Symposia & Conferences
Permanent URI for this communityhttp://repository.kln.ac.lk/handle/123456789/10213
Browse
11 results
Search Results
Item Sinhala Handwritten Postal Address Recognition for Postal Sorting(IEEE International Research Conference on Smart computing & Systems Engineering (SCSE) 2019, Department of Industrial Management, Faculty of Science, University of Kelaniya, Sri Lanka, 2019) Ifhaam, M.F.A.; Jayalal, S.Sri Lankan post office mail sorting process is done manually, even today. Though employees are well experienced, it takes considerable time and employees need to work overtime in places like Central Mail Exchange (CME). With major issues like unclear handwriting, having trouble to recognize some uncommon or ambiguous names, and carrying these duties twice a day create a negative impact on the efficiency of the postal delivery system. In the prevailing system, forward mails and delivery mails are the two categories of separating mails at the sorting centers. Delivery mails are the posts which can be delivered to its destination directly. Forward mails are the ones which need to be sent to an appropriate post office that can deliver the particular post to its destination. Majority of Sri Lankans use Sinhala language for their day to day activities. The primary objective of the research is to identify the automatic way of forwarding the letter to the next post office from the current post office. Proposed system is focused on the recognition of Sinhala handwriting using Optical Character Recognition (OCR) and image processing technologies. Data collected under different criteria were used for training and testing the solution. Genetic Algorithm (GA) was used to generate more optimized results faster with higher accuracy. Given addresses are written in the default format. This format can be extended to more formats as improvements in future. The algorithm shows accuracy over 92% for addresses which are recognized with 3 misrecognized characters. This algorithm can be used on practice scenario as the AI Recognition has more than 79 % of accuracy.Item Passion Fruit Disease Detection using Image Processing(IEEE International Research Conference on Smart computing & Systems Engineering (SCSE) 2019, Department of Industrial Management, Faculty of Science, University of Kelaniya, Sri Lanka, 2019) Dharmasiri, S.B.D.H.; Jayalal, S.Fruit diseases are a major problem in economic and production losses in the agricultural industry worldwide. In this paper, an image processing approach is proposed for identifying passion fruit diseases. According to the Sri Lankan context, treatment details are taken by the farmers from the field officers. However, it can take a few days. So, this proposed system can be used to identify passion fruit diseases quickly and automatically. This proposed approach is composed of the following main steps; Image Acquisition, Image Preprocessing, Image Segmentation, Feature Extraction, Dataset Preparation, Training & Testing. Healthy and two types of passion fruit diseases, namely passion fruit scab and woodiness images were used for this approach. This approach was tested according to passion fruit disease type and its’ stages, such as mild, moderate and severe. K-Means clustering was used for segmentation. Images were clustered according to k values, such as 2, 4, 6 and 8. Before the segmentation, images were converted to RGB, L*a*b, HSV and Grey colour models, because of find out the most suitable colour model for this approach. Local Binary Pattern was used for feature extraction and Support Vector Machine was used for creating the model. Seventy percent (70%) of each dataset was used to train the SVM and the other thirty percent (30%) was used to test the model. According to this approach, passion fruit diseases can be identified in the average accuracy of 79% and its’ stage can be identified in average accuracy 66%.Item Prediction of User Intentions Using Web History(IEEE International Research Conference on Smart computing & Systems Engineering (SCSE) 2019, Department of Industrial Management, Faculty of Science, University of Kelaniya, Sri Lanka, 2019) Samarasinghe, K.T.C.S.; Jayalal, S.In the present internet has become much more necessary thing to humans and we use it as a way of sharing information and way of communication. If the networks can identify the user’s intentions, it will be affecting to increase productivity and personalization. Predicting user intention(s) is interesting and useful for many applications such as threat identification, imposing restrictions and cashing web details. The aim of this research is to develop a method to predict user intention using supervised machine learning methods with user’s past historical behaviours. Experiments in this study used access log on a local server and focused on creating single user prediction and multiuser generalize prediction models. Experimental models were created based on several multi-classifier algorithms, such as Support Vector Machine (SVM), Multilayer Perceptron (MLP) and K-Nearest Neighbor (KNN). KNN based models outperform other used algorithms. Also results in this study show that there is some sort of behavioural patterns for peoples to use the internet according to the time and the groups they interactItem Classification of vehicles by video analytics for unorganized traffic environments(International Research Conference on Smart Computing and Systems Engineering - SCSE 2018, 2018) Arachchi, I.M.R.; Jayalal, S.; Rajapakse, C.Traffic monitoring is essential for infrastructure planning and transportation. The objective of traffic monitoring is to have an effective traffic management system. Traffic management systems would be effective in well-organized traffic environments, where it has very disciplinary behaviors and less in inefficiencies. But in unorganized urban environments like Sri Lanka, road traffic behaviours are varying from standard structured ways which lead to discompose the traffic management. An effective monitoring system requires short processing time, low processing cost and high reliability. The paper proposes a novel vehicle detection and classification algorithm based on background filtering and re-engineered with suitable changes in order to be applicable to challenging unorganized traffic environments. The solution is successfully classifying vehicles individually and their trajectories in unorganized traffic environments in order to monitor the behaviors of the drivers. The system gives 74.4% average accuracy in vehicle detection and 55% accuracy in vehicle classification while counting each vehicle passed by. We used OpenCV functions for implementing and testing algorithms. Data was collected through pre-recorded video clips from footbridge crossing at Colombo Fort in western province Sri Lanka, for the testing. The ultimate objective of this research was to come up with a best-suited algorithm for vehicle detection and classification (hybrid solution) in unorganized traffic environments which would help to analyze the behaviors of road users. The solution will lead to help reduce unorganized traffic congestions by enhancing the efficiency and effectiveness of traffic monitoring and analyzing systems those are used for intelligent traffic management systems and traffic simulation models.Item Study of machine learning algorithms for Sinhala speech recognition(International Research Conference on Smart Computing and Systems Engineering - SCSE 2018, 2018) Shaminda, S.; Jayalal, S.Speech is the primary mode of communication among humans and the most natural and efficient form of exchanging information. Therefore, it is logical that the next technological development in natural language speech recognition for Human Computer Interaction is, Artificial Intelligence. Speech recognition can be defined as the process of converting speech signal to a sequence of words by an algorithm implemented using a computer program. Speech processing is one of the challenging areas of signal processing. The main objective of the study was to conduct a study on speech recognition approaches to improve the accuracy level of Sinhala speech recognition. This study was conducted in order to find the optimal algorithm for accurate Sinhala speech recognition. According to the implementation architecture of speech recognition, feature extraction and the pattern recognition phases can be varied with different algorithms. The study identified that Linear Predictive Coding (LPC) and Hidden Markov Model (HMM) gives most accurate results than other combine algorithms.Item Analysis and detection of potentially harmful Android applications using machine learning(International Research Conference on Smart Computing and Systems Engineering - SCSE 2018, 2018) Kavneth, G.A.S.; Jayalal, S.With the rapid advancement of technology today, smartphones have become more and more powerful and attract a huge number of users with new features provided by mobile device operating systems such as Android and iOS. Android extended its lead by capturing 86% of the total market in 2017 (Gartner, 2017) and became the most popular mobile operating system. However, this huge demand and freedom has made the hackers and cybercriminals more curious to generate malicious apps towards the Android operating system. Thus, research on effective and efficient mobile threat analysis becomes an emerging and important topic in cybersecurity research area. This paper proposes a static-dynamic hybrid malware detecting scheme for Android applications. While the static analysis could be fast, and less resource consuming technique and dynamic analysis can be used for high complexity and deep analysis. The suggested methods can automatically deliver an unknown application for both static and dynamic analysis and determine whether Android application is a malware or not. The experimental results show that the suggested scheme is effective as its detection accuracy can achieve to 93% ∼ 100%. The findings have been more accurate in identifying Android malwares rather than separating those two static and dynamic behaviors. Furthermore, this research compares the machine learning algorithms for static and dynamic analysis of the Android malwares and compare the accuracy by the data used to train the machine learning models. It reveals Deep Neural Networks and SVM can be used for and higher accuracy. In addition, era of the training and testing dataset highly effect the accuracy of the results regarding Android applications.Item Machine learning based model for Android malware analysis and detection.(International Research Symposium on Pure and Applied Sciences, 2017 Faculty of Science, University of Kelaniya, Sri Lanka., 2017) Kavneth, G. A. S.; Jayalal, S.Rapid advancement of technology has enabled smartphones to become extremely powerful. They are capable of attracting a considerable amount of users with new features provided by mobile device operating systems such as Android and iOS. Android extended its lead by capturing 86 percent of the total market in 2017, and became the most popular mobile operating system. The Android operating system, which is found on a wide range of devices is owned by Google and powered by the Linux kernel. It is an open source operating system that enables mobile application developers to access unlocked hardware and develop new apps as they wish. However, this huge demand and freedom has made the hackers and cybercriminals more curious to generate malicious apps towards the Android operating system. They constantly target the security vulnerabilities in the operating system to gain access within the system. Even though, Google provides a primary set of security services, there are possibilities for potentially harmful applications in the Google Play store and other third party application stores. Thus, research on effective and efficient mobile threat analysis becomes an emerging and important topic in cybersecurity research area. Many researchers proposed various security analysis and evaluation strategies such as static analysis and dynamic analysis. In this research, we propose a hybrid approach, which aggregates the static and dynamic analysis for detecting security threats and attacks by Android malware application. This approach has two phases. First phase is the static analysis for applications, which will analyze the mobile application without execution. This focuses on extracting app APK file and examining permission requests made by Android apps that have declared in AndroidManifest.xml, as a means for detecting malwares. Because, in most of cases extra permissions granted by applications will allow the attacker to exploit the device. As the next phase, we perform dynamic analysis for mobile application. This phase focuses on runtime data obtained from the applications such as CPU, scheduler information from every running application, network calls, sensor data and so forth. For both phases, we have used supervised, machine learning based algorithms to train models and recognize malwares. In the first phase, potentially harmful applications can be identified as well as in the proposed hybrid mechanism, which is a combination of both phases. Data that was collected by several cybersecurity research centers were used for the evaluation of the proposed hybrid approach and both real-life malware and benign app data demonstrated a good detection performance with high scalability. The initial findings have been more accurate in identifying Android malwares rather than separating those two static and dynamic behaviors.Item Text mining approach on on-line guest reviews: A case study from hospitality industry.(International Research Symposium on Pure and Applied Sciences, 2017 Faculty of Science, University of Kelaniya, Sri Lanka., 2017) Hewage, N. I.; Jayalal, S.Social media and consumer-generated context on the Internet have become an integral part of the modern society with millions of users. With the increase of its popularity, tourism industry has shifted towards electronic transactions. Tourists now tend to use online social media reviews and ratings posted by hotel guests to make decisions before booking a hotel, which is an impossible task for a single user due to its high volume of reviews. Hence, to make better decisions, ranking of hotels for a specific region will be beneficial for the tourists who are willing to travel in that region and for the management of the hotel as well. However, while a handful of studies have employed on hotel guest satisfaction and experience by analyzing online hotel guest reviews collected from online travel agencies, there is a significant research problem with ranking hotels by analyzing hotel guest reviews in aspect level consideration. Expedia.com, Agoda.com and Booking.com are some of the leading online travel agencies that have millions of users. Online reviews used for this study are collected from 18 hotels that belong to all these three online travel agencies, and from that dataset, 6 hotels are selected as testing dataset. The dataset contains reviews from year 2010 to 2016. In this study, we propose a ranking mechanism, that ranks hotels by using the overall rating values, sentiment scores and the reviewed year. For computing sentiment scores, each review is split into sentences and they are categorized in to six attributes as Location, Service quality, Cleanliness, Comfort of rooms, Value for money and other. Thereafter, the sentiment analysis is done by considering the weight of the positive and negative words. In this research, we present a novel ranking algorithm to rank hotels, considering the reviewed year and computing the ranking score by getting the variance of the polarity rate and variance of rate of overall rating from initial year to the last. The results were taken by considering specific time period and without considering specific time period. Therefore, when using all the reviews without considering a time period, the rankings deviate from the Booking.com and TripAdvisor.com rankings. When using the reviews within 3 years of time, the ranking results are almost equal to the TripAdvisor.com rankings. When the time period is reduced for 18 months, the accuracy is 50% with Booking.com rankings and 33.3% with TripAdvisor.com rankings. This could be due to the fact that the above mentioned online travel agencies use online reviews more than two years, and therefore it perhaps causes for the deviation of rankings. Since sample dataset is used in this study, the accuracy can be increased by using a large dataset. Since this ranking mechanism considered variance to clarify the performance of hotels, and it is not only depending on the number of positive reviews or star ratings of the hotel, this is beneficial for the hotels which are not much popular, but having good standards.Item Real-time big data video analytics for unorganized traffic environments(Faculty of Science, University of Kelaniya, Sri Lanka, 2016) Ranaweera Arachchi, I.; Jayalal, S.; Rajapakse, C.Traffic on local roads has reached such a level that it is necessary to address the issue of traffic congestion and seek complex transport solutions for the city. Increase of the number of vehicles on the road becomes one of the key reasons for increasing traffic congestion. Traffic congestion is associated with massive financial and manhour loss and therefore attempts to alleviate this has been of keen interest. The basis of almost all those approaches is traffic monitoring and analysis, leading to having an effective traffic management system. Most traffic management systems are applied in well-organized traffic environments such as highways, where driver discipline is high. But in unorganized urban environments as seen in Sri Lanka, road traffic behavior vary from the accepted standards. Driver and pedestrian indiscipline cause huge traffic congestions in urban areas. Hence in such a scenario, a system that monitors road traffic on different traffic environments is very useful. There are several existing techniques such as Magnetic Loops, Microwave RADAR, Infrared Detectors, Ultrasonic Detectors and Camera Based Systems. Traffic monitoring systems require short processing time, low processing cost and high reliability. Therefore, according to the literature, camera-based monitoring is the best-suited technique for traffic monitoring. Real-time video analytics are part of a centralized approach to modern traffic management which is defined as computer vision-based surveillance that provides algorithms for object detection, tracking, classification and trajectory analysis using real-time traffic surveillance video. It usually uses roadside cameras (CCTV) to obtain traffic information and transmit it to central servers, exhibiting real-time operability of big data. In this study, several approaches and algorithms for moving object detection, based on temporal differencing method, optical flow method, background filtering are compared and a novel real-time vehicle detection and classification algorithm based on background filtering will be proposed and re-engineered in order to be applicable to challenging unorganized traffic environments. The solution will classify vehicles individually and their trajectories in real time in unorganized traffic environments in order to analyze the behaviors of the drivers as well as pedestrians on the road. We use OpenCV which is a library of programming functions mainly aimed at real-time computer vision, for implementing and testing algorithms. Data will be collected via pre-recorded video clips from Kiribathgoda junction in the western province, for the testing purpose and real- time CCTV surveillance video is going to be used as the input for implementation. A comprehensive data analysis is required to be conducted to address the higher processing requirement of such videos. The solution will be validated for performance subsequently. The final objective of this research is to come up with an optimum algorithm for vehicle detection and classification in unorganized traffic environments which would help to analyze the behavior of road users. The solution will lead to reduced traffic congestion in the country by enhancing the efficiency and effectiveness of traffic monitoring and analyzing systems.Item Detection of cyber bullying on social media networks(Faculty of Science, University of Kelaniya, Sri Lanka, 2016) Priyangika. S.; Jayalal, S.Social Media is becoming an integral part of people’s daily lives today. It is an effective way of sharing one’s life experiences, special occasions, achievements and other events with their friends and family. Although it is a fruitful way to communicate with groups, some people find themselves being insulted or offended by others who are involved in certain post or conversations. These insulations can be based on racism, using profanity or any other vulgar or lewd language. This cyber bullying needs to be monitored and controlled by the social media site owners since it will highly effect on the number and safety of the active site membership. Currently, there is no automated process of identifying offensive comments by the social network site itself. It can be only diagnosed by humans after reading the comments, flagging or reporting them to the owner of the site or blocking the offender. Considering the massive big data set generated in social media daily, automatically detection of offensive statements is required to reduce insulation effectively. For this purpose, text classification approach can be applied where a given text will be categorized as insulting or not, through learning from a pre-learned model. In order to develop the model, data was collected from the popular data repository site named www.kaggle.com. The dataset consists of comments posted on Facebook and Twitter. Firstly the dataset was divided into training data set and test data set. Then the collected data was preprocessed by removing the unwanted strings, correcting words and eliminating duplicate data fields. In the next step, features or keywords were extracted which are qualified to distinguish a statement as ‘insulting’ using N-grams model and counting methods. Feature selection is done using Chi- Squared test and finally apply classification algorithms for separating insulting comments and non-insulting comments from a dataset given. Machine learning algorithms such as Support Vector Machines (SVM), Naïve Bayes, Logistic Regression and Random Forest are used for this. Out of the classification algorithms, SVM is to be performed better than other algorithms since this is a two-class classification problem and a comment is to be classified only into two separate classes which are ‘insulting’ and ‘neutral’. With an exact separation of a given comment into ‘insulting’ and ‘neutral’ category, cyberbullying happening through offensive comments posted on social media sites can be detected.