International Postgraduate Research Conference (IPRC)
Permanent URI for this communityhttp://repository.kln.ac.lk/handle/123456789/155
Browse
7 results
Search Results
Item A Trust Framework for Social Networks in MANET Environment(International Postgraduate Research Conference 2019, Faculty of Graduate Studies, University of Kelaniya, Sri Lanka, 2019) Mayadunna, H.; Liyanage, S.R.The improvement of online social networks such as Facebook, Twitter, Instagram has been expanded the idea of using social networks wider. The utilization of mobile phones of general public that given access to social networks makes such platforms popular. Node to node communication in a network gives a discussion to their individuals to associate with different individuals in the systems and share hobbies, opinions, and educational involvements including daily experiences. A significant number of these online social networks are operated with the point of associating to connect many people. Hence, it is important to enhance trustworthiness in social networks. This research is focusing on implementing a trust factor in the device layer. Information within the social networks can be used to get additional trust value for the devices. Hence trust can be calculated at the upper layers to be used at the device level. Thereby, research has developed a social trust framework to allow MANET (Mobile Ad-hoc Network) environment to move cross layer to find trust-related information which can be used at the device level for decision making. The captured social network behavior will provide an indication of how trustworthy the same device by capturing upper layer information. The intent of this research project is to create a trust layer on top of a social environment, in order to achieve the advantages of trustworthy connections. A network structure has been developed in order to complete that achievement. Prior to that, information of Facebook personal friend networks has been extracted and analyzed. Analyzing the parameters which are related to security of the social network is done through a literature survey. While examining the information from social networks, appropriate security-related parameters were selected with their possible states and values. A social network is a group of people or organizations or other entities that connected by a social relationship including friendship, information exchange or corporative working. Social network analysis is the process of mapping and measuring relationships, interactions and flows between people, groups, organizations or other social entities. In general, social network trust can be defined as a measure of confidence that an entity or entities behaves in an expected manner. The research work is reviewing the definitions and measures of trust by focusing on social networks where it can be utilizing within further achievements such as improving security within any kind of networkItem Code Vulnerability Identification and Code Improvement Methods using Advanced Machine Learning(International Postgraduate Research Conference 2019, Faculty of Graduate Studies, University of Kelaniya, Sri Lanka, 2019) Ruggahakotuwa, L.O.; Liyanage, S.R.Dealing with cyber-attacks has become a routine task of modern information systems. The misconfigurations of source code can result in security vulnerabilities that potentially encourage attackers to exploit them and compromise the system. The security of a software critically depends on its underlying source code because hackers always hunt for the loopholes of the software which reflects the vulnerabilities of the source code. To mitigate the above-identified threats the researches have produced several commercially used tools such as Vera code, OWASP, source clear, etc. But still, the frequency of threats and data breaches is very high. ‘Veracode’ is capable of doing both static and dynamic analysis but it is very expensive software. ‘OWASP’ uses only the static analysis to automate the detection of the vulnerabilities. ‘Source clear’ is capable of scanning the repositories either manually or automatically. As all the above-mentioned tools are in their testing phase, number of false positives of the results can be high. In this research, we investigated how to automatically identify the software vulnerabilities by conducting a live scan to detect the error fragments and how to correct the detected source code vulnerabilities automatically at the development stage. This system consists of mainly two phases, Error Detection, and Error Correction. Error Detection is done through a live scan. In the live scan, both Static analysis and Dynamic Analysis will run in parallel. In the dynamic analysis the source code was run in the background and checked with random input data. In Static analysis, the source code was checked line by line and verified by another Rule-Based Engine. The source code is highlighted with markers based on the two outputs of static and dynamic analysis. Research analyzed several machining learning models for better accuracy and performance. After the most suitable machine learning model was identified, the model was trained with enough training samples to develop a generalized model. The final system was implemented to identify vulnerable code segments in Java source codes and suggest corrected code fragment to the developerItem Recommendations for Students in Higher Education: A Machine Learning Approach.(2017) Kasthuriarachchi, K.S.T.; Liyanage, S.R.Educational Data Mining is a rising discipline in Data Mining setting which concentrated on creating systems for investigating one of a kind data that starts from educational settings, and utilizing those procedures to better comprehend students and the settings which they learn in. There were numerous potential circumstances for applying data mining in education, such as; predicting the performance of students in education domain, advancement of student models, making methodologies for instructive help, settling on decisions to growing better learning systems, upgrading the execution of students and lessening the dropout rate of students and so on. There were sure examinations directed in dissecting students' data to foresee the execution in light of data mining approaches utilizing machine learning algorithms. However, a few of them were guiding the students using the recommendations of educators to success in their academic lives. The key objective of this research is to provide educators‘ recommendations to students in higher education through data analysis using machine learning algorithms. In this experiment, the data about more than 3000 students with eight attributes; age, gender, A/L Stream, A/L English Grade, does the student has repeat modules, GPA of Semester1, GPA of Semester 2 and Pass status of year 1 were included into the research sample who registered and were following their first academic year of an Information Technology degree in an institute. Three classification type machine learning algorithms were used to build the predictive model. They were Naïve Bayes algorithm, Decision Tree algorithm and Support Vector Machine algorithm. The accuracy of the models built by each algorithm have been tested against each other to identify the best model and extracted the most influencing/ important attributes in the model to predict the final grade (pass/ fail) in the end of first year of the students. Accordingly, the accuracy measures of Naïve Bayes, Decision tree andSupport Vector Machine were recorded as 74.67%, 74.01% and 74.01% respectively and it was clear that all three algorithms were holding almost same accuracy level. However, the model generated by Naïve Bayes algorithm has been selected since it was outperformed the rest. Then rank features by importance method was used as the feature selection method to identify the most influencing factors of the predictive model. As the result of it, past repeat modules, GPA of Semester1, GPA of Semester 2 were extracted as the most influencing attributes. Furthermore, these attributes were tested using correlation analysis to measure the significance of the relationship with the target attribute. According to this study, the educators will be able to recommend the students to score good marks for assessments of the subjects to obtain a better GPA to semester 1 and semester 2 without failing the modules to successfully complete the first year of the degree course which make more beneficial for educators as well as students to be success.Item A Review of Data Mining Methods for Educational Decision Support(Faculty of Graduate Studies, University of Kelaniya, Sri Lanka, 2016) Kasthuriarachchi, K.T.S.; Bhatt, C.M.; Liyanage, S.R.Data mining is a computer based information system which is devoted to scanning huge data repositories, generate information and discover knowledge. It attempts to uncover data patterns, organize information of hidden relationships, structure association rules and many more operations that cannot be performed using traditional computer based information systems. Therefore, data mining outcomes represent a valuable support for decisions making in various industries and education is one domain that can benefit from data mining. Application of data mining in education is living in its spring time and preparing for a hot summer season. Educational data mining emerges as a paradigm oriented to design models, tasks, methods, and algorithms for exploring data from educational settings. Educational Data Mining develops and adopts statistical methods, machine- learning and data mining methods to study educational data generated basically by students and educational instructors. The main goal of applying data mining in education is largely to improve learning by enabling data driven decision making for improve current educational practices and learning materials. Educational knowledge discovery, in data mining point of view can be seen as a similar process of applying the general knowledge discovery and data mining process and in experimental point of view, it can be seen as an iterative cycle of hypothesis formation, testing and refinement which not just turn data into knowledge but, also to filter the mined knowledge for decision making. There are many applications in education arena that have been resolved using data mining. There are more research studies have also been conducted under various educational problem categories and also there are a number of frequently used data mining methods use in Educational Data Mining. Various open source and commercial tools are available to apply data mining methods on the educational data. This study focuses on the identification of various educational problem domains where data mining methods can be applied and to study the suitability of the available data mining methods and the tools to perform Educational Data Mining in Sri Lankan Educational Institutes. The knowledge discovered by this review is expected to generate meaningful insight and provide guidance for important decisions made by educators.Item Sales Prediction with Data Mining Algorithms(Faculty of Graduate Studies, University of Kelaniya, 2015) Rathnadiwakara, A.S.K.; Liyanage, S.R.Nowadays most business fields using many strategies to improve business profits. Most of them used traditional methods. Therefore, those company‘s efficiency and profit goes to the critical situation. So the improve efficiency of the company is a major requirement for nowadays business platform. Using new technologies companies can improve their profit and efficiency. Also companies can identify their sales life-cycle. This sales prediction was carried for Alfred Edirisinghe (PVT) LTD which is a medium scale tyre dealer in Colombo. Decision Tree, Association Rules and Naïve Bayes data mining models were attempted for the prediction. The best algorithm was selected for each model. Item Code, Item Type, Item Quantity, Item Value, Item Sold Date, etc. variables were used in data mining process. Among those variables five variables were selected for the mining process. A sales data sample with 5000 records were provided by the client for the analysis. Out of the 5000 records 30% was used in the mining process. According to the predicting probabilities, Decision Tree algorithm were performed 98.65%, Association Rules algorithm were performed 100.00% and Naïve Bayes algorithm were performed 99.57%. Decision Tree belongs to the lowest predict probability value. Therefore Decision Tree model was the worst model. Association Rule model contains highest predicted value 100.00%. Therefore it was the best model. Naïve Bayes model was also a good model. The Score results indicate that Decision Trees and Naïve Bayes mining model has the best score 1.00 and followed by Association Rule mining algorithm with score of 0.99. By considering score and target population with predicting probabilities, Association Rule algorithm was the best one for prediction process. Data mining model was implemented using Association Rule algorithm. According to these predicting results, the company can handle their imports optimizing the available resources; storage, time, money. Therefore this research would benefit the Company to improve their incomes.Item Real Time Emotion Based Music Player for Android(Faculty of Graduate Studies, University of Kelaniya, 2015) Dissanayaka, D.M.M.T.; Liyanage, S.R.Listening to music has been found to affect the human brain activities. Emotion based music players with automated playlists can help users to maintain a selected emotional state. This research proposes an emotion based music player that create playlists based on real time photos of the user. Two emotional statuses, happy and not-happy were considered in this study. User‘s images were captured in real-time using an android device camera. Grey scaled images were used to compress the image files. Eye and lip areas were cropped and sent to the MATLAB backend via client server-socket connections. Gaussian filtering was applied to reduce noise. Canny Edge Detection algorithm was used for edge detection. Eigen face-based pattern recognition was used for emotion recognition. PCA eigenvectors were learnt from the dataset via unsupervised training to learn the Eigen face models. The dissimilarity between pairs of face images projected to the Eigen space were measured using the Euclidean distance. The matched image was the one with the lowest dissimilarity. The identified label, happy/not-happy was transmitted back to the Android music player via a client server socket connection. Songs that are pre-categorised as happy/ not-happy are stored in the android application. When emotional label of the perceived face image is received, songs relevant to the received emotional label are loaded to the android music player. 120 face images were collected at the Department of Statistics & Computer Science, University of Kelaniya for validation. Another 100 happy and 100 not-happy images were collected for testing. Out of the 100 test cases with happy faces 75 were detected as happy and out of the 100 not-happy faces 66 were classified as not-happy. The overall accuracy of the developed system for the 200 test cases was 70.5%. This concept can be extended from a single face to multiple faces and the system has to be made more robust to noises, different poses, and structural components. The system can be extended to include other emotions that are recognizable via facial expressions.Item Eliminating the storage wastage of CCTV cameras by motion detection(Faculty of Graduate Studies, University of Kelaniya, 2015) Ranasinghe, A.N.; Liyanage, S.R.Despite the ever increasing capacity of data storage mediums, there is a wider appeal for studies on efficient storage management to avoid the wastage of capacity due to unwanted data volumes. In line with the demand for research on capacity optimization, this study focuses on the efficient use of storage space by avoiding unwanted data with respect to the storage management in Closed-circuit television (CCTV) camera systems. Therefore, deviating from the common high end hardware solutions such as sensors, study introduces a software solution to store the video only when a motion occurs. Comparison of video frames using image processing is used as the basic method to identify motion. The grayscale version of the each frame and the calculated absolute difference between the video frame and base image are used to identify the motion. A threshold filter is employed to eliminate the unnecessary effects due to noise. The value chosen for the threshold is dependent on the noisiness of the environment as it affects the sensitivity. The threshold value can be optimized statistically using a cost function based on the errors. In this study, a threshold values between 10 and 15 were found to be suitable for the laboratory environment which is considered as low noise indoor environment. Finally, an edge filter can be applied to identify the moving object in the video. The study has utilized the advantages of gradual update (blending the base image with current video frame in a lower rate than actual changing rate of the current frame) of base image rather than using a static image to compare with the live image. In a commercial perspective, this study focuses on a mechanism that can be used to transfer the live feed of CCTV cameras at a very high speed to an Android mobile phone which is connected to the same network. According to the test results, the solution proposed in this study saves about 50% of storage space of CCTV cameras in an environment with limited motions while providing a very fast live streaming of the video footage. This would be an ideal storage solution for domestic CCTV camera systems which generally deal with limited motions.