Master of Philosophy (M.Phil.)
Permanent URI for this collectionhttp://192.248.9.226/handle/123/18731
Browse
Recent Submissions
- item: Thesis-Full-textOriental musical instruments identification by selecting optimized features and suitable classifier(2023) Uruthiran P; Ranathunga UThe research field of Music Information Retrieval is a particular subcategory, which brings out data from the audio signal by the expedient of digital signal analysis. This thesis deals with temporal and spectral features of music instruments. Particularly, the formant concept of timbre is the main subject all through. This theory expresses that auditory musical instrument sounds may be classified with the aid of their formant structures. Ensuring this concept, our method aims to suggest a computer based implementation for constructing tools for musical instrument recognizable proof and grouping systems. One of the most crucial aspects of musical instrument classification is selecting the relevant set of features, which are very important steps in musical instrument identification. Feature selection is an important task in musical instrument identification. Feature selection is one routine of reaching dimension reduction, and after an ephemeral debate of various feature selection techniques, the study endorses a derived technique for predominant feature selection in a sequential forward feature selection with a greedy algorithm. This technique is empirically selected to optimize the best set of features by using train data and it is displayed to gain classification accuracy with a diminished predominant set of features much like that gained with a complete set of features. This study extracted the 44 features from 20 musical instruments with three musical families. The three classifiers used in this task, were Decision Tree, kNN, SVM and CNN. The best-selected features have been used in the classification. The confusion matrix got from each classification for evaluation to the performance of the classifiers. The SVM classifier contains the lowest error rate, and the highest AUC scores most values are 1 and a few are within the range of 0.99 - 0.98. Finally, the approval results are finished. SVM classifier is found to be the best classifier among the four classifiers. The predominant features are selected by the Greedy algorithm with SFFS technique for individual musical instrument and selected features are used for polyphonic music identification.
- item: Thesis-Full-textSelection of JPEG steganography algorithms using a feature based modelSenthooran, V; Ranathunga, LJPEG image steganographic techniques use the DCT coefficients scaled by quantization table to make secure data hiding without degrading the image quality. The selection process of data embedding locations in lower frequency DCT coefficients should be carefully considered in each image blocks as these lower frequency coefficients are high sensitive to human eyes. Some of the existing related JPEG steganographic methods have been proposed with primary quantization table modification to hide message bits in the quantized DCT coefficients with minimal distortion by analyzing the properties of quantization table entry and relevant DCT coefficients. The performance of the JPEG steganographic methods is evaluated by the imperceptibility and embedding capacity. In the literature of quantization table modification based JPEG steganography, the middle frequency coefficients in each image block are utilized to embed maximum message size by modifying the middle part of the relevant quantization table values with minimizing the effect of visual perception. However, the data hiding techniques in lower frequency coefficients from the existing studies endure from imperceptibility while increasing the message size. This study suggests the lower frequency data hiding algorithms with utilizing middle frequency data hiding in terms of the modification of lower and middle part of the quantization table values by evaluating image quality parameters and it doesn’t affect the perceptual detectability and improves embedding capacity. The proposed JPEG steganography investigates the modification of quantization table values with regarding to selected lower frequency DCT coefficients for data hiding and selects different data hiding patterns in lower frequency area in terms of modification of quantization table. Finally, it returns the pair of relevant modified quantization table and generated data hiding pattern for an image based on the empirical results of the PSNR values. The pair that contains modified quantization table and data hiding pattern shared by the sender is used as a secrete key to extract the message at the receiver side. From the preliminary studies, the selection of appropriate lower frequency coefficients in image block to hide the optimum size of secrete message with perceptual un-detectability is dependent on the combination of image features, message size and the hiding algorithm. Further, this study recommends a dynamic model to keep the consistency of the combination of image features, message size and the hiding algorithm in terms of quantization table modification and this model based steganography suggests a dynamic model to cover image statistics. Eventually, the model prevents visually perceptible changes for maximum embedding message bits. The proposed method achieves a good imperceptibility level and it is evaluated by the PSNR value range 30dB to 45dB and maximum message size more than 52 bits per block for the selected JPEG image dataset. The dynamic model fitted between the quantization tables and cover image statistics shows the statistical significance with the p-value 0.0007634 and the model generated between the data hiding pattern and statistical features of DCT coefficients shows the statistical significance with the p-value 4.598e-13. The dynamic model for the selected data hiding patterns in the lower frequency coefficients hides the message and it is stego invariant for message analyzers.
- item: Thesis-Full-textImprovement of coronary angiography for quantitative coronary analysis by using a computer vision techniqueKulathilake, KASH; Ranathunga, LCoronary cine-angiography is an invasive medical image modality, which is widely used in Interventional Cardiology for the detection of stenosis in Coronary arteries. Quantitative coronary analysis is one of the demanding areas in medical imaging and in this study a semi automated quantitative coronary analysis method has been proposed. Direct coronary cineangiogram frames are processed in order to obtain the features of lumen such as, vessel boundary, skeleton and luminal diameter along the vessels’ skeleton as the results. The proposed method consists of four main implementation phases namely, pre-processing, segmentation, vessel path tracking and quantitative analysis. The visual quality of the input frames is enhanced within the pre-processing phase. The proposed segmentation phase is implemented based on a spatial filtering and region growing approach. A clinically important vessel region is processed to detect the vessel boundary and skeleton, which is required as prior knowledge for quantitative analysis. Moreover, the vessel diameter is computed while tracking the vessel skeleton path starting from a given seed. The proposed segmentation method possesses 93.73% mean segmentation accuracy and 0.053 mean fallout rate. Moreover, the proposed quantitative analysis method has been validated for assessing its’ technical supportability using a clinically approved data set. As a result of that, this proposed method computes the vessel diameter along the vessel skeleton in single pixel gap and develops the ability to determine the diameter stenosis as the quantitative analysis results. Additionally, the clinical feasibility of the proposed method has been validated to emphasize the clinical usability. Moreover, this study can be further extended to make clinical decisions on stenosis through the functional significance of the vasculature by using proper medical image modality like biplane angiography.
- item: Thesis-Full-textAutomated pedagogical expert for evaluating web-based e-learning content(2018) Sirisuriya SCMDS; Ranathunga L; Karunanayaka SP; Abdullah NAe-Learning has been revolutionizing education system based on the concept of learning occurring at any time and any place. The advent of e-Learning has not only bridged the gap between distance and education but also in student learning and student performance by allowing for more personalized teaching. Behind any successful e-Learning program, it is a necessity to maintain careful design and attractive content that can keep the audience focused and interested. Hence, the importance of evaluating web-based e-Learning content is non-secondary in the e-Learning content development. The evaluation process usually consists of pedagogical evaluation and content evaluation, because e-Learning course material is a combination of the course’s content, as well as the way it is delivered. This research study is mainly focused on automating the pedagogical evaluation component of web-based e-Learning content. In automating the pedagogical evaluation, identifying inconsistencies is the biggest challenge faced by pedagogical experts in the current manual reviewing process, because different institutions use different checklists to pedagogically evaluate their web-based e-Learning content. Developing a calibrated checklist that can be used in the pedagogical evaluation process is the solution to this matter. This calibrated checklist was devised based on studying existing checklists and then a questionnaire was created, and a survey conducted with pedagogical experts to identify the most important review factors which are considered in the pedagogical evaluation process. Additionally, a quantitative formula was devised to weigh the importance of each review factor along with their related SRFs. This study achieves the following objectives. First to build a calibrated checklist that indicates the most important factors for evaluating the pedagogical effectiveness of Web based e-Learning content. Secondly, to prepare a quantitative formulation for determining the pedagogical effectiveness of Web based e-Learning content. Both the checklist and the quantitative formulation can be instrumental towards the development of a theoretical framework for pedagogical compliance of e-Learning content. This framework can provide the foundation to design and develop a tool for assisting pedagogical experts in their evaluation process prior to making a decision whether a particular e-Learning content is well designed or not. Further, it will pave the path to elicit a quantitative approach for pedagogical evaluation. The benchmarked results of automated pedagogical expert results and the manual evaluation results with respect to the variation within one times standard deviation of mean values of manual evaluation have shown the validity of the framework. Further, this study has elicited a quantitative measure to align with manual evaluation to provide consistence evaluation framework.
- item: Thesis-AbstractA Computational model for recognising students emotions in E-learning systems(2014-08-06) Sandanayake, TC; Madurapperuma, APOnline learning is a support tool for educators as well as a medium of delivery of any-time, any-where delivery of a content to a dispersed learner community. Web-based learning environments are a relatively new medium of learning to Sri Lankan universities. Like any learning process, online learning depends on effective communication of human knowledge, whether this occurs in a face-to-face classroom or across the Internet. The effectiveness of online learning also depends on establishing two-way communication between facilitators and learners, and among learners themselves. Although both emotions and interest can increase learners’ likelihood to engage in traditional learning, little is known about the influence of emotions and interest in learning activities in a digital environment. Emotions play an essential role in decision making, managing, perceiving and learning and influence the rational thinking process of humans. Emotions are also important in teaching and learning and often find expression in particular ways, such as interactions with others and motivation in learning. The influence of emotions on e-learning is still not emphasized. Continuous and increasing exploration of the complex set of parameters surrounding online learning reveals the importance of the emotional states of learners and especially the relationship between learning and affective behaviour. Previous research have identified that emotions occur while individuals assess events in their environment that are related to the needs, goals and well-being. Moreover, recent research on the emotional response to online learning has focused on the importance of learners’ feelings in relation to the community of learning. The aim of the research is to develop a model to recognize leaner emotions in online learning environment. Through a critical literature review on affective computing, the study has identified several models and selected Barry Kort’s Learning Spiral Model as the prototype model of the research study. The learning spiral model is a four quadrant learning model in which emotions change while the learner moves through quadrants and up the spiral. This study will be presenting a model which describes the relationship between the online learners learning performances and emotions that occur during online learning process. The research study has built a high-level architecture which consists of three sub modules representing the current context on online learning and two sub modules representing the novel approach of affective learning. Experiments were conducted based on the sub modules developed. The research was focused on identifying a suitable tool to recognise the online learner’s emotions. During the comprehensive literature survey, different tools enabling recognising learner emotions were identified and the study has selected Achievement of Emotions Questionnaire (AEQ) by Pekrun et al. to be applied in recognising learner emotions. Therefore, the study has developed Online AEQ based on the AEQ which is suited for the online learning environment. The study has identified six parameters which represent the learner’s level of learning during the learning experience. These parameters are analysed using multiple regression analysis and a model equation was developed to compute the online learner’s level of learning. Finally the study has analysed and evaluated the correlation between the learner emotions and the observed behaviour. This research study therefore developed a novel model of affective online learning which can be use as a tool to recognise online learner’s emotions with regard to the performance in learning.
- item: Thesis-AbstractA Computational grammar of Sinhala for English-Sinhala machine translation(5/26/2011) Hettige, B; Karunananda, ASCommunication is fundamental to the evolution and development of all kinds of living beings. With no disputes, languages should be recognized as the most amazing artifacts ever developed by mankind to enable communication. Computer has also become such a unique machine, due to its capacity to communicate with humans through languages. It is worth mentioning that the languages understood by computers and humans are quite different, yet people can communicate with computers. This has been possible since the computer is fundamentally an artifact that can translate one language to another. Therefore, computers must be able to do language translations than any other computing task. Nowadays, computing is evolving to enable machine-machine communication with no or little human intervention, yet humans continue to face with what is called language barrier for communication. In particular, a vast collection of world knowledge written in English has been inaccessible to communities who cannot communicate in English. Such communities are unable to contribute to the development of world knowledge due to the language barrier. As a result many people have embarked into research in computer aided natural language translation. This area is commonly known as Machine Translation. Among others, Aptium, Bable fish, Google translator, SYSTRAN, EDR, Anusaaraka, AngalaHindi, AnagalaBarathi, and Mantra are some examples for popular machine translation systems. These systems use various approaches including Human-assisted, Rule-based, Corpus-based, Knowledge-based, Hybrid and Agent-based to translate from one language to another. However, due to inherent diversifications of natural languages, a generic machine translation approach is far from reality. This thesis presents a computational grammar for Sinhala language to develop English to Sinhala machine translation system with an underlying theoretical basis. This system is known as BEES, an acronym for Bilingual Expert for English to Sinhala machine translation. The concept of Varanegeema (conjugation) in Sinhala language has been considered as the philosophical basis of this approach to the development of BEES. The Varanegeema in Sinhala language is able to handle large number of language primitives associated with nouns and verbs. For instance, Varanegeema handles the language primitives such as person, gender, tense, number, preposition and subjectivity/objectivity. More importantly, Varanegeema allows deriving all associated word forms from a given base word. This enables to drastically reduce the size of the Sinhala dictionary. Since the concept of Varanegeema can be expressed by a set of rules, it nicely goes with rule-based implementation of machine translation systems. BEES implements 85 grammar rules for Sinhala nouns and 18 rules for Sinhala verbs. BEES compresses with seven modules namely English Morphological analyzer, English Parser, English to Sinhala base word translator, Sinhala Morphological Generator, Sinhala Parser, Transliteration module and Intermediate Editor. In addition to the main modules, system comprises of four dictionaries, namely, English dictionary, Sinhala dictionary, English-Sinhala Bilingual dictionary and the Concept dictionary. BEES primarily shares the features with the Rule-based, Context-based and Human-assisted approaches to machine translation. The BEES has been implemented using Java and Swi-Prolog to run on both Linux and Windows environments. The English to Sinhala Machine Translation system, BEES has been evaluated to test the hypothesis that concepts of Varanegeema can be used to drive English to Sinhala machine translation. The English to Sinhala machine translation system has been evaluated through three steps. As the first step, all the language processing primitives such as morphological analyzers, parsers, translator and the transliteration module have been tested through the white box testing approach. In order to test each module, several online testing tools ii including English morphological analyzer, English parser and Sinhala word generator have been implemented. By using these online tools each module has been completely tested through a carefully created test plan. In addition, an online evaluation test bed has also been implemented to continuously capture feedback from online users. This online evaluation test bed gives facilities to make different types of sentences using a given set of words. Word Error Rate and the Sentence Error Rate were calculated by using these evaluation results. Finally the intelligibility and the accuracy tests have been conducted through the human support. In order to evaluate the intelligibility and the accuracy of the English to Sinhala machine translation system, following steps were followed. Two hundred sample sentences were collected and grouped into 20 sets (10 sentences per each set). Then each sentence was translated using the English to Sinhala Machine Translation system. Each set was given to the human translators and scored. The intelligibility and the accuracy were calculated through the above evaluation results. The experimental result shows that English morphological analyzer, English parser, English to Sinhala base word translator, Sinhala morphological generator and the Sinhala sentence generator successfully work with more than 90% accuracy. Overall result of the evaluation shows 89% accuracy with the word error rate of 7.2% and the sentence error rate of 5.4%. The BEES successfully translates English sentences with simple or complex subjects and objects. The translation system successfully handles most commonly used patterns of the tenses including active and passive voice forms.