Computer science

Permanent URI for this collectionhttps://repository.chuka.ac.ke/handle/chuka/15557

Browse

Now showing 1 - 4 of 4

SECURE CLOUD BASED APPROACH FOR MOBILE DEVICES USER DATA
(Chuka University, 2022-02) Mbae, Oscar
ABSTRACT In this era characterized by rapid technological innovations, mobile devices such as tablets and smartphones have become inevitable due to the variety of services they offer. As a result, computing capacity as well as storage needs of these devices are increasing tremendously. To ensure users continue to enjoy the portability, flexibility and accessibility that these devices continue to provide, there is need for a secure and user friendly data storage solution. However, despite the benefits of this technology, there are increased risks to the information that is accessible from the mobile devices. The main problem is the risk of private and confidential data being exposed to unauthorized persons and the risk of permanent loss or damage of that data. These problems are escalated by the fact that most information is stored in the devices’ internal memory, making them easily accessible. Mobile devices are susceptible to loss and the pins and patterns used as security controls are easy to by-pass because they have minimal encryptions. In the event of human error whereby the user forgets to delete downloaded confidential content from cloud-based platforms, it remains in the mobile device from where it can be accessed easily. This creates a need for secure ways of storing data in a cost-effective and convenient manner. The objective of this study was to design, implement and validate a secure cloud based approach for mobile devices’ user data. The study adopted design and development methodology which followed the entire design and development process from analyses to evaluation. From this methodology, the research employed the strategy of mixed method using a systematic process of collecting data, at first during prototype and then throughout the rest of study. This method allowed for continued development and implementation of the product. The solution was prototyped in an android based environment and developed using Java programming language together with MySQL for the database. The mobile data privacy solution proposed by this study provides a security solution to users to be able to store sensitive data and access it on their mobile devices. The solution focuses on securing the data on the mobile devices by storing it in an encrypted format and uploading it to the cloud. In addition, the downloaded data is timed to self-destruct after user consumption, eliminating unauthorized person or application from reading the information without a decryption key. The developed solution was able to provide security for the users’ confidential data while making it available. The tool enabled its users to store the selected sensitive files from their mobile devices in an encrypted format. To achieve this, we used the algorithm AES 256 to encrypt the data with a key only known to the user and upload it to the cloud for secure storage. The secure mobile tool was developed and fulfilled the requirements specification successfully. It met all the security parameters stated hence optimizing mobile user data storage security.
A CONVOLUTIONAL NEURAL NETWORKS AND SUPPORT VECTOR MACHINES HYBRID MODEL FOR NUMBER PLATE RECOGNITION
(Chuka University, 2022-03) Kibaara, Peter
ABSTRACT Automatic Number Plate Recognition (ANPR) systems are applied in many fields such as automatic electronic toll collection, car park management and access control, logistics and vehicle tracking, traffic law enforcement and crime resolution amongst others. Motion blur, plate orientation, lighting changes and image noise severely lower the detection speed and recognition accuracy of these systems. The incorporation of machine learning algorithms in ANPRs has seen Convolutional Neural Network (CNN) being used to develop ANPR models with improved performance in license plate detection. CNNs are best suited for image data where the number of features is large such as license plate detection. This is attributed to their design architecture which enables them to perform feature extraction automatically. However, their speed of execution is slow as the model has to learn a lot of features. Support Vector Machine (SVM) is a supervised machine learning algorithm suitable for classification and regression problems with datasets that have a small number of features. It doesn’t scale up well for large datasets with many features. It has demonstrated high speed and accuracy when used for classification in small datasets such as character recognition. The final stage in ANPR is a character recognition phase and involves few features. These two algorithms have been deployed independently, however the concept of combining the two algorithms for ANPR models remains highly unexplored. The research therefore combines the two models (CNN and SVM) to come up with an efficient hybrid ANPR system with improved number plate recognition accuracy. The two models were developed using a deep cascade framework; a CNN with a SoftMax classifier and a hybrid CNN with a SVM classifier. The Universidade Federal do Paraná (UFPR-ALPR) dataset was used to train validate and test the models. Recognition accuracy, precision, recall and F1 score metrics were used to evaluate the model. The hybrid CNN-SVM model had a recognition accuracy of 91.25% against 89.07 % from the pure CNN model. The weighted average precision, recall, and F1-score of the hybrid CNN-SVM was 92%, 91% and 91% respectively, which was better compared to that of pure CNN. The hybrid model was tested for external validity using the Smart Sense Laboratory (SSIG) dataset. The hybrid CNN-SVM model had a recognition accuracy of 91% against 89 % from the pure CNN model. The weighted average precision, recall, and F1 score of the hybrid CNN-SVM was 91%, 91% and 91% respectively which was better compared to that of pure CNN, which had 90%, 89% and 89% respectively.
FACTORS AFFECTING ADOPTION OF MANURE ON BANANA YIELDS IN SMALLHOLDER FARMS IN MAARA SUB COUNTY, THARAKA NITHI COUNTY, KENYA
(Chuka University, 2022-09) Kimetto, Jerono
ABSTRACT Banana (Musa ssp) is a widely distributed fruit and which is contributing greatly towards food security in the developing countries. In Kenya, banana is among the major food and cash crop produced by smallholder farmers as a source of income and food security. However, it has been associated with low productivity and income. Currently, banana production in Kenya is 14 t/ha-1yr-1, which is below the global average of 20.5 t/ha-1 yr1. The low production has been attributed to various factors such as poor agronomic practices, low soil fertility, poor markets, lack of farming capital and sociodemographic factors. Despite studies on most of these factors, information on dynamics of socio-economic factors on banana farming by smallholder farmers is ever changing due to change in climate and lifestyles. Therefore, there is need to continuously monitor and study the effect of these factors on productivity and income of banana farming. The study, therefore determined the effect of manure adoption on banana production in smallholder farms in Maara Sub County, Tharaka-Nithi County. Descriptive research design was used in the study. The study targeted 34,779 smallholder banana farmers. A structured questionnaire, open and closed ended, were used to collect the data. Data was collected on demographic socio economic, institutional factors and banana production. Data obtained was analysed using the SPSS version 26. Descriptive statistics were employed in the presentation of the results. The study revealed that majority of the banana farmers are above 53 years and majority (54.8%) of the farmers derived their seedlings from their own suckers. About 83.8% of the farmers have adopted manure while 16.2% of the respondents have not. The common variety planted by the farmers is Kampala represented by 30.73%. About 29.7% banana farmers own land sizes between 0.5-1.0 ha of land. The majority (97.3%) of the farmers privately owned their lands while 2.7 3% of the respondents leased the lands for growing bananas. It was established that majority (88.0%) of the adopters received extension services while adopters who had not received any extension services were being represented by 8.0 %. Majority of the farmers who belonged to a farmer group and were adopters of manure being represented by 82.0% respondents while farmers who and were adopters and did not belong to any farmer group were represented by 14.0%. The study sought to determine the socioeconomic factors that affect adoption of manure in smallholder farms in Maara Sub County. Logistic regression model was used to find out whether gender, age, highest level of education of decision maker, access to extension service, participation in a farmer group, labour and land size were significant in the adoption of manure. Education level of decision maker p = 0.007, Gender p = 0.000, land size p = 0.000 and participation in farmer group p = 0.003 and extension services p = 0. 027 were the factors that were found to significantly affect adoption of manure. Multiple regression model was carried out to determine the impact of adoption of manure on banana yield. The result showed that adoption of manure had an association with banana yield and was statically significant at p = 0.000. Most of the respondents were found to be literate. Therefore, illiterate farmers should be enlightened as education boost a farmer’s ability to decode information. Awareness should be created to encourage any member who has not registered to any group as is through such groups the information is disseminated. Extension services should also be well strengthened. The study recommends farmers to adopt manure as a way of increasing banana yields.
QUANTUM-ENHANCED NEURAL NETWORK FOR FORECASTING KENYAN ECONOMIC GROWTH
(Chuka University, 2023-10) SAIF KINYORI
Machine learning has had success in solving real world problems using classical computers. Since its adoption, it has undergone tremendous algorithms improvements. One of the most important advancements in this area of computer science is deep learning. Deep learning has already outperformed humans in a number of fields, e.g stock market forecasting. On the other hand, the data that machine learning algorithms consume becomes more complex and keeps on growing through the use of personal computers and mobile phones. Deep learning algorithms have been employed in these data analytics to come up with trends or classifications that can be translated into actionable results that are useful in many areas. However, these very large or complex datasets take a very long time to train. This is due to the fundamentals of classical computing operations in processing data in the basic binary of 0s and 1s. Quantum computers run on qubits and researchers have been able to prove that they have an advantage over the current classical computers in processing of data. Therefore, this study employed experimental quantum-enhanced paradigms to develop a quantum-enhanced neural network model to forecast Kenyan economic growth. It took advantage of quantum-enhanced simulators, currently in place to experiment on the efficiency of data analysis. Analysis on Kenyan economy indicators datasets from the World Bank was used to evaluate the performance and how fast actionable results and economic growth forecasts can be obtained. The data was transformed to a dimension of quantum vector to allow for the data dimensions to be within the boundaries for quantum computers. The quantum-enhanced neural network model demonstrated a computing time reduction of approximately 97.7% when compared to the artificial neural network model, demonstrating a remarkable increase in efficiency. The model mean absolute error indicated a relatively small average deviation of 0.01047 from the actual values. In addition, the mean squared error indicated a low average squared deviation of 0.00025 from the true values. The discrepancy from the expected and actual values was 0.99775, which indicated a high degree of predictability and a strong fit of the model to the data. The quantum-enhanced neural network model demonstrated an overall good performance in all areas of the forecast study.

Browse

Browsing Computer science by Issue Date