Computing and Information Systems - Theses
Permanent URI for this collection
Search Results
1 - 10 of 418
-
ItemPractical declarative debugging of mercury programs(University of Melbourne, 2006)
-
ItemA multistage computer model of picture scanning, image understanding, and environment analysis, guided by research into human and primate visual systems(University of Melbourne, Faculty of Engineering,, 1983)This paper describes the design and some testing of a computational model of picture scanning and image understanding (TRIPS), which outputs a description of the scene in a subset of English. This model can be extended to control the analysis of a three dimensional environment and changes of the viewing system's position within that environment. The model design is guided by a summary of neurophysiological, psychological, and psychophysical observations and theories concerning visual perception in humans and other primates, with an emphasis on eye movements. These results indicate that lower level visual information is processed in parallel in a spatial representation while higher level processing is mostly sequential, using a symbolic, post iconic, representation. The emphasis in this paper is on simulating the cognitive aspects of eye movement control and the higher level post iconic representation of images. The design incorporates several subsystems. The highest level control module is described in detail, since computer models Of eye movement which use cognitively guided saccade selection are not common. For other modules, the interfaces with the whole system and the internal computations required are out lined, as existing image processing techniques can be applied to perform these computations. Control is based on a production . system, which uses an "hypothesising" system - a simplified probabilistic associative production system - to determine which production to apply. A framework for an image analysis language (TRIAL), based on "THINGS". and "RELATIONS" is presented, with algorithms described in detail for the matching procedure and the transformations of size, orientation, position, and so On. TRIAL expressions in the productions are used to generate "cognitive expectations" concerning future eye movements and their effects which can influence the control of the system. Models of low level feature extraction, with parallel processing of iconic representations have been common in computer vision literature, as are techniques for image manipulation and syntactic and statistical analysis� Parallel and serial systems have also been extensively investigated. This model proposes an integration Of these approaches using each technique in the domain to which it is suited. The model proposed for the inferotemporal cortex could be also suitable as a model of the posterior parietal cortex. A restricted version of the picture scanning model (TRIPS) has been implemented, which demonstrates the consistency of the model and also exhibits some behavioural characteristics qualitatively similar to primate visual systems. A TRIAL language is shown to be a useful representation for the analysis and description of scenes. key words: simulation, eye movements, computer vision systems, inferotemporal, parietal, image representation, TRIPS, TRIAL.
-
ItemLearned Hashmap for Spatial Queries( 2020)Spatial indexes such as R-Tree are widely used for managing spatial objects data efficiently, which is influenced by the popular one-dimensional range index B-Tree. Research has suggested that applying machine learning techniques such as linear regression or a neural network can improve the performance of traditional data structures. However, most studies are focused on tuning recursive learned models or learning a different ordering of data items. Many of them cannot guarantee query accuracy as in the traditional methods. This study investigates a different approach to the learned spatial index, namely Learned Spatial Hashmap (LSPH), which combines the learned model and hashmap. It only requires values from one of the data dimensions to build. Results from experiments on both synthetic and real-world datasets show that our approach significantly reduces the query processing time and maintains 100% accuracy, which is more efficient than traditional spatial indexes and more robust than recently proposed learned spatial indexes.
-
ItemConstructing authentic esports spectatorship: An ethnography( 2020)Formally organised competitive video gaming, commonly known as esports, has seen a rapid rise in popularity over the past decade as a form of spectator entertainment. Thousands of fans pack grand stadia across the globe to watch their favourite esports teams and players compete for extravagant prize pools, while millions watch online remotely in their homes via livestreaming platforms like Twitch. Unlike conventional sports however, there is no inherent need for esports to take place in any particular physically-situated site like a stadium. The computerised nature of esports lends the practice a perceived placeless quality. No matter where an esports event is situated, be it in a stadium or networked across homes, the digital synthetic environments in which players’ avatars compete remain the same as objective constructs of code and graphical assets. If the perspective of watching esports is largely the same across different sites by virtue of its inherent mediation, then why is esports spectated beyond the comfort of the home? The aim of this thesis is to address this conundrum by exploring the experiences esports spectatorship in varying contexts. In particular, this thesis seeks to understand how experiences of esports spectatorship are influenced by and differ across various sites of spectatorship. This aim is achieved through an ethnographic methodology, where data is generated through embedded interactions and experiences with esports spectators at three distinct sites of esports spectatorship. Data generation methods including semi-structured interviews and various forms of participant observation are employed across the ethnographic fieldwork, while data analysis is largely conducted through reflexive thematic analysis and a thick description approach. Three studies are conducted, each looking at a separate site of esports spectatorship: the home, the stadium, and the esports bar. Study 1 focuses on the experiences of domestic esports spectatorship. The findings of the study demonstrate that in the mundanity of the home, domestic spectators perform laborious spectatorship to authenticate and make spectacular their experiences of spectating esports. It also demonstrates that despite a perceived sense of autonomy, domestic spectators commonly encounter numerous compromising factors in their homes which often prevents them from constructing an ideal spectating experience. Study 2 focuses on experiences of and motivations for esports spectatorship in stadia. It reveals that spectators seek to affirm their expectations of authentic esports by attending events held in stadia, thus establishing notions of esporting authenticity. Besides seeking to partake in an authentic experience of esports spectatorship, those attending stadium-situated events seek to present themselves as authentic esports spectators. Aware of their status as props in the mediated spectacle of esports events broadcast to remote audiences, those spectating in stadia seek to present themselves in a perceived authentic manner to convince event organisers to host future esports events in Australia. Study 3 focuses on the experiences of spectating esports in an esports bar, representing a site of public communal spectatorship between the stadium and the home. Despite being a public place, the bar is in many ways more homely than the home. Void of many compromising factors commonly found in domestic environments, spectators at the esports bar are free to exercise a greater degree of autonomy over the construction of authentic esports spectatorship experiences. Taken together, the three studies reveal ways in which esports spectators construct authenticity in their experiences of spectatorship by creating a sense of placefulness. In doing so they establish a convention of esports authenticity for both those within and outside of the esports fandom. Different sites of spectatorship offer different tools, resources, and opportunities to construct experiences of esports spectatorship. Spectators choose accesible sites that best allow them to construct what they percieve an experience of esports spectatorship ought to be.
-
ItemSafe acceptance of zero-confirmation transactions in Bitcoin( 2016)Acceptance of zero confirmation transactions in Bitcoin is inherently unsafe due to the lack of consistency in states between nodes in the network. As a consequence of this, Bitcoin users must endure a mean wait time of 10 minutes to accept confirmed transactions. Even so, due to the possibility of forks in the Blockchain, users who may want to avoid invalidation risks completely may have to wait up to 6 confirmations, which in turn results in a 60 minute mean wait time. This is untenable and remains a deterrent to the utility of Bitcoin as a payment method for merchants. Our work seeks to address this problem by introducing a novel insurance scheme to guarantee a deterministic outcome for transaction recipients. The proposed insurance scheme utilizes standard Bitcoin scripts and transactions to produce inter-dependent transactions which will be triggered or invalidated based on the occurance of potential doublespend attacks. A library to setup the insurance scheme and a test suite was implemented for anyone who may be interested in using this scheme to setup a fully anonymous and trustless insurance scheme. Based on our test in Testnet, our insurance scheme was successful at defending against 10 out of 10 doublespend attacks.
-
ItemStatistical Approaches for Entity Resolution under Uncertainty( 2020)When real-world entities are referenced in data, their identities are often obscured. This presents a problem for data cleaning and integration, as references to an entity may be scattered across multiple records or sources, without a means to identify and consolidate them. Entity resolution (ER; also known as record linkage and deduplication) seeks to address this problem by linking references to the same entity, based on imprecise information. It has diverse applications: from resolving references to individuals in administrative data for public health research, to resolving product listings on the web for a shopping aggregation service. While many methods have been developed to automate the ER process, it can be difficult to guarantee accurate results for a number of reasons, such as poor data quality, heterogeneity across data sources, and lack of ground truth. It is therefore essential to recognise and account for sources of uncertainty throughout the ER process. In this thesis, I explore statistical approaches for managing uncertainty—both in quantifying the uncertainty of ER predictions, and in evaluating the accuracy of ER to high precision. In doing so, I focus on methods that require minimal input from humans as a source of ground truth. This is important, as many ER methods require vast quantities of human-labelled data to achieve sufficient accuracy. In the first part of this thesis, I focus on Bayesian models for ER, owing to their ability to capture uncertainty, and their robustness in settings where labelled training data is limited. I identify scalability as a major obstacle to the use of Bayesian ER models in practice, and propose a suite of methods aimed at improving the scalability of an ER model proposed by Steorts (2015). These methods include an auxiliary variable scheme for probabilistic blocking, a distributed partially-collapsed Gibbs sampler, and fast algorithms for performing Gibbs updates. I also propose modelling refinements, aimed at improving ER accuracy and reducing sensitivity to hyperparameters. These refinements include the use of Ewens-Pitman random partitions as a prior on the linkage structure, corrections to logic in the record distortion model and an additional level of priors to improve flexibility. I then turn to the problem of ER evaluation, which is particularly challenging due to the fact that coreferent pairs of records (which refer to the same entity) are extremely rare. As a result, estimates of ER performance typically exhibit high levels of statistical uncertainty, as they are most sensitive to the rare coreferent (and predicted coreferent) pairs of records. In order to address this challenge, I propose a framework for online supervised evaluation based on adaptive importance sampling. Given a target performance measure and set of ER systems to evaluate, the framework adaptively selects pairs of records to label in order to approximately minimise statistical uncertainty. Under verifiable conditions on the performance measure and adaptive policy, I establish strong consistency and a central limit theorem for the resulting performance estimates. I conduct empirical studies, which demonstrate that the framework can yield dramatic reductions in labelling requirements when estimating ER performance to a fixed precision.
-
ItemPredicting Students' Intention to Use Gamified Mobile Learning in Higher Education( 2020)Background and Objectives: While gamified mobile learning holds the potential to offer an interactive learning environment that can improve students’ engagement, the predictors of its adoption remain underexplored, especially in a higher education context. The aim of this study, therefore, is threefold: (i) to identify predictors of higher education students’ intention to use gamified mobile learning; (ii) to examine the correlations among these predictors; (iii) to explore students’ attitudes towards different game elements. Methods: For the first and second objectives, the study extended the Unified Theory of Acceptance and Use of Technology (UTAUT) with cognitive gratification and perceived enjoyment. For the third objective, the study explored students’ attitudes towards five popular game elements in gamifying learning systems; these are Points, Levels, Leaderboard, Teams, and Gifting. A total of 440 responses were collected from higher education students from different regions of Saudi Arabia, using Qualtrics survey tool. After conducting data screening, 271 valid responses were considered in the analysis of Structural Equation Modeling (SEM), particularly in the items of the hypothesised model, using AMOS 27. For the third research objective, 399 valid responses were obtained and analysed, using SPSS 27. Results: Our findings reveal that perceived enjoyment (β= .507, p < .001) and social influence (β= .261, p < .001) had the strongest positive effects on intention to use gamified mobile learning, followed by performance expectancy (β= .179, p= .008) and effort expectancy (β= .138, p= .034), while cognitive gratification had no influence (β= -.020, p= .770). The proposed model was able to explain 71% of the data variance in usage intentions. For the third objective, the results showed that game elements that allow students to quantify their achievements as individuals, i.e., Points and Levels, are the most favourable, while there seemed to be high variation with the one that encourages competition, Leaderboard, especially among female groups. Lastly, the elements that encourage collaboration, Teams and Gifting, received the lowest positive perceptions. Originality: The original contribution of this study is the empirically backed impact of the extended UTAUT on students’ intention to use gamified mobile learning in higher education. It also contributes in shedding light on which game elements are most promising in this context. The study offers a set of practical outcomes to contribute to realising successful adoption of gamified mobile learning in higher education.
-
ItemEfficient algorithms to improve feature selection accuracy in high dimensional data( 2020)Feature selection plays a vital role in machine learning by removing the features which are irrelevant for the learning task in high dimensional datasets. Selecting the relevant features before the learning task often improves the prediction accuracy of the learning models, reduces the training and prediction times of the learning models and results in simple learning models. However, selecting the optimal feature subset is an NP-hard problem and achieving high prediction accuracy with low computational costs remains a challenge. In this thesis, we develop efficient feature selection algorithms which gain high prediction accuracy in machine learning tasks to address this problem. First, we propose a method to improve the feature selection accuracy in multi-class binary data compared to state-of-the-art methods for binary feature selection. Despite many feature selection methods specifically designed for different data types, only a few methods are specifically designed for binary data. These few methods also result in poor class separation; therefore, low prediction accuracy in multi-class binary datasets. In our feature selection method, we first propose two new feature selection measures which exploit the special properties in binary data. Second, we propose a feature selection algorithm, which uses these measures to achieve good class separation in multi-class binary data. We experimentally show that compared to the existing feature selection methods, the proposed method achieves high classification accuracy in multi-class binary datasets with reasonable computational costs. The proposed method has wide applicability in domains such as image and gene analysis and text classification in which binary datasets are commonly used. Second, we propose a framework, which facilitates the supervised filter feature selection algorithms to exploit feature group information from external sources of knowledge. Among the various feature selection methods available, filter methods use statistical measures, and wrapper and embedded methods use the classification accuracy of a learning model to evaluate the features. Compared to the wrapper and embedded methods, filter methods are advantageous in terms of the interpretability of the results, simplicity and computational efficiency. However, in the literature, feature group information, obtained from external sources of knowledge, is rarely used to improve the filter feature selection approach. We theoretically and experimentally show that using the proposed framework, external feature group information can be incorporated into some filter feature selection algorithms with no additional computational costs. We also show that the new algorithm which exploits feature group information achieves higher classification accuracy than the original algorithm and other existing feature selection methods. Third, we propose a framework to incorporate feature group information from the external sources of knowledge into unsupervised feature selection methods. Due to the unavailability of the class labels, unsupervised feature selection is more challenging than supervised feature selection, yet has wide applicability in real world applications. We experimentally show that the proposed framework can be used to develop new unsupervised feature selection algorithms which achieve high clustering performance with low computational costs. As the current framework is limited to disjoint feature groups, in our fourth contribution, we extend it to exploit pairwise feature correlations as well, resulting in a more generic framework. Proposed group based and pairwise correlation based feature selection methods have wide applicability in areas such as genomic, text and image data analysis which involve datasets whose features show group behaviours and pairwise correlations.
-
ItemTowards Sensor-based Learning Analytics: A Contactless Approach( 2020)Learning analytics is an emerging field in which sophisticated analytic tools are used to improve students’ learning. Driven by the data from heterogeneous resources and the latest data mining techniques, learning analytics mainly attempts at creating a more integrated and personalized learning experience for each student. Most of the studies in the field rely exclusively on keyboard-mouse interactions. Though these interactions are effective in capturing overall sentiments and reactions, they do not provide enough granularity to conduct detailed analyses about students’ learning patterns. This dissertation will provide an interdisciplinary perspective on sensor-based learning analytics by discussing data-science techniques, HCI-based solutions in relation to educational theories. The thesis begins by providing an overview of the field of learning analytics and discusses the limitations of traditional contact-based sensor technologies in education research. We then discuss how data generated from contactless sensors such as an eye-tracker and a thermal camera can provide an informative window on students’ learning experiences. By conducting two lab-based studies, we analysed students’ eye movements and facial temperature to answer our four research questions related to predicting students’ desktop activities, evaluating learning task design, measuring students’ cognitive load, and measuring students’ attention patterns while students’ perform a learning task in a digital learning environment. The first study investigates the role of gaze-based features in predicting the desktop activities of the students. The outcomes of the study reveal that gaze-based features can be used to predict desktop activities of the students, and the design of a novel set of features (mid-level gaze features) can improve the accuracy of the prediction. The study lacks strong ground truth and contains a small sample size with only one sensor data. The limitations are addressed by the second study, which is a more extensive study and utilises a multi-sensor setup consisting of an eye-tracker, a thermal camera, a web camera and a physical slider. The study focuses on understanding students’ learning process in video-based learning. The data is analysed at different granularity, and results are present as separate chapters in the thesis. The first analysis compares two types of instructional designs of the video lectures (text vs. animation) using a physical slider. The results suggest that continuous evaluation of video lectures using a slider provides deeper insight into how students experience video lectures. To confirm these findings using a physiological marker, we again compare different instructional designs of the video lecture by measuring the cognitive load of the students using a thermal camera. The results suggest that thermal imaging is a reliable psycho-physiological indicator of cognitive load, and text-based video lectures induce higher cognitive load than animation-based video lectures. The final analysis focuses on measuring the coattention between students’ attention and instructor’s dialogues by measuring how much student follows instructor dialogues in a video lecture using an eye-tracker. The results suggest that students’ attention mostly follows the instructor’s dialogues on the screen, and their direction of attention is influenced by their prior knowledge. The thesis contributes to the fields of learning analytics by combining knowledge from different disciplines – HCI-design based on contactless sensors, video lectures derived from educational psychology, and novel data science techniques to extract features from eye movements and facial temperature to understand students’ learning process in a digital learning environment. The thesis’s final chapter discusses these contributions and provides future directions for sensor-based learning analytics using contactless sensors.
-
ItemTowards Developing Theoretical Foundations for Personalisation: Local Intrinsic Dimensionality Perspective( 2020)Personalised data analytics can make a significant improvement to the individuals' career, lifestyle, and health by providing them with an in-depth analysis of their profiles in contrast to other people. This thesis aims to obtain such personalised analytics by developing effective query-driven data-mining algorithms and strategies. However, existing research lacks in measures and strategies for the discovery of query aware patterns. State-of-the-art techniques mainly focus on mining patterns/clusters for illustration of underlying trends or association across the entire data rather than paying attention to an individual object of interest. The usefulness of feature space is typically evaluated in terms of its ability to preserve the global or/and local intrinsic structures of data without considering its relevance in relation to a specific object's interest, e.g., preserving its neighbourhood structure. Customisation to the general patterns according to the user given criteria is challenging, and often it is not possible to obtain the optimal or best solutions. In this thesis, we demonstrate the significance of personalised patterns over general patterns and identify the challenges of adapting existing research to the personalised case. We propose a novel query driven framework, \texttt{PRESS} that mines objects with subspace similarity for discrete data. Under this framework, we develop multiple algorithms by adopting a top-k pattern enumeration technique to mine personalised clusters, where the objects are homogeneous with respect to the query subspaces. We exploit the strictness and flexibility of query subspaces to effectively reduce the exponential search space. Our empirical case studies on real and artificial data showcase \texttt{PRESS}’s ability to identify cohesive groups, its scalability, and its interpretability while providing users with personalised feedback and recommendation. We conduct enhanced experimental and theoretical investigations on the query's local neighbourhood with respect to different size feature-set in order to get a better understanding of its latent behaviour in numerical data. The statistical modelling of real data enables us to identify the role of two key phenomena influencing the local complexity of neighbourhood surrounding the query object for different feature combinations: correlation and dominance. We develop theoretical foundations in the light of local intrinsic dimensionality (\texttt{LID}) model for assessing the usefulness of a feature space in relation to the query object’s interest. Our proposed LID oriented \texttt{dominance} measure (estimator) quantifies the degree of relevance of a feature space by its ability to preserve the local latent neighbouring structure of the given query object during the projection of data from higher to lower dimensional space. Such query aware local measure of feature relevance would facilitate the subspace learning process, i.e., feature selection, ranking and enumeration, of the query focused machine learning algorithms. It is worth mentioning that our latent analysis of individual characteristics not only benefits the personalised learning but could also enhance the general learning process.