Mining discriminative Itemsets in data streams

This paper presents a single pass algorithm for mining discriminative Itemsets in data streams using a novel data structure and the tilted-time window model. Discriminative Itemsets are defined as Itemsets that are frequent in one data stream and their frequency in that stream is much higher than the rest of the streams in the dataset. In order to deal with the data structure size, we propose a pruning process that results in the compact tree structure containing discriminative Itemsets. Empirical analysis shows the sound time and space complexity of the proposed method.

Developing a dashboard for real-time data stream composition and visualization

Technological advances have led to an influx of affordable hardware that supports sensing, computation and communication. This hardware is increasingly deployed in public and private spaces, tracking and aggregating a wealth of real-time environmental data. Although these technologies are the focus of several research areas, there is a lack of research dealing with the problem of making these capabilities accessible to everyday users. This thesis represents a first step towards developing systems that will allow users to leverage the available infrastructure and create custom tailored solutions. It explores how this notion can be utilized in the context of energy monitoring to improve conventional approaches. The project adopted a user-centered design process to inform the development of a flexible system for real-time data stream composition and visualization. This system features an extensible architecture and defines a unified API for heterogeneous data streams. Rather than displaying the data in a predetermined fashion, it makes this information available as building blocks that can be combined and shared. It is based on the insight that individual users have diverse information needs and presentation preferences. Therefore, it allows users to compose rich information displays, incorporating personally relevant data from an extensive information ecosystem. The prototype was evaluated in an exploratory study to observe its natural use in a real-world setting, gathering empirical usage statistics and conducting semi-structured interviews. The results show that a high degree of customization does not warrant sustained usage. Other factors were identified, yielding recommendations for increasing the impact on energy consumption.

New variational Bayesian approaches for statistical data mining : with applications to profiling and differentiating habitual consumption behaviour of customers in the wireless telecommunication industry

This thesis investigates profiling and differentiating customers through the use of statistical data mining techniques. The business application of our work centres on examining individuals’ seldomly studied yet critical consumption behaviour over an extensive time period within the context of the wireless telecommunication industry; consumption behaviour (as oppose to purchasing behaviour) is behaviour that has been performed so frequently that it become habitual and involves minimal intentions or decision making. Key variables investigated are the activity initialised timestamp and cell tower location as well as the activity type and usage quantity (e.g., voice call with duration in seconds); and the research focuses are on customers’ spatial and temporal usage behaviour. The main methodological emphasis is on the development of clustering models based on Gaussian mixture models (GMMs) which are fitted with the use of the recently developed variational Bayesian (VB) method. VB is an efficient deterministic alternative to the popular but computationally demandingMarkov chainMonte Carlo (MCMC) methods. The standard VBGMMalgorithm is extended by allowing component splitting such that it is robust to initial parameter choices and can automatically and efficiently determine the number of components. The new algorithm we propose allows more effective modelling of individuals’ highly heterogeneous and spiky spatial usage behaviour, or more generally human mobility patterns; the term spiky describes data patterns with large areas of low probability mixed with small areas of high probability. Customers are then characterised and segmented based on the fitted GMM which corresponds to how each of them uses the products/services spatially in their daily lives; this is essentially their likely lifestyle and occupational traits. Other significant research contributions include fitting GMMs using VB to circular data i.e., the temporal usage behaviour, and developing clustering algorithms suitable for high dimensional data based on the use of VB-GMM.

Adaptive unsupervised learning of human actions

Automatic detection of suspicious activities in CCTV camera feeds is crucial to the success of video surveillance systems. Such a capability can help transform the dumb CCTV cameras into smart surveillance tools for fighting crime and terror. Learning and classification of basic human actions is a precursor to detecting suspicious activities. Most of the current approaches rely on a non-realistic assumption that a complete dataset of normal human actions is available. This paper presents a different approach to deal with the problem of understanding human actions in video when no prior information is available. This is achieved by working with an incomplete dataset of basic actions which are continuously updated. Initially, all video segments are represented by Bags-Of-Words (BOW) method using only Term Frequency-Inverse Document Frequency (TF-IDF) features. Then, a data-stream clustering algorithm is applied for updating the system's knowledge from the incoming video feeds. Finally, all the actions are classified into different sets. Experiments and comparisons are conducted on the well known Weizmann and KTH datasets to show the efficacy of the proposed approach.

A suspicious behaviour detection using a context space model for smart surveillance systems

Video surveillance systems using Closed Circuit Television (CCTV) cameras, is one of the fastest growing areas in the field of security technologies. However, the existing video surveillance systems are still not at a stage where they can be used for crime prevention. The systems rely heavily on human observers and are therefore limited by factors such as fatigue and monitoring capabilities over long periods of time. This work attempts to address these problems by proposing an automatic suspicious behaviour detection which utilises contextual information. The utilisation of contextual information is done via three main components: a context space model, a data stream clustering algorithm, and an inference algorithm. The utilisation of contextual information is still limited in the domain of suspicious behaviour detection. Furthermore, it is nearly impossible to correctly understand human behaviour without considering the context where it is observed. This work presents experiments using video feeds taken from CAVIAR dataset and a camera mounted on one of the buildings Z-Block) at the Queensland University of Technology, Australia. From these experiments, it is shown that by exploiting contextual information, the proposed system is able to make more accurate detections, especially of those behaviours which are only suspicious in some contexts while being normal in the others. Moreover, this information gives critical feedback to the system designers to refine the system.

A comparison of methods to identify alcohol involvement in youth injury-related emergency department presentation data

Aims: To compare different methods for identifying alcohol involvement in injury-related emergency department presentation in Queensland youth, and to explore the alcohol terminology used in triage text. Methods: Emergency Department Information System data were provided for patients aged 12-24 years with an injury-related diagnosis code for a 5 year period 2006-2010 presenting to a Queensland emergency department (N=348895). Three approaches were used to estimate alcohol involvement: 1) analysis of coded data, 2) mining of triage text, and 3) estimation using an adaptation of alcohol attributable fractions (AAF). Cases were identified as ‘alcohol-involved’ by code and text, as well as AAF weighted. Results: Around 6.4% of these injury presentations overall had some documentation of alcohol involvement, with higher proportions of alcohol involvement documented for 18-24 year olds, females, indigenous youth, where presentations occurred on a Saturday or Sunday, and where presentations occurred between midnight and 5am. The most common alcohol terms identified for all subgroups were generic alcohol terms (eg. ETOH or alcohol) with almost half of the cases where alcohol involvement was documented having a generic alcohol term recorded in the triage text. Conclusions: Emergency department data is a useful source of information for identification of high risk sub-groups to target intervention opportunities, though it is not a reliable source of data for incidence or trend estimation in its current unstandardised form. Improving the accuracy and consistency of identification, documenting and coding of alcohol-involvement at the point of data capture in the emergency department is the most desirable long term approach to produce a more solid evidence base to support policy and practice in this field.

An hybrid approach for efficient multicast stream authentication over unsecured channels

We study the multicast stream authentication problem when an opponent can drop, reorder and inject data packets into the communication channel. In this context, bandwidth limitation and fast authentication are the core concerns. Therefore any authentication scheme is to reduce as much as possible the packet overhead and the time spent at the receiver to check the authenticity of collected elements. Recently, Tartary and Wang developed a provably secure protocol with small packet overhead and a reduced number of signature verifications to be performed at the receiver. In this paper, we propose an hybrid scheme based on Tartary and Wang’s approach and Merkle hash trees. Our construction will exhibit a smaller overhead and a much faster processing at the receiver making it even more suitable for multicast than the earlier approach. As Tartary and Wang’s protocol, our construction is provably secure and allows the total recovery of the data stream despite erasures and injections occurred during transmission.

Representing team behaviours from noisy data using player role

Due to their unobtrusive nature, vision-based approaches to tracking sports players have been preferred over wearable sensors as they do not require the players to be instrumented for each match. Unfortunately however, due to the heavy occlusion between players, variation in resolution and pose, in addition to fluctuating illumination conditions, tracking players continuously is still an unsolved vision problem. For tasks like clustering and retrieval, having noisy data (i.e. missing and false player detections) is problematic as it generates discontinuities in the input data stream. One method of circumventing this issue is to use an occupancy map, where the field is discretised into a series of zones and a count of player detections in each zone is obtained. A series of frames can then be concatenated to represent a set-play or example of team behaviour. A problem with this approach though is that the compressibility is low (i.e. the variability in the feature space is incredibly high). In this paper, we propose the use of a bilinear spatiotemporal basis model using a role representation to clean-up the noisy detections which operates in a low-dimensional space. To evaluate our approach, we used a fully instrumented field-hockey pitch with 8 fixed high-definition (HD) cameras and evaluated our approach on approximately 200,000 frames of data from a state-of-the-art real-time player detector and compare it to manually labeled data.

Comprehensive evaluation of driver gaze pattern using fuzzy rules for driver training

The over representation of novice drivers in crashes is alarming. Research indicates that one in five drivers’ crashes within their first year of driving. Driver training is one of the interventions aimed at decreasing the number of crashes that involve young drivers. Currently, there is a need to develop comprehensive driver evaluation system that benefits from the advances in Driver Assistance Systems. Since driving is dependent on fuzzy inputs from the driver (i.e. approximate distance calculation from the other vehicles, approximate assumption of the other vehicle speed), it is necessary that the evaluation system is based on criteria and rules that handles uncertain and fuzzy characteristics of the drive. This paper presents a system that evaluates the data stream acquired from multiple in-vehicle sensors (acquired from Driver Vehicle Environment-DVE) using fuzzy rules and classifies the driving manoeuvres (i.e. overtake, lane change and turn) as low risk or high risk. The fuzzy rules use parameters such as following distance, frequency of mirror checks, gaze depth and scan area, distance with respect to lanes and excessive acceleration or braking during the manoeuvre to assess risk. The fuzzy rules to estimate risk are designed after analysing the selected driving manoeuvres performed by driver trainers. This paper focuses mainly on the difference in gaze pattern for experienced and novice drivers during the selected manoeuvres. Using this system, trainers of novice drivers would be able to empirically evaluate and give feedback to the novice drivers regarding their driving behaviour.

An assessment system for evaluation of driving competencies

Automobiles have deeply impacted the way in which we travel but they have also contributed to many deaths and injury due to crashes. A number of reasons for these crashes have been pointed out by researchers. Inexperience has been identified as a contributing factor to road crashes. Driver’s driving abilities also play a vital role in judging the road environment and reacting in-time to avoid any possible collision. Therefore driver’s perceptual and motor skills remain the key factors impacting on road safety. Our failure to understand what is really important for learners, in terms of competent driving, is one of the many challenges for building better training programs. Driver training is one of the interventions aimed at decreasing the number of crashes that involve young drivers. Currently, there is a need to develop comprehensive driver evaluation system that benefits from the advances in Driver Assistance Systems. A multidisciplinary approach is necessary to explain how driving abilities evolves with on-road driving experience. To our knowledge, driver assistance systems have never been comprehensively used in a driver training context to assess the safety aspect of driving. The aim and novelty of this thesis is to develop and evaluate an Intelligent Driver Training System (IDTS) as an automated assessment tool that will help drivers and their trainers to comprehensively view complex driving manoeuvres and potentially provide effective feedback by post processing the data recorded during driving. This system is designed to help driver trainers to accurately evaluate driver performance and has the potential to provide valuable feedback to the drivers. Since driving is dependent on fuzzy inputs from the driver (i.e. approximate distance calculation from the other vehicles, approximate assumption of the other vehicle speed), it is necessary that the evaluation system is based on criteria and rules that handles uncertain and fuzzy characteristics of the driving tasks. Therefore, the proposed IDTS utilizes fuzzy set theory for the assessment of driver performance. The proposed research program focuses on integrating the multi-sensory information acquired from the vehicle, driver and environment to assess driving competencies. After information acquisition, the current research focuses on automated segmentation of the selected manoeuvres from the driving scenario. This leads to the creation of a model that determines a “competency” criterion through the driving performance protocol used by driver trainers (i.e. expert knowledge) to assess drivers. This is achieved by comprehensively evaluating and assessing the data stream acquired from multiple in-vehicle sensors using fuzzy rules and classifying the driving manoeuvres (i.e. overtake, lane change, T-crossing and turn) between low and high competency. The fuzzy rules use parameters such as following distance, gaze depth and scan area, distance with respect to lanes and excessive acceleration or braking during the manoeuvres to assess competency. These rules that identify driving competency were initially designed with the help of expert’s knowledge (i.e. driver trainers). In-order to fine tune these rules and the parameters that define these rules, a driving experiment was conducted to identify the empirical differences between novice and experienced drivers. The results from the driving experiment indicated that significant differences existed between novice and experienced driver, in terms of their gaze pattern and duration, speed, stop time at the T-crossing, lane keeping and the time spent in lanes while performing the selected manoeuvres. These differences were used to refine the fuzzy membership functions and rules that govern the assessments of the driving tasks. Next, this research focused on providing an integrated visual assessment interface to both driver trainers and their trainees. By providing a rich set of interactive graphical interfaces, displaying information about the driving tasks, Intelligent Driver Training System (IDTS) visualisation module has the potential to give empirical feedback to its users. Lastly, the validation of the IDTS system’s assessment was conducted by comparing IDTS objective assessments, for the driving experiment, with the subjective assessments of the driver trainers for particular manoeuvres. Results show that not only IDTS was able to match the subjective assessments made by driver trainers during the driving experiment but also identified some additional driving manoeuvres performed in low competency that were not identified by the driver trainers due to increased mental workload of trainers when assessing multiple variables that constitute driving. The validation of IDTS emphasized the need for an automated assessment tool that can segment the manoeuvres from the driving scenario, further investigate the variables within that manoeuvre to determine the manoeuvre’s competency and provide integrated visualisation regarding the manoeuvre to its users (i.e. trainers and trainees). Through analysis and validation it was shown that IDTS is a useful assistance tool for driver trainers to empirically assess and potentially provide feedback regarding the manoeuvres undertaken by the drivers.

Robust suspicious behaviour detection for smart surveillance systems

Video surveillance technology, based on Closed Circuit Television (CCTV) cameras, is one of the fastest growing markets in the field of security technologies. However, the existing video surveillance systems are still not at a stage where they can be used for crime prevention. The systems rely heavily on human observers and are therefore limited by factors such as fatigue and monitoring capabilities over long periods of time. To overcome this limitation, it is necessary to have “intelligent” processes which are able to highlight the salient data and filter out normal conditions that do not pose a threat to security. In order to create such intelligent systems, an understanding of human behaviour, specifically, suspicious behaviour is required. One of the challenges in achieving this is that human behaviour can only be understood correctly in the context in which it appears. Although context has been exploited in the general computer vision domain, it has not been widely used in the automatic suspicious behaviour detection domain. So, it is essential that context has to be formulated, stored and used by the system in order to understand human behaviour. Finally, since surveillance systems could be modeled as largescale data stream systems, it is difficult to have a complete knowledge base. In this case, the systems need to not only continuously update their knowledge but also be able to retrieve the extracted information which is related to the given context. To address these issues, a context-based approach for detecting suspicious behaviour is proposed. In this approach, contextual information is exploited in order to make a better detection. The proposed approach utilises a data stream clustering algorithm in order to discover the behaviour classes and their frequency of occurrences from the incoming behaviour instances. Contextual information is then used in addition to the above information to detect suspicious behaviour. The proposed approach is able to detect observed, unobserved and contextual suspicious behaviour. Two case studies using video feeds taken from CAVIAR dataset and Z-block building, Queensland University of Technology are presented in order to test the proposed approach. From these experiments, it is shown that by using information about context, the proposed system is able to make a more accurate detection, especially those behaviours which are only suspicious in some contexts while being normal in the others. Moreover, this information give critical feedback to the system designers to refine the system. Finally, the proposed modified Clustream algorithm enables the system to both continuously update the system’s knowledge and to effectively retrieve the information learned in a given context. The outcomes from this research are: (a) A context-based framework for automatic detecting suspicious behaviour which can be used by an intelligent video surveillance in making decisions; (b) A modified Clustream data stream clustering algorithm which continuously updates the system knowledge and is able to retrieve contextually related information effectively; and (c) An update-describe approach which extends the capability of the existing human local motion features called interest points based features to the data stream environment.

Supermassive

“Supermassive” is a synchronised four-channel video installation with sound. Each video channel shows a different camera view of an animated three-dimensional scene, which visually references galactic or astral imagery. This scene is comprised of forty-four separate clusters of slowly orbiting white text. Each cluster refers to a different topic that has been sourced online. The topics are diverse with recurring subjects relating to spirituality, science, popular culture, food and experiences of contemporary urban life. The slow movements of the text and camera views are reinforced through a rhythmic, contemplative soundtrack. As an immersive installation, “Supermassive” operates somewhere between a meditational mind map and a representation of a contemporary data stream. “Supermassive” contributes to studies in the field of contemporary art. It is particularly concerned with the ways that graphic representations of language can operate in the exploration of contemporary lived experiences, whether actual or virtual. Artists such as Ed Ruscha and Christopher Wool have long explored the emotive and psychological potentials of graphic text. Other artists such as Doug Aitken and Pipilotti Rist have engaged with the physical and spatial potentials of audio-visual installations to create emotive and symbolic experiences for their audiences. Using a practice-led research methodology, “Supermassive” extends these creative inquiries. By creating a reflective atmosphere in which divergent textual subjects are pictured together, the work explores not only how we navigate information, but also how such navigations inform understandings of our physical and psychological realities. “Supermassive” has been exhibited internationally at LA Louver Gallery, Venice, California in 2013 and nationally with GBK as part of Art Month Sydney, also in 2013. It has been critically reviewed in The Los Angeles Times.

A framework for cost-aware process management : cost reporting and cost prediction

Organisations are constantly seeking efficiency gains for their business processes in terms of time and cost. Management accounting enables detailed cost reporting of business operations for decision making purposes, although significant effort is required to gather accurate operational data. Process mining, on the other hand, may provide valuable insight into processes through analysis of events recorded in logs by IT systems, but its primary focus is not on cost implications. In this paper, a framework is proposed which aims to exploit the strengths of both fields in order to better support management decisions on cost control. This is achieved by automatically merging cost data with historical data from event logs for the purposes of monitoring, predicting, and reporting process-related costs. The on-demand generation of accurate, relevant and timely cost reports, in a style akin to reports in the area of management accounting, will also be illustrated. This is achieved through extending the open-source process mining framework ProM.