Department of Computer Engineering

Permanent URI for this collection

https://repository.run.edu.ng/handle/123456789/2323

Browse

Now showing 1 - 20 of 29

A Comprehensive Analysis of COVID-19 Spread in Nigeria
(Baze Universityc, 2021) Omolayo Abegunde
The COVID-19 pandemic emanated from China was not only unexpected by the rest of the world, but it also resulted in an economic downturn. In Nigeria, attempts have been made at different levels of government to combat the virus' spread, with some promising results. In this paper, we looked at the impact of the spread from February 29 to December 27, 2020 to see what it was like. The findings were based on data from reported cases, deaths, recoveries, and active cases. The data was preprocessed and feature engineering was performed by adding new features (active, days, month). The pandas library were used to analyse the two sets of data. The results of the analysis provide us with a description of the COVID-19 pandemic's spread in Nigeria and the need to slow it down even further.
A SIMULATION-BASED COMPARATIVE STUDY OF CONTROLLED DELAY (CODEL) WITH RANDOM EARLY DETECTION (RED) FOR NETWORK PERFORMANCE EVALUATION
(British Journal of Computer, Networking and Information Technolog, 2023) KOMOLAFE Temitope Amos
The rising need to use the internet for time/delaysensitive applications with different Quality of Service (QOS) requirements has made network management and control even more challenging. The current congestion avoidance and control mechanisms for Transport Control Protocol (TCP) are insufficient to offer good service in all circumstances. A few decades ago, the TCP successfully regulated Internet congestion control. However, it is already widely acknowledged that TCP has reached its limits and that new congestion control protocols will be required in the near future. This has prompted a significant amount of research on novel congestion control designs that will meet the demands of the future Internet. With widespread public attention and study, the full buffer problem has not gone away, but rather worsened. As a result, there has been a surge in interest in using Active Queue Management (AQM) in Internet routers to minimize queue latency. The effectiveness of a recently developed AQM, Controlled Delay (CoDel) algorithm, designed to work in today’s network setups and can be deployed as a main part of the bufferbloat solution, is evaluated in this research study. CoDel's effectiveness is evaluated by running simulations in ns-3 and comparing its results to that of Random Early Detection (RED), another promising network queue management technique
An Automatic Door Lock Security System Based on Convolutional Neural Network
(Dutse Journal of Pure and Applied Sciences (DUJOPAS),, 2023) Janet O. Jooda
Door lock security is important because it prevents intruders or unauthorized individuals from entering our homes or offices. Previous door lock security systems based on password, radio frequency identification and facial recognition are unreliable. Therefore, this project developed an automatic door lock security system using Convolutional Neural Network. Fifty faces of home owners were captured and trained using CNN. The system was tested at different distances and lightning condition and evaluated using accuracy, precision and f1 score. Results show that the developed system is 91.67% accurate. However, it is recommended that future work considers increasing dataset for model training to obtain more accurate results.
Authorship Veriﬁcation of Yorùbá Blog Posts using Character N-grams
(ICMCECS (IEEE), 2020) Omolayo Abegunde
The task of determining whether a pair (or more) documents were written by the same author comes under authorship veriﬁcation. N-grams are sequences of elements appearing in texts; they can be words, POS tags, characters, or some other elements that can be encountered one after another in texts. The tasks in authorship veriﬁcation were more challenging as it focused on whether the target author and the text to be used have a closely related style. In this paper, an authorship veriﬁcation task on Yorùbá blog posts is hereby presented. N-grams features were extracted from the corpus, and inductive learning techniques were applied to build feature-based models in order to perform the automatic author identiﬁcation. The K-means clustering algorithm was used in the study since the supervised algorithm cannot be applied to the one-class classiﬁcation of the dataset. The evaluation was done with the Silhouette Coefﬁcient algorithm, which is used to evaluate unlabeled data. The result obtained is positive, which indicates the data points have a strong relationship with the dataset. The obtained result signiﬁes a yes relationship between the posts. This signiﬁes that the posts were from the same author.
Automatic Plagiarism Detection Using Fuzzy-Logic
(Dutse Journal of Pure and Applied Sciences (DUJOPAS),, 2023) Janet O. Jooda
Plagiarism occurs when a researcher copies fellow researcher’s work verbatim without acknowledging the author. This work developed an automatic plagiarism detector using fuzzy logic. The system developed was tested with 4 different text documents and evaluated using portability, efficiency, functionality, ease of use and accuracy metrics. Results show that the developed plagiarism detector is very easy to use with high functionality and accuracy as well as moderate efficiency and portability based on user’s assessment. However, future work can increase the data size for model training and consider machine learning techniques to improve accuracy.
Comparative Analysis of Feature Level Fusion Bimodal Biometrics for Access Control
(International Journal of Progressive Sciences and Technologies,, 2021) Janet O. Jooda
The increasing quest for dependable, robust and secure recognition systems led to combining two or more biometric modalities for improved performance of a biometric system. Bimodal biometric systems have proven to achieve obvious advantages over unimodal systems in various applications such as access control, surveillance, forensics, deduplication and border control etc. In this study, a comparative analysis of combination of three biometric trait at feature level of fusion was carried out. Face, fingerprint and iris images were acquired from LAUTECH biometric database. The bimodal setup consists of face-iris, face-fingerprint and iris-finger modalities. Principal Component Analysis (PCA) was employed for feature extraction, weighted sum technique was used to fuse the images at feature level while Support Vector Machine (SVM) was used for classification. Experimental result revealed that the bimodal biometric achieved an improved performance than the unimodal biometric. The performances of the bimodal systems indicated that combination of face and iris features achieved the best performance with FAR, FRR and accuracy of 0.00%, 1.42% and 99.00% at 38.15 seconds. Hence, a bimodal face-iris recognition system would produce a more reliable security surveillance system for access control than other combination compared in this study.
Design and Implementation of Mobile Information System for Federal Road Safety Corps (FRSC) of Nigeria
(International Journal of Sensor Networks and Data Communications, 2021) Omolayo Abegunde
With a daily increase in the use of mobile devices in the 21st century, handheld devices are fast reaching the unreached and information is now easily disseminated. Nigeria, as a developing nation in the western Africa needs to be all information technology compliant. Far from this, vehicles have been registered manually. This mobile information system is designed to aid the every member of the Nigeria community in building an information network with the Federal Road Safety Corps (FRSC). Motorists, drivers and others who had registered their vehicles manually would be able to register their vehicles number plates and report accident victims to the Corps with ease from their mobile devices. This work focuses mainly on the vehicle registration, issuing number plates and information dissemination to the Federal Road Safety Corps, Nigeria.
Design Issues in Sentiment Analysis for Yorùbá Written Text
(Ife Journal of Science and Technology, 2019) Omolayo Abegunde
Sentiment Analysis (SA) is an exciting and important field in Artificial Intelligence combining Human Language Processing, Machine Learning and Psychology. It is a means of understanding a user’s opinion about an event. The goal of SA is to get opinion expressed in implied text, targets of the opinion and reason for the opinion. Conversely, a great number of research efforts are dedicated to English language data, while a countless share of information is obtainable in other languages as well but none yet for Yorùbá. This work examines the design issues with respect to automating SA for standard Yorùbá language. The process of SA which includes data cleaning, data annotation etc. is highlighted. The structure of the Yorùbá text is described and a text corpus design for Yorùbá sentiment analysis system is presented. The outcome of this work provided suitable requirements for the design.
Development of a face recognition system using hybrid Genetic-principal component analysis
(1st International Conference on Electrical, Electronic, Computer Engineering & Allied Multidisciplinary Field, 2021-12) Ibikunle, Akinola
Humans have been using physical attributes such as face, voice gait and fingerprints to recognize each other for ages. With the recent technological advancement, face recognition is a branch of biometrics system which has received considerable interest because of its ease in collecting, analysing and recognising face images. It is a system which compares an unknown image against the trained images in a database in order to identify the image. It has a number of applications such as Automatic Teller Machine (ATM), credit card, physical access control, National Identity card and correctional facilities. It has been found to be one of the ways of controlling and reducing crime rate. The development and evaluation of the performance of a face recognition system using hybrid Genetic- principal component Analysis technique is presented. The system consists of three major subsystems. Initial preprocessing procedures are applied on the input face images selected from the ORL Database. Consequently, face features are extracted from the processed images by principal component analysis and finally face identification is carried out using Genetic algorithm. Image resolutions of 50 x 50, 70 x 70, 100 x 100 and 140 x 140 are used in training and testing the system. The identification rates obtained were 100%, 96.36%, 93.63% and 90.90% for 50 x 50, 70 x 70, 100 x 100 and 140 x 140 respectively. This experimental result revealed that the lower the resolution of the cropped images, the higher the number of the correctly identified face images. The reason is attributed to the fact that there is variation in the features considered for recognition for each resolution. Hence, this technique has been proved to be more robust and suitable for low resolution.
Development of a Modified Simulated Annealing to School Timetabling Problem
(International Journal of Applied information Systems, 2015) Janet O. Jooda
This work presents a modified simulated annealing applied to the process of solving a typical high school timetabling problem. Preparation of a high school timetable consists basically of fixing a sequence of meetings between teachers and students in a prefixed period of time in such a way that a certain set of constraints of various types is satisfied. The approach presented in the paper has been successfully used to schedule the first time school timetable of Fakunle Comprehensive High School, Osogbo Nigeria during the 2012/2013 session and it was capable of generating timetables for complex problem instances. A task involving 18 Classes, 45 Teachers and 15 Subjects for Junior Secondary School (JSS) with 3 Levels (JSS 1 to JSS 3), and 6 arms each; and 24 Classes, 77 Teachers and 19 Subjects for Senior Secondary School (SSS), with 3 Levels (SSS 1 to SSS 3), and 8 arms (3 for Science Group, 3 for Commercial Group, and 2 for Art Group), for 6 hours, 5 days respectively. The use of the implemented model resulted in significant time saving in the scheduling of the timetables, and a well spread lessons for the teachers. Also none of the teachers and classes was double booked. It was clearly evident that the developed modified simulated annealing reduces the major weakness of slow convergence (convergence at excessive time) associated with the classical simulated annealing.
DEVELOPMENT OF A PREDICTIVE FUZZY LOGIC MODEL FOR MONITORING THE RISK OF SEXUALLY TRANSMITTED DISEASES (STD) IN FEMALE HUMAN
(International Research Journal of Engineering and Technology, 2020) Janet O. Jooda
The purpose of this study is to develop a classification model for monitoring the risk of sexually transmitted diseases (STDs) among females using information about non-invasive risk factors. The specific research objectives are to identify the risk factors that are associated with the risk of STDs; formulate the classification model; and simulate the model. Structured interview with expert physicians was done in order to identify the risk factors that are associated with the risk of STDs Nigeria following which relevant data was collected. Fuzzy Triangular Membership functions was used to map labels of the input risk factors and output STDs risk of the classification model identified to associated linguistic variables. The inference engine of the classification model was formulated using IF-THEN rules to associate the labels of the input risk factors to their respective risk of still birth. The model was simulated using the fuzzy logic toolbox available in the MATLAB® R2015a Simulation Software. The results showed that 9 non-invasive risk factors were associated with the risk of STDs among female patients in Nigeria. The risk factors identified were marital status, socio- economic status, toilet facility used, age at first sexual intercourse, practice sex protection, sexual activity (in last 2 weeks), lifetime partners, practice casual sex and history of STDs. 2, 3 and 4 triangular membership functions were appropriate for the formulation of the linguistic variables of the factors while the target risk was formulated using four triangular membership functions for the linguistic variables no risk, low risk, moderate risk and high risk. The 2304 inferred rules were formulated using IF-THEN statements which adopted the values of the factors as antecedent and the STDs as consequent part of each rule. This study concluded that using information about the risk factors that are associated with the risk of STDs, fuzzy logic modeling was adopted for predicting the risk of STD based on knowledge about the risk factors.
Development of an Enhanced Convolutional Neural Network for Self-proctoring in Online Examination Systems
(IEEE 5th International Conference on Electro-Computing Technologies for Humanity (NIGERCON)2024, 2024) Janet O. Jooda
Self- proctoring systems leverage on artificial intelligence algorithms to detect abnormal behaviour during examination, and the Convolutional Neural Network (CNN) has been proven to be one of the most preferred deep learning algorithms to serve this purpose. Convolutional neural networks (CNN) are good for human activity recognition, detection and classification purposes, however, they require a lot of training data which can slow down execution and increase computational time during classification. Therefore, this research addressed the aforementioned problem by developing an enhanced convolutional neural network for online examination self-proctoring systems. The technique was achieved by selecting optimal weight values in the Convolutional network with a Gravitational Search Algorithm optimizer which ultimately improved the model’s computation time. The result of the CNN-GSA technique achieved an F1-Score of 91%, Accuracy of 83.33%, Precision of 83% and a recall of 99.98% at 0.657 seconds.
Development of Turbo Code Error Detection and Correction Scheme for Wireless Telemedicine Video Transmission
(LAUTECH Journal of Computing and Informatics (LAUJCI), 2021-09) OLAYINKA OYERONKE OYEFUNKE
Transmission of medical images and videos to a distant location over a wireless network for a proper diagnosis of the patient is a core aspect of telemedicine. During data transmission over the wireless communication channel, noise and other impairments are introduced into data and this causes error in the transmitted data. Hence, there is a need for a method to detect and correct error which may lead to an erroneous diagnosis. Turbo codes happens to be the earliest error-correcting codes with the intention of establishing a dependable communications near the channel capacity with basically possible hardware. It has an excellent error correcting capabilities, which make it appropriate for many internet communications technology. This paper has come up with an effective and efficient method to detect and correct error encountered during the transmission of telemedicine video over the wireless channel with the use of a parallel concatenated Turbo code error detection and correction scheme. A MATLAB Simulation was carried out to investigate and demonstrate the performance of the proposed system. The performance of the developed system was taken at different ranges of SNR with BER, PSNR, MSE and the processing time of the decoders. The developed system was compared with when turbo code is not applied during transmission. The results of the simulation shows a better performance when turbo code was applied compared to when it was not applied
Facial Emotion Recognition And Detection Using Convolutional Neural Network
(Journal of Computing, Science &Technology, 2024) Janet O. Jooda
Human emotions have important role in communication especially to understand the emotions of those with speech problems. Various facial emotion recognition and detection systems have been developed but most of these systems have difficulty in performing a muti-class classification and yielded lower accuracy. Therefore, this research employed convolutional neural network for recognition and detection of four basic emotions: happy, sad, angry, neutral. The dataset training the convolutional neural network model was obtained human emotion, accuracy, machine locally and it include about 133 images. Results show that the system developed performed well with an learning, communication, detection accuracy of 0.9533, precision of 0.97, F1-score of 0.94 and recall of 0.93. The approach used showed a significant improvement over traditional machine learning methods and be a useful tool for those with speech problems and visually to predict human emotion.
Feature Fusion Using GSA for Multi Instance Authentication System
(Asian Research Journal of Current Science, 2023-11-15) Ibikunle, Akinola
Multi-instance fusion of fingerprint authentication system at score level overcomes a few of the shortcomings of a Unimodal Biometric System (UBS) and enhanced the efficiency of the system. However, due to loss of information at higher levels, the features fused at the score level are confined in comparison to feature level fusion and could lead to poor performance. In this study, multi-instance fusion of fingerprints was done at feature level using Gravitational Search Algorithm (GSA) to select and combine minimal relevant informative texture features subsets from multi-instances of fingerprint and considerably improves the performance of the system. The approach was validated by creation of multi-instances of fingerprint database acquired locally from 150 subjects in an uncontrolled environment and texture based feature extraction was considered and classification of fused texture feature was done using back propagation neural network. The results show that the presented technique was effective in subject authentication with accuracy of 97.09%, indicating that it can successfully secure fingerprint authentication systems from unauthorized attacks.
Feature Fusion Using GSA for Multi- Instance Authentication System
(Asian Research Journal of Current Science, 2023) Janet O. Jooda
Multi-instance fusion of fingerprint authentication system at score level overcomes a few of the shortcomings of a Unimodal Biometric System (UBS) and enhanced the efficiency of the system. However, due to loss of information at higher levels, the features fused at the score level are confined in comparison to feature level fusion and could lead to poor performance. In this study, multi-instance fusion of fingerprints was done at feature level using Gravitational Search Algorithm (GSA) to select and combine minimal relevant informative texture features subsets from multi- instances of fingerprint and considerably improves the performance of the system. The approach was validated by creation of multi-instances of fingerprint database acquired locally from 150 subjects in an uncontrolled environment and texture based feature extraction was considered and classification of fused texture feature was done using back propagation neural network. The results show that the presented technique was effective in subject authentication with accuracy of 97.09%, indicating that it can successfully secure fingerprint authentication systems from unauthorized attacks.
Fingerprint Intramodal Biometric System Based on ABC Feature Fusion
(Asian Journal of Research in Computer Science, 2021-08-13) Jooda, Janet
Unimodal biometrics system (UBS) drawbacks include noisy data, intra-class variance, inter-class similarities, non-universality, which all affect the system's classification performance. Intramodal fingerprint fusion can overcome the limitations imposed by UBS when features are fused at the feature level as it is a good approach to boost the performance of the biometric system. However, feature level fusion leads to high dimensionality of feature space which can be overcame by Feature Selection (FS). FS improves the performance of classification by selecting only relevant and useful information from extracted feature sets being an optimization problem. Artificial Bee Colony (ABC) is an optimizing algorithm that has been frequently used in solving FS problems because of its simple concept, use of few control parameters, easy implementation and good exploration characteristics. ABC was proposed for optimized feature selection prior to the classification of Fingerprint Intramodal Biometric System (FIBS). Performance evaluation of ABCbased FIBS showed the system had a Sensitivity of 97.69% and RA of 96.76%. The developed ABC optimized feature selection reduced the high dimensionality of features space prior to classification tasks thereby increasing sensitivity and recognition accuracy of FIBS.
Improving millimetre-wave path loss estimation using automated hyperparameter-tuned stacking ensemble regression machine learning
(Results in Engineering, 2024) Janet O. Jooda
Path loss prediction is a crucial aspect of designing and operating wireless communication systems, especially in the millimetre-waves (mmWaves) frequency bands. However, these bands are associated with climate-related challenges: rain attenuation, and free space path loss. To address these challenges, an advanced stacking ensemble-regression machine learning (SEML) model with automated hyperparameter tuning (AHT) was pro posed. The AHT-SEML model leverages multiple base regressors integrated with a meta-regressor. The model’s performance was optimised using the AHT tuning technique. The AHT-SEML model’s efficiency was tested using simulated path loss data from a Composite 3D Raytracing-Image-Method propagation model across four sub- Saharan cities, at mmWaves frequencies. The AHT-SEML model’s performance was compared to three empir ical path loss models, namely Close-In (CI), Floating Intercept (FI), and Alpha-Beta-Gamma (ABG), using eval uation metrics such as Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). AHT-SEML outperformed other models in the four cities across all frequencies and scenarios with the highest Index of Agreement and lowest Bayesian information criterion. Model confidence set (MCS) analysis with CI benchmark indicates that all the models except AHT-SEML performed below the critical t-value of 2.3530 at 95% confidence level with a degree of freedom of 3, implying no significant differences in their MAEs compared to the CI. However, AHT-SEML’s t-statistic values exceed this critical t-value, indicating statistically significant differences and better performance than the CI benchmark models. Similarly, F-statistics of 29.45 and 26.54 correspond to p- values of 1.91 × 10− 14 and 2.50 × 10− 13 for MAE and RMSE, respectively, corroborating significant differences in the AHT-SEML’s performance.
Influence of Eigenvector on Selected Facial Biometric Identification Strategies
(World Journal of Engineering Research and Technology, 2020-02-16) Jooda, Janet
Face identification strategies are becoming more popular among biometric-based strategies as it measures an individual‟s natural data to authenticate and identify individuals by analyzing their physical characteristics. For face identification system to be efficient and robust to serve it purpose of security, there is need to use the best strategy out of the many strategies that have been proposed in literatures for face identification. Amidst the most popularly used face identification strategies, Principal Component Analysis PCA, Binary Principal Component Analysis BPCA, and Principal Component Analysis – Artificial Neural Network PCA-ANN were selected for performance evaluation. The research was experimented by varying the eigenvector of the training images for each strategy to compare the performance using Recognition Rate RR and Total Recognition Time TR as performance metrics. Results showed that PCA – ANN strategy gave the best recognition rate of 94% with a trade-off in recognition time. Also, the recognition rates of PCA and B-PCA increased with decreasing number of eigenvectors but PCA-ANN recognition rate was negligible. Hence PCA-ANN outperforms the other face identification strategies.
Long-Short-Term Memory Model for Fake News Detection in Nigeria
(Ianna Journal of Interdisciplinary Studies, 2023) Janet O. Jooda
Background: The advent of technology allows information to be passed through the Internet at a breakneck speed and enables the involvement of many individuals in the use of different social media platforms. Propagation of fake news through the Internet has become rampant due to digitalisation, and the spread of fake news can cause irreparable damage to the victims. The conventional approach to fake news detection is time-consuming, hence introducing fake news detection systems. Existing fake news detection systems have yielded low accuracy and are unsuitable in Nigeria. Objective: This research aims to design and implement a framework for fake news detection using the Long-Short Term Memory (LSTM) model. Methodology: The dataset for the model was obtained from Nigerian dailies and Kaggle and pre-processed by removing punctuation marks and stop words, stemming, tokenisation and one hot representation. Feature extraction was done on the datasets to remove outliers. The locally acquired dataset from Nigeria was balanced using Synthetic Minority Oversampling Techniques (SMOTE) Long-Short Term Memory (LSTM), a variant of Recurrent Neural Network (RNN)- which solved the problem of losing gained knowledge and information over a long period faced by RNN- was used as the detection model This model was implemented using Python 3.9. The model detected fake news by classifying real and fake news approaches. The dataset was fed into the model, and the model classified them as either fake or real news by processing the dataset through input and hidden layers of varying numbers of neurons. accuracy F1 score and detection time were used as the evaluation metrics. The results were then compared to some selected machine learning models and a hybrid of convolutional neural networks and long short-term memory models (CNN-LSTM). Results: The result shows that the LSTM model on a balanced dataset performed best as the two news classes were accurately classified, giving an average detection accuracy of 92.86%, which took the model 0.42 seconds to detect whether news was real or fake. Also, 87.50% average detection accuracy was obtained from an imbalanced dataset. Compared to other machine learning models, SVM and CNN-LSTM gave 81.25% accuracy for imbalanced datasets and 82.14% and 78.57% for balanced datasets, respectively. Conclusion: The outcome of this research shows that the deep learning approach outperformed some machine learning models for fake news detection in terms of performance accuracy.

Browse

Browsing Department of Computer Engineering by Title

Results Per Page

Sort Options