Malaria diagnosis relies heavily on time-consuming manual microscopy. Automated AI offers solutions; however, remote regions often lack the advanced technical infrastructure to execute them. Most research work heavily relies on massive deep convolutional neural networks. Although accurate, these uninterpretable models demand substantial computational resources, making them unrealistic for low-resource clinics. Existing studies also miss critical methodological shortcomings, such as data leakage. Our aim is to develop an accurate, interpretable, and lightweight automated malaria detection framework efficiently engineered for low-resource edge deployment. Thus, we propose an Efficient Mathematical Feature Extraction (EMFE) pipeline using Gray World normalization and adaptive thresholding to extract biological features. These classical machine learning models are optimized using Bayesian Optimization and validated using nested 20-Fold Stratified Cross-Validation for strict data separation. The optimized Random Forest classifier achieved a robust 94.96% accuracy (p < 0.0001). Further, this pipeline reduces the memory footprint to 0.180 MB, reduces power consumption below 10W, and accelerates CPU inference to 1.8 milliseconds per image compared to deep learning baselines, eliminating GPU dependency. This study bridges the gap between diagnostic accuracy and real-world deployability by replacing complex deep learning architectures with a transparent, mathematically grounded framework, providing a sustainable blueprint for global infectious disease screening.

Automated malaria diagnosis from stained blood-smear microscopy is dominated by deep convolutional neural networks that are accurate but computationally expensive, poorly interpretable, and rarely validated with patient-level rigor. We present EMFE (Efficient Mathematical Feature Extraction), a five-feature framework for classifying single red-blood-cell images as parasitized or uninfected using Gray World color normalization, adaptive green-channel thresholding, morphological spot detection, and classical machine learning. Using the NIH LHNCBC malaria dataset (27,558 images from 200 patients), we evaluate Random Forest, Histogram Gradient Boosting, and Support Vector Machine classifiers under patient-grouped nested cross-validation (K_outer=20, K_inner=3), ensuring that cells from each patient remain within a single fold. The optimized Random Forest achieves 94.6

In real-world agriculture, healthy plant leaves are significantly more common than diseased ones. This natural class imbalance presents challenges in automated plant disease detection, as analyzing each leaf with computationally intensive deep-learning models is problematic, leading to inefficiency and increased resource consumption. To tackle this challenge and promote sustainable AI solutions, this study presents an iterative, hybrid AI approach that boosts computational efficiency, interpretability, and scalability for real-time disease detection. This hybrid system operates in two stages: first, a lightweight traditional machine learning classifier performs binary classification to quickly separate and exclude healthy leaves, followed by a deep learning model (ResNet, DenseNet, MobileNet, and EfficientNet) that classifies the specific disease in the smaller group of diseased leaves. This two-stage method minimizes computational load while maintaining high classification accuracy. Additionally, this study uses Explainable AI (XAI) methods, particularly Gradient-weighted Class Activation Mapping (Grad-CAM), to generate heatmaps. These heatmaps highlight the image areas that most significantly influence the model's predictions, thereby improving transparency and refining the feature extraction process. The proposed hybrid model, comprising Logistic Regression and Mobilenetv3, offers up to 77.6% faster inference than conventional deep learning models with only about 3% accuracy loss. For a large-scale test of 1,227 images on an entry-level laptop, the hybrid model reduced the total inference time from 4,548 seconds to just 1,010.13 seconds, with minimal CPU load. By addressing class imbalance, optimizing inference efficiency, and incorporating explainable AI, this work contributes a scalable, sustainable, and trustworthy solution for plant disease detection in precision agriculture.

Precise feature extraction is vital in traffic scene analysis, especially in vehicle detection and segmentation. In this research, the performance of Histogram of Oriented Gradients (HOG) in extracting meaningful visual features from traffic images is assessed. Two instances are under examination: one with an empty road and the other with vehicles on the road. By comparing and analyzing HOG features of both scenarios, we see how traffic affects feature intricacy. The features are then processed further with Decision Tree, Support Vector Classifier (SVC), Multi-Layer Perceptron (MLP), Random Forest and Logistic Regression in order to analyze their contribution towards object detection. The findings reveal that under notraffic conditions, HOG is seen to target static background structures most of the time, while under vehicle conditions, the detected features are more complex with the presence of added edges and contours. This work emphasizes the effect of traffic on feature extraction and validates the application of HOG-based methods for automated traffic surveillance and vehicle detection.

Smoking is a severe threat to public health that impacts Bangladesh's general population as well as university students. Many people engage up smoking as a habit as a result of personal stressors or social trends, frequently disregarding the adverse consequences of smoking. Thirty-five percent of Bangladeshi adults smoke or use electronic cigarettes, according to the WHO. This study investigates the root causes of smoking and its broader implications through a sustainable AI-driven approach. This is because our study to find out the reason behind it and the consequence of smoking risk in personal life. In this study, various aspects of life were considered to establish the effects of smoking. Most people who used to do daily smoke of what with a huge amount did not lead a better life in personal life. The use of these findings assists people, corporations, and authorities in identifying the primary hazards of smoking that may be lethal. To predict the Risk of smoking KNN, Logistic Regression, Random Forest, XGBoost, LightGBM, Decision Tree, Naive Bayes and SVM were practiced to get the model. By collecting data via the survey form and keeping nine questions from the service, 501 data collections were made from various persons. Among the models, SVM and XGBoost perform better than all other working classifiers attaining 98.40% and 97% accuracy. In this study, the integration of sustainable AI practice enhances resource scalability and long-term planning. It enables the allocation of resources better to achieve the desired outcomes besides other institution's health services to introduce and implement successful interventions. Specifically, this work underscores the significance of sustainable AI for development for improving various public health causes and forming a healthier and more equitable society.

Background and Objective: Deep learning models have high computational needs and lack interpretability but are often the first choice for medical image classification tasks. This study addresses whether complex neural networks are essential for the simple binary classification task of malaria. We introduce the Extracted Morphological Feature Engineered (EMFE) pipeline, a transparent, reproducible, and low compute machine learning approach tailored explicitly for simple cell morphology, designed to achieve deep learning performance levels on a simple CPU only setup with the practical aim of real world deployment. Methods: The study used the NIH Malaria Cell Images dataset, with two features extracted from each cell image: the number of non background pixels and the number of holes within the cell. Logistic Regression and Random Forest were compared against ResNet18, DenseNet121, MobileNetV2, and EfficientNet across accuracy, model size, and CPU inference time. An ensemble model was created by combining Logistic Regression and Random Forests to achieve higher accuracy while retaining efficiency. Results: The single variable Logistic Regression model achieved a test accuracy of 94.80 percent with a file size of 1.2 kB and negligible inference latency (2.3 ms). The two stage ensemble improved accuracy to 97.15 percent. In contrast, the deep learning methods require 13.6 MB to 44.7 MB of storage and show significantly higher inference times (68 ms). Conclusion: This study shows that a compact feature engineering approach can produce clinically meaningful classification performance while offering gains in transparency, reproducibility, speed, and deployment feasibility. The proposed pipeline demonstrates that simple interpretable features paired with lightweight models can serve as a practical diagnostic solution for environments with limited computational resources.

Pneumonia remains a significant public health challenge in underdeveloped regions, where timely diagnosis is often hindered by limited resources and delayed medical intervention. To address this issue, an automated diagnostic system was developed using the ResNet50 convolutional neural network architecture and the FastAI deep learning framework. Leveraging transfer learning, the model was trained on chest X-ray images to accurately detect pneumonia. Despite being designed for short-term training, the system achieved strong diagnostic performance, highlighting the potential of deep learning in medical image analysis. The results indicate that such models can significantly reduce diagnostic time and improve early detection of pneumonia in underserved areas.

One of the most alarming issues in digital society is hate speech (HS) on social media. The severity is so high that researchers across the globe are captivated by this domain. A notable amount of work has been conducted to address the identification and alarm system. However, a noticeable gap exists, especially for low-resource languages. Comprehensive datasets are the main problem among the constrained resource languages, such as Bangla. Interestingly, hate speech or any particular speech has no single dimensionality. Similarly, the hate component can simultaneously have multiple abusive attributes, which seems to be missed in the existing datasets. Thus, a multi-label Bangla hate speech dataset named BOISHOMMO has been compiled and evaluated in this work. That includes categories of HS across race, gender, religion, politics, and more. With over two thousand annotated examples, BOISHOMMO provides a nuanced understanding of hate speech in Bangla and highlights the complexities of processing non-Latin scripts. Apart from evaluating with multiple algorithmic approaches, it also highlights the complexities of processing Bangla text and assesses model performance. This unique multi-label approach enriches future hate speech detection and analysis studies for low-resource languages by providing a more nuanced, diverse dataset.

Video game addiction is a rapidly growing concern among young people, emerging as a major issue due to its harmful impact on their mental health. This study investigates how predictive models might reveal subtle patterns in gaming habits by examining the relationship between video games, youth development, and machine learning. The research will investigate social dynamics as well as cognitive, academic, and mental health implications in order to educate individuals about responsible gaming habits and teaching tactics. The study covers ethical considerations and long-term societal ramifications, emphasizing the significance of a compatible way to navigating the IT based world because of young people's prosperity. Its purpose is to evaluate whether a gamer is career-minded. That's why we gathered 804 data points from them, of which 780 were utilized. The key point that we utilized were name, gender, age, amount of time spent studying, amount of sleep or wakefulness, amount of time spent playing games, reason for playing games, amount of time spent playing games more than with family, time spent honing skills, and time spent worried about a career. and after that, they underwent reprocessing and were checked before being used with certain machine learning algorithms. Various prediction and find ways use machine learning, artificial intelligence, and deep learning method. Our approaches include Gaussian Naive Bayes (GNB), Random Forest (RF), Adaptive Boosting (ADA Boosting), Decision Tree (DT), Extra Trees (ET), Gradient Boosting (GB) classifier, KNN (K-Nearest Neighbors), Support Vector Machine (SVM). The Random Forest classifier performed better in our analysis than the other eight methods. While all eight classifiers have great prediction accuracy, Random Forest Classifier (RF) has the highest accuracy of 97.00 %. This study analyses the various behaviours or habits of the current youth society. Based on the results obtained from that analysis, a machine learning-based model has been proposed. This proposed model will help in understanding the behavior of young people and their current situation.

The recent era has seen significant growth of technologies in the field of Natural Language Processing (NLP). But the scarce resource languages like Bengali have not got much attention from the research community. The BERT language model has laid a very positive impact on the performance of the NLP tasks. Although several others language models came into the scenario, we investigate the performance of BERT model and other conventional methods for the sentiment classification task in Bengali text. The obtained result shows that BERT overperformed other conventional machine learning and lexicon-based methods in all aspects of the performance metrics. Along with BERT, conventional methods namely Logistic Regression, Decision Tree, SVM, Random Forest, Naïve Bayes and Neural Network were implemented. Besides these methods a lexicon-based approach was used to see the overall variation in the results. The lexicon resource for Benali was created for this implementation.

Advancements in technology have revolutionized various sectors, including agriculture, which serves as the backbone of many economies, particularly in Asian countries. The integration of new technologies and research has consistently aimed to enhance cultivation rates and reduce reliance on manual labor. Two key technologies, Artificial Intelligence (AI) and the Internet of Things (IoT), have emerged as pivotal tools in automating processes, providing recommendations, and monitoring agricultural activities to optimize results. While traditional soil cultivation has been the preferred method, the increasing urbanization trend necessitates alternative approaches such as hydroponics, which replaces soil with water as the medium for crop cultivation. Having many significant advantages, hydroponics serves a crucial role in achieving efficient space utilization. To get a higher density of plants in a confined area hydroponic approach provides water, nutrients and other essential elements directly to the plant's root. To utilize the hydroponic system more effectively, our proposed method, integrating AI and IoT helps to provide suitable crop recommendations, monitor the parameters of the plants and also suggest the necessary changes required for gaining optimal parameters. To ensure optimal resource allocation and maximize yields we have used machine learning models and trained them to recommend suitable crops from the given parameters and also refer to the changes in parameters that are needed for better plant growth. We have used the crop recommendation dataset from the Indian Chamber of Food and Agriculture to train our proposed machinelearning model. Our selected machine learning algorithms to predict the best crops are Random forests, Decision trees, SVM, KNN, and XGBoost. Our research combines AI and IoT with hydroponic systems to streamline crop recommendations, automate monitoring processes, and provide real-time guidance for optimized cultivation. Among them, the Random forest algorithm outperformed other algorithms with an accuracy of 97.5%.

There are many varieties of local fruits in Bangladesh. Now some of them are moribund. The aim of this paper is to introduce these local and moribund fruits to the next generation and to build a local fruit model that can be used beneficially for future research in the field of local fruits. Though all Bangladeshi fruit names and their identities are already entered in the database, but all the information is not accessible always, and most people don’t know about them. There are a lot of local fruits in Bangladesh which contain so many calories that it’s beneficial to us, but the calorie content of the same is unknown. This problem can be solved by an automatic fruit identifier, which would make many people familiar with the fruits in a timely manner. The calories of the fruit detected can also be calculated using an Android app. The proposed app is not only calculating calories, but it’s introducing local fruits to the next generation. A method of deep convolutional neural networks (CNN) model is used to classify fruits from fruit images. To detect fruit and calculate its calorie content, this model is incorporated into the mobile app. In terms of accuracy, the model has reached about 98.03%. In the field of real-time fruit identification, this method is very useful.

This paper proposes a new method called RL- RNN that combines the power of Reinforcement Learning and Recurrent Neural Networks in furthering malicious bot detection. Malicious social bots on Twitter have emerged as a significant threat to this platform, spreading disinformation, manipulating public opinion, and influencing political events. The existing traditional detection methods, such as Support Vector Machines, are proven to be unable to deal with bot behaviors dynamically and complicatedly. The proposed model inculcates that URL features with frequency and patterns of postings are hot indicators of bot activity. The RL component enables the model to adjust dynamically to changing bot behaviors. It also adapts the RNN type to capture the sequential structure of the tweets with the associated URLs. Thus, experimental results show that this approach using RLRNN hybrids significantly outperforms traditional SVM-based methods by precision, recall, and overall detection accuracy. The experimental results point out the RL-RNN model's ability to scalability in detecting evolving bot strategies on Twitter. This makes the approach more effective as a mitigation method for preventing malicious activities, thus increasing security in social media systems.

Aim/Background: This research aims to develop an automated contextual classifier for scholarly papers by utilizing established algorithms and understanding the information retention of different parts of a scholarly article, such as the Abstract, Article Title, and Keywords. It also seeks to recommend a contextual classifier-based recommender system to help academics identify credible sources. Scholarly articles from various study fields often use similar terms in their titles and keywords. However, finding a publication venue can be challenging for researchers at the beginning of a scientific inquiry. Thus, it is crucial to classify information based on its context, especially when abstracts, keywords, and titles receive equal attention. Materials and Methods: An ensembled model was developed and trained using 114K instances from 38 classes of the Web of Science (WoS) dataset and 40 classes of the Dimensions dataset. The ensemble approach incorporated both machine learning and deep learning algorithms to build a diverse classifier. The model was evaluated by testing it with an 80:20 train-test split to assess performance. The classifier was further integrated into a recommender system designed to suggest probable publication sources based on given article information. Results: The ensemble classification approach demonstrated superior performance with faster inference and efficient training time. The balanced training model, tested on 114K instances, effectively categorized scholarly articles into one of 40 categories. The recommender system was capable of recommending up to 10 probable publication sources based on the article's Title, Keywords, and Abstract. Models utilizing abstractions yielded the best results and provided a better understanding of the context in every iteration of the experiment. Conclusion: This study successfully developed an ensemble-based contextual classifier for academic papers, which can also function as a recommender system. The system aids researchers in choosing the most appropriate sources to publish by categorizing articles into 40 categories and suggesting credible publication venues. This approach simplifies the decision-making process for academics, enabling them to identify relevant publications and suitable sources for their work more efficiently.

This research investigates the application of Adaptive Hierarchical Reinforcement Learning with Probabilistic Decision Networks (AHRL-PDN) within personalized learning systems. AHRL-PDN’s performance is meticulously compared with a wide range of dynamic learning path models, including Reinforcement Learning (RL), Bayesian Knowledge Tracing (BKT), Markov Decision Processes (MDPs), Neural Networks, Genetic Algorithms, Fuzzy Logic Systems, and Ant Colony Optimization (ACO). The study involves simulations with 50 students and 5 distinct learning objectives. It achieved the highest adaptability and scalability scores, as clearly illustrated in the accompanying comparative bar graph. Notably, AHRL-PDN demonstrated a 1.2-second response time and a $95 \%$ accuracy rate in dynamically adjusting learning paths, outperforming the other models in these key metrics. Additionally, AHRL-PDN exhibited remarkable robustness, maintaining its superior performance as the number of students was scaled up to 200 and the learning objectives expanded to 20. This scalability highlights the framework’s effectiveness in managing large-scale educational environments. These results underscore the significant potential of AHRL-PDN in enhancing personalized learning experiences by enabling rapid, real-time adaptation to individual student feedback, thereby optimizing educational outcomes.

Social media has transformed the way we interact with each other in the twenty-first century. As it continues to evolve and shape our lives, it is important to consider both the benefits and the potential risks of these platforms. One of these risks is the spread of misinformation over global platforms like Twitter. The rapid dissemination of false information can have far-reaching consequences, influencing public opinion and potentially leading to real-world harm. One critical area of concern is the impact of fake news on mental health and well-being. This study aims to explore the emotional responses elicited by fake news and provide insights into the psychological effects of false information circulated through social media. To achieve this, a novel Ensemble Hard Voting Classifier (EHVC), which accurately classifies tweets as either fake or real with an accuracy of 93.84

This paper examines the convergence of cloud computing, facts science, and facts engineering, providing a primer for college kids getting into those fields. The examine highlights the synergistic courting among those domains, displaying how cloud infrastructure complements collaboration and efficiency. By illuminating those interconnections, the paper presents college students with a holistic view of the current facts landscape, making ready them to leverage current equipment withinside the cloud era.

Altmetrics, or alternative metrics, refer to the newer kind of events around scholarly articles, such as the number of times the article is read, tweeted, mentioned in blog posts etc. These metrics have gained a lot of popularity during last few years and are now being collected and used in several ways, ranging from early measure of article impact to a potential indicator of societal relevance of research. However, there are several studies which have cautioned about use of altmetrics on account of quality and reliability of altmetric data, as they may be more prone to manipulations and artificial inflations. This study proposes a framework based on application of Benford’s Law to evaluate the quality of altmetric data. A large sized altmetric data sample is considered and the fits with Benford’s Law are computed. The analysis is performed by doing plots of the empirical data distributions and the theoretical Benford's, and by employing relevant statistical measures and tests. Results for fit on first and second leading digit of altmetric data show conformity to Benford's distribution. To further explore the usefulness of the framework, the altmetric data is subjected to artificial manipulations through a systematic process and the fits to Benford’s law are reassessed to see if there are distortions. The results and analysis suggest that Benford’s Law based framework can be used to test the quality of altmetric data. Relevant implications of the research are discussed.

The Smart Plant Watering System is an Internet of Things (IoT) solution for automated plant irrigation. It combines an ESP8266 Wi-Fi module for real-time monitoring with soil moisture and DHT11 sensors for data collection. It makes use of tools such as Blynk and Thing Speak to facilitate data visualization and remote control. An online crop recommendation systems using ML approaches and written in Python makes recommendations for the best crops to plant depending on the analysis of sensor data. With its ability to maximize irrigation, reduce water use, and improve crop management, this system represents a major breakthrough in precision agriculture for increased productivity and environmental sustainability.