource efficiency and reducing the environmental impact of healthcare facilities. The sustainability of these facilities can be further improved using advanced technologies such as Building Information Modeling (BIM), which simultaneously enhances the well-being of patients and staff. The study also highlights the importance of adopting globally recognized sustainability assessment systems and adapting them to suit the local context to ensure effective sustainability in future medical centers.

groups

Yasmmen Mashtah mail -

Alaa J. kadi mail -

Batoul Hasanin mail

link https://doi.org/10.54216/IJBES.100209

Volume & Issue

Vol. Volume 10 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

AlertFusion-OptiNet: An Advanced SIEM Alert Management System for IoT Environments using CMRO and AlertQ-Net

SIEM, which stands for Security Information and Event Management, is a collection of services and solutions that give businesses the capacity to gather, examine, and handle security-related data in real time from all areas of their IT infrastructure. This study presents AlertFusion-OptiNet, a sophisticated SIEM alert management architecture intended for effective alert handling and intrusion detection. The proposed CMRO algorithm (a hybrid of Coot Bird Optimization and Mug Ring Algorithm) is used to select the best features after the system integrates data from multiple sources (raw logs, network traffic, and security alerts), applies preprocessing to eliminate redundancy and inconsistencies, and extracts features using techniques like LDA, GloVe, statistical analysis, and DWT. PCA is then used to reduce dimensionality. The shortcomings of current intrusion detection systems include delayed alert replies, poor feature selection, and ineffective management of heterogeneous datasets. Two-channel CNNs, LSTM, and Bi-RNNs are used in AlertFusion-OptiNet's hybrid detection model to improve accuracy and real-time detection, while AlertQ-Net uses reinforcement learning to handle and monitor alerts continuously. The proposed AlertFusion-OptiNet accomplished 99.43% and outruns SOTA models.

groups

Abdullah Alenizi mail

link https://doi.org/10.54216/FPA.180201

Volume & Issue

Vol. Volume 18 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

Boosting Financial Risk Prediction Model Using Attention Mechanism Based Recurrent Neural Networks with Red‐Tailed Hawks Algorithm

The systemic prediction of financial risk issues has become a main attention in the area of finance. Financial risk is the main likelihood that stockholders will lose currency after they finance a business that has debt if the business flow of cash demonstrates insufficient to see its economic requirements. The incorporation of deep learning (DL) methods into financial risk forecast and investigation has altered conventional techniques. While traditional quantitative systems often trust basic metrics such as the highest reduction, the arrival of DL requires a more nuanced assessment, highlighting the model's generalization capability, particularly in market crises like stock market crashes. DL techniques are efficient in removing intricate patterns from massive data collections and become an effective model for forecasting financial trends. In this paper, we offer Boosting Financial Risk Prediction Model Using Attention Mechanism with Red‐Tailed Hawk (BFRPM-AMRTH) Algorithm. The presented BFRPM-AMRTH model aims to address the challenges of identifying and mitigating potential financial threats in a dynamic environment. Initially, the BFRPM-AMRTH technique applies the linear scaling normalization (LSN) data normalization technique to standardize the input features and ensure consistency across the dataset. In addition, the long short-term memory auto encoder with attention mechanism (LSTMA-AE) technique can be employed for classifying financial risks. Eventually, the red‐tailed hawk (RTH) algorithm adjusts the hyperparameter values of the LSTMA-AE algorithm optimally and outcomes in greater classification performance. To ensure the improved performance of BFRPM-AMRTH system, a huge range of simulation studies has been achieved and the obtained outcomes establish the advancement of the BFRPM-AMRTH system over the existing techniques

groups

Ilyos Abdullayev mail -

Hafis Hajiyev mail -

Mahfuza Sattarova mail -

Elena Klochko mail

link https://doi.org/10.54216/FPA.180202

Volume & Issue

Vol. Volume 18 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

Enhancing Visibility on Social Media with Categorization Machine Learning Analysis

A considerable number of individuals concentrate their engagement on social media sites, notably Instagram, YouTube, and Facebook, where they may adeptly exploit their popularity. A considerable volume of research studies has been conducted across diverse social networks to examine user profiles and their relationship with popularity. The primary emphasis of research concerning social media has centered on theme analysis, encompassing domains such as health, creativity, and awareness. This study utilizes K-means clustering to classify social media articles and determine the characteristics that contribute to their popularity. Gathered data from publications by international influencers during an eight-month duration. Producing roughly 161 posts daily and around 1092 posts monthly seeks to improve metrics including views, likes, and dislikes on social media. This strategy is designed to facilitate the growth of the platform's popularity, thereby maximizing visibility and outreach. The analysis focused on three factors: virility, appeal, and publicity rates. Classified the posts into five fundamental groups: casual, quality, number, support, and leader. The study yielded important findings on optimal publication timing, ideal video length, follower metrics, biography and caption lengths, and hashtag utilization. Researchers found some interesting things that will help people who use social media and brand owners make better marketing plans.

groups

Haitham S. Hasan mail

link https://doi.org/10.54216/FPA.180203

Volume & Issue

Vol. Volume 18 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

Financial Data Analysis for Financial Management Based on Cloud Computing Using Deep Reinforcement Learning Model

Maintainable ﬁnancial fraud detection includes the usage of viable and decent performs in the recognition of fraudulent actions in ﬁnancial region. A credit card is susceptible to cyber threats, which leads to a fraud of credit card. The fraudster does dishonest action by attaining illegal access to credit card information and this action affects an economic loss for the user as well as company. At present, deep learning (DL) and machine learning (ML), systems were deployed in financial fraud detection owing to their features’ ability of making a great device to find out fraudulent dealings. This paper presents a Financial Data Analysis for Financial Management Based on Cloud Computing Using Deep Reinforcement Learning Model (FDAFM-CCDRLM). The main intention of FDAFM-CCDRLM model is to improve analysis of financial data in the economic management. Initially, the min-max normalization is employed in the data normalization stage to convert a data of input into a suitable format. Besides, the proposed FDAFM-CCDRLM model designs a black‐winged kite algorithm (BKA) for the subset of feature selection process. For the classification process, the double deep Q‐network (DDQN) algorithm has been executed. At last, the artificial bee colony (ABC) algorithm-based hyperparameter range method is done for improving the classification outcomes of the DDQN model. The experimental evaluation of the FDAFM-CCDRLM system can be tested on a benchmark database. The extensive outcomes highlight the significant solution of the FDAFM-CCDRLM approach to the financial data analysis classification process

groups

Parviz Gurbanov mail -

Mansur Matkarimov mail -

Nilufar Sapayeva mail -

Alexey Nedelkin mail -

Andrey Kulik mail -

Olga Zanina mail

link https://doi.org/10.54216/FPA.180204

Volume & Issue

Vol. Volume 18 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

An Intelligent Model to combat Soybean Plant Disease based on Random Forest and Support Vector Machine Algorithms

Given that plant disease is the primary factor contributing to damage in most plants, decision makers in the agriculture industry are highly interested in enhancing prediction strategies to detect illness in plants at an early stage. This is crucial for ensuring timely and effective plant care. Classifying healthy soybean plants is a dependable and efficient use of noninvasive techniques like machine learning (ML). In this work, we used ML to enhance a smart forecasting model for the prediction of soybean diseases. We utilized two feature selection techniques, namely gain ratio and correlation, two supervised ML algorithms (support vector machine and Random forest) and the cross-validation technique was used for assessing the proposed system, such as accuracy, F-measure, specificity, executing time, and sensitivity. The suggested technique can readily differentiate between soybean plants that are infected and those that are healthy. The suggested approach has undergone testing using a comprehensive collection of soybean characteristics, as well as a subset of attributes. The findings show that performance metrics are impacted when soybean traits are reduced.

groups

Zainab A. Abdulazeez mail -

Israa Abdulkadhim Jabbar Al Ali mail -

Basma Mustafa M. H. mail -

Ghada Kamil Mustafa mail -

Refed Adnan Jaleel mail

link https://doi.org/10.54216/FPA.180205

Volume & Issue

Vol. Volume 18 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

Multi-Criteria Decision Support System for Predicting Financial Futures Using Ensemble of Deep Learning Algorithms with Heuristic Search Mechanisms

Financial markets are an intricate dynamic system. The difficulty comes from the contact among a market and its applicants, which means, the integrated consequence of the activities of whole applicants decides the market trend, while the market trend disturbs the actions of applicants. These linked interactions make financial markets keep developing. Financial markets are interchange financial instruments like savings certificates, bonds, stocks, and much more. Particularly in stocks, because variations in stock prices are inclined by numerous factors, with economic cycles, financial trends, financial structure, and other macro issues, as well as industry growth, listed businesses’ financial quality. In the last few years, deep learning (DL) and machine learning (ML) techniques have been very effective in predicting financial futures. This study develops a Multi-Criteria Decision Support System for Predicting Financial Futures Using Ensemble of Deep Learning Algorithms with Heuristic Search Mechanisms (MDSSPFF-EDLAHS) model. The main intention of the MDSSPFF-EDLAHS method is to predict future of finances using advanced ensemble models. At first, the data normalization stage applies min-max normalization for transforming input data into a beneficial format. Besides, the ensemble of deep learning models namely variational auto encoder (VAE), bidirectional long short-term memory (Bi-LSTM) technique, and dueling double deep Q-network (DDQN) system have been executed for the prediction of financial futures. At last, the spider wasp optimization (SWO) algorithm adjusts the hyperparameter values of the ensemble models optimally and outcomes in greater prediction performance. The experimental evaluation of the MDSSPFF-EDLAHS is examined on a benchmark dataset. The extensive outcomes highlight the significant solution of the MDSSPFF-EDLAHS approach to the financial future predicting process

groups

Elvir Akhmetshin mail -

Sanatbek Yakubov mail -

Khurshid Zaripov mail -

Rustem Shichiyakh mail

link https://doi.org/10.54216/FPA.180206

Volume & Issue

Vol. Volume 18 / Iss. Issue 2

Details open_in_new

Fusion: Practice and Applications | 2025

Transforming Education with Deep Learning: A Systematic Review on Predicting Student Performance and Critical Challenges

Deep learning (DL) is recognized as a breakthrough in the educational technology arena, more so in the sense that it can be applied for forecasting student performance and critical issues in academic systems. This systematic review is used to investigate advances in the DL-based system-to-predicting student performance and emphasizes its applicability, methodologies, and limitations. The paper analyses key technologies such as neural networks (NNs) and ensemble models used in educational data mining. The paper also points out limitations in previous studies, for example, data imbalance model interpretability, and issues of scalability. This review highlights the potential of DL to improve educational quality, provide personalized learning experiences, and mitigate learning hazards by synthesizing ideas from different studies. Future directions will comprise hybrid models, improvements in data preprocessing, and merging with real-time educational systems to optimize the performance of the prediction model in several academic environments. For this review, 58 papers were collected from the year 2017-2024 respectively based on DL in education, Risk in education, and student education performance analysis. Subsequently, the aim, technique used, dataset used, performance score attained, significance, and limitations of the existing studies were discussed in this review.

groups

M. Nazir mail -

A. Noraziah mail -

M. Rahmah mail -

Mohammed Fakherldin mail -

Ahmad Khawaji mail

link https://doi.org/10.54216/FPA.180207

Volume & Issue

Vol. Volume 18 / Iss. Issue 2

Details open_in_new