VFAST - Virtual Foundation for Advancement of Science and Technology (Pakistan)
Not a member yet
1255 research outputs found
Sort by
Artificial intelligence based real-time smoke and fire detection and security management algorithms
Full text linkSmart fire detection is essential for people’s safety and property. The effective utilization of innovative technologies provides fast fire detection before intensification. Automatic fire systems commonly utilize passive sensors that are damaged by sunlight and environmental conditions. To address this problem, this study provides AI-based fire and smoke detection system that uses a You Only Look Once (YOLO) smart object detection algorithm integrated with a deep learning convolutional neural network architecture(CNN) and Android Studio to achieve the desired requirements. This prototype uses Common Objects in Context (COCO) datasheets for YOLO modelling. The incorporated camera continuously monitors the consumer for immediate notification. The system uses Android applications to monitor the parameters. The application architecture uses the Django framework to communicate the developed system with the Android application and the YOLO model. The Android application was designed using Android studio software to provide online information via cloud-based systems. Compared with conventional fire detection systems that consist of heat, flame, gas, and smoke sensors with high power consumption, installation, and preventive maintenance. The designed system considers AI fire detection algorithms using images and video forms. Further advancements in this state-of-the-art technology can improve the industrial application of early fire detection
Data to Diagnosis: Evaluating Machine Learning Algorithms for Predictive Healthcare in Diabetes
Full text linkDiabetes mellitus, a chronic metabolic disease, presents alarming challenges to world health. It is vital to diagnose it early to prevent serious complications. In this research, eight machine learning algorithms—SVM, XGBoost, Naive Bayes, Logistic Regression, Gradient Boosting, KNN, Decision Tree, and Random Forest—are used on a formatted dataset with clinical and demographic attributes. Normalization and categorical encoding were done for preprocessing. Although no class-balancing methods (e.g., SMOTE or weighting) were used or hyperparameter tuning was performed, models were tested with accuracy, precision, recall, F1-score, and confusion matrices. Interestingly, the dataset is very imbalanced (~10% diabetic cases), and thus may influence sensitivity. Ensemble models, particularly Gradient Boosting and XGBoost, reported more than 91% accuracy. In spite of limitations, findings suggest the promise of ML in early prediction of diabetes
Enhancing Interpretability in Anxiety Detection on Reddit: A Machine Learning Approach with LIME and Topic Modeling
Full text linkIn modern society, mental disorders, particularly anxiety, are becoming more and more prevalent concerns. Individuals express their opinions and feelings on social media platforms like Reddit which offers valuable information for understanding mental health. This study applies BERTopic and Local Interpretable Model-agnostic Explanations (LIME) to demonstrate the interpretation of machine learning models in anxiety detection. To analyze and identify the linguistic patterns, a novel dataset has been collected from Reddit communities utilizing multiple subreddits pertaining to anxiety and casual conversations. For topic modeling BERTopic was used to discover key topics in discussions. In addition, TF-IDF features were used to train a Random Forest Classifier, which obtained an accuracy of 88% in classifying the post between anxiety and non-anxiety. Furthermore, to ensure transparency in model decision making process, LIME was used to examine textual features that influence models. This study emphasizes the importance of explainability with regards to AI-assisted mental health solutions while also demonstrating the usefulness of social media data in analyzing how anxiety is articulated, and language is employed differently
Complexity Analysis of LLM-Generated Recursive Code: A Systematic Evaluation
Full text linkProgramming is an essential skill, but it can be difficult for beginners, especially when it comes to logical concepts like recursion. Despite the development of many computational and pedagogical methods to simplify programming, recursion remains a challenging topic to understand, implement, and debug. Artificial intelligence has led to the development of large language models (LLMs), such as ChatGPT, Gemini, and DeepSeek that can generate programming source code. Various studies have analyzed the quality of code produced by LLM. However, the complexity of the recursive code generated by these models has not been studied. This study compared and analyzed recursive Python programs generated by Gemini (2.5 Pro), DeepSeek (V3.1) and ChatGPT (GPT-5) in an attempt to fill this gap. For the study, 250 programs generated by each model were examined using Halsted and cyclomatic complexity metrics. The results showed that ChatGPT produced less complex code, indicating easier recursion, while DeepSeek produced more complex programs due to higher Halstead and cyclomatic complexity scores. Gemini programs have a medium level of difficulty. The Kruskal-Wallis test was used to further analyze the data, and it revealed significant differences between the recursive code generated by ChatGPT, DeepSeek, and Gemini. Overall, the study found that each LLM has a distinct pattern: ChatGPT emphasizes simplicity, Gemini takes a balanced approach, and DeepSeek\u27s generated code promotes clarity but suffers from complexity. More comprehensive analysis will be conducted in the future by expanding the dataset and including larger language models
Public Perception of Chinese Language Education in Saudi Arabia: A Keyword-Enhanced Aspect-Based Sentiment Analysis of Social Media Discourse
Full text linkThe increasing popularity of Chinese as a second language in Saudi Arabia offers a unique chance to study and explore public perceptions and opinions through computational methods. Natural Language Processing (NLP) is one such method, widely adopted in research for this type of analysis. It offers techniques to extract insights from large volumes of unstructured data. However, sentiment analysis on multilingual, culturally specific online discourse remains a challenging task in NLP. The aim of this research work is to address the problem of accurately detecting sentiments and topics in contexts influenced by culture; we discuss the adoption of Chinese language education in Saudi Arabia. For this, we implemented a transformer-based sentiment analysis model on a custom domain-specific dataset with LDA for topic modeling. In this way, we identified key thematic clusters related to globalization, education, and cultural exchange. The research results indicate that topics associated with globalization carry the most positive sentiment, reflecting optimistic public attitudes toward linguistic expansion. This work contributes to the field of applied NLP by demonstrating the feasibility of sentiment and topic modeling in low-resource, culturally diverse environments and contexts
Establishing Neural Network Models for Predicting Flood Propagation and Recession in Urban Roads
Full text linkIn this analysis, a contagion model is a straightforward yet effective mathematical approach—that was used to forecast the temporal change of the outset and contiguous distribution and recession of flooding in metropolitan roadway networks. A system of metropolitan roadways must be flood-resistant in order to provide public services and deal with emergencies. The dispersion of floodwaters is a complicated temporal-spatial process that affects urban networks. In comparison to the SEIR (Susceptible-Exposed-Infected-Recovered) prototype, a system of ordinary differential equations, four macroscopic characteristics, rate of including flood spreading represented by (), rate of flood incubation symbolized by (), and rate of recovery highlighted by (), can be used to understand how floods evolve within networks. Additionally, by joining the backpropagated neural network and the Levenberg—Marquardt algorithm (NN-BLMA), surrogate solutions to the model are discovered. This method has some clear advantages over conventional ones, including flexibility, comparatively simple implementation, and fastest results. Reference solutions are generated using the Runge-Kutta of order four (RK4) method. We have examined three distinct scenarios to analyze our surrogate solution models. By changing , , , and k, the mathematical model\u27s stability and equilibrium are examined. To gauge the validity of our machine learning process, we categorize our candidate solutions into training, testing, and experimental class. The efficacy of the NN-BLMA scheme has been confirmed by comparative examinations of statistical values based on mean squared error function (MSEF), effectiveness, regression plots, and failure histograms
Pakistan\u27s response to the Afghan war (1979-1988): In the Neo-classical Realism perspective
Full text linkMost researchers condemn the fact that the foreign policy of Pakistan was not befitting during the war in Afghanistan since its consequences outweighed the gains to the country during the period. This is one of the criticisms that are repeatedly emphasized and strengthened on the foreign policy debate, as the war in Afghanistan made a lasting influence on the politics, economy and security of Pakistan. But little focus is given in terms of the examination of the Pakistani reaction to the Afghan war in the view of Neo-classical Realism. Since the world is anarchic, all the states have to act on their national interests, and Pakistan as any other state had to react with the system of anarchy. A nation can further be seen to be in terms of the international relations theory, since theories offer an analysis model to explain judgments that might seem conflicting or incongruent.Among others, there exist two international relations theories Realism and Liberalism, which are more appropriate to examine the behavior of a nation under pressure. Realism focuses on power, security, and survival whereas Liberalism focuses on cooperation, institutions and interdependence. But the Neo-classical Realism is of the opinion that internal and external factors should be combined to comprehend the behaviour of states. The external influences include the force of balance, alliances, and world pressures which interacts with internal influences of leadership, political constraints, economic conditions, and domestic pressures.The paper will thus seek to discuss the internal and external forces that forced Pakistan to take a position in the Afghan war. By concentrating on Neo-classical Realism, this paper will indicate why Pakistan had to do what it did, how its leadership had to juggle between national interests and international demands, as well as why their foreign policy decisions were influenced by a set of national and global realities
A Fault Prognostic System for the Turbine Guide Bearings of a Hydropower Plant Using Long-Short Term Memory (LSTM)
Full text linkHydroelectricity, being a renewable source of energy, globally fulfills the electricity demand. Hence, Hydropower Plants (HPPs) have always been in the limelight of research. The fast-paced technological advancement is enabling us to develop state-of-the-art power generation machines. This has not only resulted in improved turbine efficiency but has also increased the complexity of these systems. In lieu thereof, efficient Operation & Maintenance (O&M) of such intricate power generation systems has become a more challenging task. Therefore, there has been a shift from conventional reactive approaches to more intelligent predictive approaches in maintaining the HPPs. The research is therefore targeted at developing an artificially intelligent fault prognostics system for the turbine bearings of an HPP. The proposed method utilizes the Long Short-Term Memory (LSTM) algorithm in developing the model. Initially, the model is trained and tested by bearing vibration data from a test rig. Subsequently, it is further trained and tested with realistic bearing vibration data obtained from an HPP operating in Pakistan via the Supervisory Control and Data Acquisition (SCADA) system. The model demonstrates highly effective predictions of bearing vibration values, achieving a remarkably low RMSE
Covid-19 Sentiment Analysis on X (formerly Twitter) Using Machine Learning Classifiers: Performance Comparison and Key Insights
Full text linkThe current generation and widely used platforms like X (formerly Twitter) enable the study of public attitudes toward important topics, including the COVID-19 outbreak. In this paper, machine learning approaches (ML) are employed to build a sentiment analysis system for COVID-19 hashtagged tweets. We employed four ML classifiers, namely Naïve Bayes (NB), Support Vector Machine (SVM), Decision Tree (DT), and Random Forest (RF), to classify the tweets into positive, negative, and neutral sentiments. In total, the examined dataset includes 178,240 tweets that are related to COVID-19 and were preprocessed through natural language processing. To assess the performance of the classifiers, we used accuracy, precision and recall, and F1 score. The results show that the DT classifier has the highest accuracy of 94% when compared to other models concerning precision and recall. Undersampling and oversampling were the techniques examined for addressing the issue of class imbalance. Such findings imply that ML, especially the SVM and DTs, can be useful in the next large-scale public sentiment analysis during a pandemic. Among the recommendations for further enhancements of the sentiment analysis approaches and their use in monitoring people’s reactions to social media during the pandemic are included in the paper
A Decision Tree Based Approach for Pashto Coreference Resolution: The Case of Person Name Aliases
Full text linkCoreference resolution is an important problem in fields such as natural language understanding, natural language generation, named entity recognition, text summarization, and anaphora resolution. Determining whether or not two proper nouns are aliases of each other (i.e. aliases identification) is a classification problem. A binary classifier for alias identification is needed which returns “Yes” if the two input nouns are aliases and “No” otherwise. In this research paper, a binary decision tree based classifier is proposed that is augmented with cosine similarity measure for personal name aliases identification in Pashto. This classifier is trained on aliases records containing features’ vectors. A total of 10000 proper nouns’ pairs examples from the Pashto corpus have been extracted and a collection of crawled Pashto text, and recorded their features in this work. This resulted in 10000 example records, having 12 attributes. The selected dataset contains examples from different genres of the corpus e.g. novels, dramas, news, sports, letters and essays. These examples contain 5000 positive instances (i.e. class “Yes”) and 5000 negative instances (i.e. class “No”). These records are divided into two parts: the training part and the testing part in the ratio of 7:3. The 7000 examples of training part are used to induct the decision tree. This decision tree is created using Rapidminer, which is a data mining tool. Then, first order logic rules are created from the decision tree. These rules are then transformed into an algorithm, which is implemented in programming language Python. These rules are tested on the testing part of examples, which contain 3000 labeled examples. A total of 2794 out of these 3000 examples are classified correctly, which means an accuracy of approximately 93%. The error analysis of the 7% classification errors is performed to improve the system in future