This page gives an overview of events and activities happening around the School of Computing and Mathematical Sciences.
News and Events
- Integration of Prediction and Rehabilitation for the Stroke, and Implementation with Software and Hardware System by Professor Zeng-Guang Hou (State Key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing)
1pm, 13 June 2013, WY315, Corner of Wakefield St and Mayoral Drive, Auckland CBD
Stroke and complications are leading reasons that cause nervous system damage for the middle-aged and senior. The patients might suffer permanent disability or casualty without effective treatment. It is important to predict and prevent a stroke using the traditional statistical methods integrating up-to-date computational intelligence techniques. For rehabilitation, at present, physical therapy, occupational therapy and exercise therapy are most popular clinical treatments, and they have been proven helpful to the recovery of patients’ nervous system and limb functions. However, the majority of rehabilitation hospitals still carry out the above treatments artificially or using simple rehabilitation medical devices. Since the training process for patients of neurological damage is repetitive, it is expected to improve the current status of rehabilitation by using robotics, and also it would accelerate the rehabilitation process for patients and reduce therapists’ labor intensity. This talk will mainly address the above issues which are the key points of a collaborative project conducted jointly by researchers at AUT and CASIA.
- 2013 NZ Game Developer’s Conference
10:00 am to 6:00 pm, 28 June 2013
WG403, WG Buidling, AUT City Campus | $70 student price
The New Zealand Game Developers Conference is a one-day action-packed conference for professional and indie game developers.
Video games continue to be a fast-growing export sector for New Zealand, with several Kiwi developers gaining global recognition. We've had 7 top 10 iPhone games in the last 9 months, not to mention PC and other smartphone successes. Kiwis game devs are at the top of their game, so come along and learn from some of the best.
Featuring leading New Zealand and Australian game developers from PikPok, NinjaKiwi, Gameloft, Cerebral Fix, Grinding Gear Games, TinMan Games, Halfbrick - plus a few surprise international guests.
All event details + registrations...
- 2013 MINA Symposium and Screening
20 - 22 November, AUT City Campus
This year CoLab and the School of Art & Design, and in conjunction with Massey University, is glad to announce our partnership with MINA (Mobile Innovation Network Aotearoa) for the 2013 Symposium & Screening.
The call has just gone out for papers and mobile phone films for the 3rd Mobile Creativity and Innovation Symposium, and 2013 Screening. The symposium will provide a platform for filmmakers, artists, designers, researchers, pro-d-users and industry professionals to debate the prospect of wireless, mobile and ubiquitous technologies in Arts and Design environments and the creative industries.
There will also be a mobile phone filmmaking workshop July 26 + 27 at CoLab.
Check out the MINA showreel
Registrations for this event will be available soon
An AUT University computer science team has won New Zealand’s first ever NASA Space Apps Challenge with a mobile phone application that allows users to share their sightings of the International Space Station. Read the full article...
As computing services are being distributed to growing numbers of users and service consumers, it has become increasingly important for IT professionals to understand how such systems work. Read the full article...
Auckland University of Technology announced it had graduated the first batch of eight students from its new Masters in Service-Oriented Computing programme. Read the full article...
Massive industry growth in the ICT/digital sector has highlighted a need for more professionals to meet future job demands. Read the full article...
Past News and Events
- Seminar: A Framework for Supporting Strict QoS Guarantee and Emergency Traffic in Wireless Local Area Networks: Research findings and Progress to date
4pm, 17 May 2013, WT126, AUT Tower, Wakefield Street
- Seminar: Bayesian inference for inverse obstacle scattering problems by Kate Lee
12.00 noon, 14 May 2013, WT126, AUT Tower, Wakefield Street
The Bayesian inference for inverse obstacle acoustic scattering problems is presented. The inverse obstacle scattering problem is to determine the image of an obstacle from measurement data of scattering eld at a large distance form the obstacle. Using these measurements statistical inference is performed to reconstruct the image of the obstacle. In Bayesian inversion, assuming Gaussian random noises to measurements the uncertainties in estimates or predictions are calculated using Markov chain Monte Carlo (MCMC) methods.
A heavy computational challenge often rises from a forward map calculation in which involves solving the boundary value problem (BVP) numerically. Various sampling techniques such as a Metropolis-Hastings algorithm, delayed acceptance MCMC and an adaptive MCMC are implemented and examined to increase the eciency of computational workload.
In numerical computations of a BVP using boundary element, the choice of a number of boundary elements is critical as it relates to the accuracy of BVP solutions and a computational workload. The ecient boundary discretization is considered borrowing a mean separation time of coupling Markov chains idea.
- Seminar: Rendezvous in Cognitive Radio Ad-Hoc Networks by Akbar Hossain
1pm, 10 May 2013, WT126, AUT Tower, Wakefield Street
- Seminar: Description Logic as a Programming Language by James Skene
12.30 noon, 7 May 2013, WY315, Corner of Wakefield St and Mayoral Drive, Auckland CBD
Description Logics (DLs) are a family of formalisms developed from earlier work on frame systems for Artificial Intelligence (AI) and currently most conspicuously employed in efforts to develop meta-data languages for the semantic web. The talk will give an overview of how sentences in a DL can be used as a programming language, why we might want to do this, and the challenges involved in making DLs a practical basis for programming. In the same way that predicate logic provided an inspiration for the development of Prolog, DLs can inform the design of a quite different programming system, one with a declarative object-oriented character and open-world semantics.
The talk represents a milestone of James research during the last several years. James will provide both theoretical and practical demonstrations that Description Logic as a programming approach works. These demonstrations will include properties of execution algorithms, a demonstration of an interpreter for a particular DL that James has developed and a discussion of the feasibility of the approach.
James hopes to get feedback from software engineers, practitioners and those with a special interest in mathematical logic, algorithmics and AI.
- Seminar: Object Identification from a Large Amount of Astronomical Data by Stuart Weston, Institute of Radio Astronomy and Space Research
12 noon, 16 April 2013, WT126, AUT Tower, Wakefield Street
Radio surveys to date such as the Australia Telescope Large Area Survey (ATLAS) have been used to identify objects for this investigation which contain 100s of identied objects, but with new instruments such as the Australian Square Kilometer Array pathnder (ASKAP) which is in the process of being commissioned and other larger survey's the number of objects identied will increase to the millions. One such future survey is the Evolutionary Map of the Universe (EMU), which will be a radio sky survey using the new ASKAP antenna array to make a deep radio survey of the Southern Sky. Due to size of the survey area and depth a very large number of new sources will be identied, it is estimated at about 70 million. The problem is how to identify the object types and provide their redshift, data from other wavelengths in dierent surveys, catalogues. Having this data we are faced with the dilemma of how to cross identify the 70 million sources with these surveys or follow up observations in other wavelengths in a automated pipeline before ant science can be started.
- Seminar: Capturing Recurrent Patterns in High Speed Data Streams by Sakthithasan Sripirakas
11.30am - 1.30pm, 9 April 2013, WT126, AUT Tower, Wakefield Street
Mining a time varying high speed stream is an interesting and a challenging problem in machine learning research. Many different concepts may appear at different times and the models built to learn and predict need to adapt in response of underlying changes. When prior knowledge on stream elements is not available, the models should recognize the change in an unsupervised way. In other words, concept changes should be detected automatically to update the models to cope with underlying changes. Previous research proposed a few change detection methods that demonstrate drift detection capability but with one or more of the following drawbacks: being computationally expensive, high memory consumption, high false detection rate and insensitive to drifts. In this research, we have proposed a superior drift detection mechanism compared to the current state-of-art drift detector ADWIN2. Identification of concepts values a system to learn recurrence of concepts over the time.
Capturing recurring concepts or patterns is another interesting and valuable theme in stream mining. Extracting knowledge on recurrent patterns minimizes or avoids the efforts to relearn and the delay in recognizing recurrent pattern. Meta-mining and statistical techniques will be used to analyse recurrence of concepts and to predict future concepts in a stream. This research addresses this problem using a forest of classifiers to archive the previous patterns for the future usage. A forest like structure with many classifiers such as decision trees consumes significant memory and computational overheads especially when used to capture recurrent concepts. Compression of decision tree is the next problem that this research addresses.
A very promising compression approach proposed in the literature constructs a Fourier Spectra of decision trees. This is not only a compact form but allows aggregation of many decision trees in Fourier space. Therefore, this research adopts this approach to compress the forest of decision trees to capture recurring patterns. In addition, this approach will be further extended to adapt Fourier Spectra in response to concept changes in the streams. In summary, this research is to model a system which can capture recurring patterns in a compact form by detecting concept changes automatically with lower memory and computational overheads.
- Seminar: Orion Health - Our Agile Journey by Nic Kenny and Rebecca Berrigan from Orion Health
12.30pm - 1.30pm, 9 April 2013, WY315, Corner of Wakefield St and Mayoral Drive, Auckland CBD
- Seminar: Data Quality Challenges in Empirical Software Engineering: An Evidence-Based Solution by Michael Franklin Bosu
9am - 11am, 27 March 2013, WT 609 AUT Tower, Wakefield Street
The management of software development and maintenance activities is informed by models that predict factors such as cost, schedule and quality. Managers therefore rely heavily on the accuracy of these models, and in turn, model accuracy is reliant on the data and methods used in their development. While extensive research has been conducted on issues of model development, the same cannot be said about the data that serve as input to these models. Multiple data quality challenges, such as noise, outliers, missing data, redundancy, small dataset size, and limited data accessibility, have been identified in regard to empirical software engineering (ESE) datasets.
Although a degree of research attention has been dedicated to resolving some of these problems, the consensus of the ESE community points to the inadequacy of effort in terms of sustained practice addressing data quality issues. Challenges in data collection, data pre-processing and data quality are generally not considered by most researchers and practitioners, hence the incidence of poor data quality which negatively affects the reliability of prediction models.
This thesis will address problems of data quality by employing a provenance solution. The notion of data provenance relates to the trustworthy and auditable recording of information regarding the source and processing of data. The intent is to follow a constructivist research approach to design, develop and evaluate a software system that is able to capture provenance information throughout the data collection and processing stages in the context of software engineering measurement.
The novel contributions of this thesis will be a provenance framework comprising the provenance system, a trust model, a data quality assessment model and an auditing model for empirical software engineering data.
The Role of Domain Knowledge and Cross-Functional Communication in Socio-Technical Coordination
- Seminars: The Role of Domain Knowledge and Cross-Functional Communication in Socio-Technical Coordination by Daniela Damian
Timely and Efficient Facilitation of Coordination of Software Developers’ Activities by Kelly Blincoe
1pm - 2.30pm, 26 March 2013, WY315, Corner of Wakefield St and Mayoral Drive, Auckland CBD
Software projects involve diverse roles and artifacts that have dependencies to requirements. Project team members in different roles need to coordinate but their coordination is affected by the availability of domain knowledge, which is distributed among different project members, and organizational structures that control cross-functional communication. Our study examines how information flowed between different roles in two software projects that had contrasting distributions of domain knowledge and different communication structures. Using observations, interviews, and surveys, we examined how diverse roles working on requirements and their related artifacts coordinated along task dependencies. We found that communication only partially matched task dependencies and that team members that are boundary spanners have extensive domain knowledge and hold key positions in the control structure. These findings have implications on how organizational structures interfere with task assignments and influence communication in the project, suggesting how practitioners can adjust team configuration and communication structures.
Timely and Efficient Facilitation of Coordination of Software Developers’ Activities
In large and distributed software development projects, work often needs to be divided among many developers. Those developers perform their work concurrently and technical dependencies arise between the various tasks. While steps can be taken to reduce these dependencies, they cannot be fully eliminated. When technical dependencies exist, developers may need to coordinate their work for a variety of reasons, for example, to prevent future integration problems. However, developers working on very large projects may not be aware of all of their coordination needs. There are methods that exist to detect coordination needs among developers by looking at the technical dependencies between tasks. However, those methods are limited in that they are not timely, and they risk overwhelming the developers by introducing too much noise since they enumerate all potential coordination needs. They do not make a distinction among those coordination needs that originate from critical as opposed to trivial inter-task technical dependencies. I propose a new method of detecting coordination requirements by looking at developer actions as they occur to detect coordination needs in real-time. Further, I propose using machine learning techniques to learn the true characteristics of dependencies that mandate coordination between tasks and the developers responsible for those tasks. Introducing these machine learning techniques can ensure that only the most crucial coordination requirements are brought to the attention of the developers. The proposed methods, therefore, can provide for timely coordination awareness without the risk of information overload.
- Seminar: "Embedded Ethnicity Inference in Fuzzy Logic System for DNA Profile Matching with Discrete Membership Function" by Reggio Hartono
1pm, 21 March 2013, WY315
- Seminar: The Unified Neutral Theory of Biodiversity by Robin Hankin
12 noon, 19 March 2013 WT126 AUT Tower, Wakefield Street
The distribution of abundance amongst species with similar ways of life is a classical problem in ecology. The unified neutral theory of biodiversity, due to Hubbell, states that observed population dynamics may be explained on the assumption of per capita equivalence amongst individuals. One can thus dispense with differences between species, and differences between abundant and rare species: all individuals behave alike in respect of their probabilities of reproducing and death. It is a striking fact that such a parsimonious theory results in non-trivial predictions, and even more striking that the theory predicts abundance curves that match observations across a wide range of ecologies.
In this short, non-technical talk, Robin will outline the Neutral Theory and show how it makes startlingly accurate predictions in the context of rainforest community ecology.
- Seminar: Adaptability in Streaming Pipelines for eScience with emphasis on the SKA
11am - 1pm, 15 March 2013, WT609, AUT Tower, Wakefield Street by Mahmoud Mahmoud
Large eScience research projects such as the Square Kilometre Array (SKA) or the Large Hadron Collider (LHC) place enormous demands on technology. Their planning, construction and operational lifetime can span decades during which technologies evolve substantially. As a consequence often such projects incorporate solutions that are obsolete by the time of deployment. In addition, scientific specifications and requirements can also change dramatically over time. Hence of benefit would be an infrastructure with architectural adaptability designed to minimize dependencies and mitigate technological evolution and changing requirements over time. Architectural adaptability means the infrastructure can be reconfigured to suit changing technology as well as changing specifications and mitigation of faults.
This research aims to improve the adaptability of eScience processing pipelines with particular emphasis on the SKA. Specifically the thesis investigates the adaptability of middleware for implementing SKA pipelines. A suitable set of fundamental pipeline operations will be determined. A thorough exploration will be conducted to evaluate an optimal initial mapping for pipeline operators as well as dynamically adapting the mapping towards optimality in the face of varying workloads, requirements and faults.
- Seminar: Intelligent Surveillance by Dr Wei Qi Yan
4pm - 5pm, 14 March 2013, 723 Meeting Room, Tamaki Campus
(Entry via Gate 1, South Entrance, Morrin Road, Glen Innes)
Surveillance is being increasingly used for traditional and non-traditional security applications such as monitoring of shopping malls and ATMs as well as industrial supervisory use. The decreasing costs coupled with rapid miniaturization of video cameras have enabled its widespread use on highways, airports, railway stations and on-board vehicles. The recent trend of coupling video cameras to cell-phones will only accelerate this trend. Therefore, research in video surveillance is moving into the mainstream with the focus on day-to-day applications and uncontrolled outdoor scenarios. And it is moving away from mere data collection with manual observation to intelligent analysis of events and actions at a semantic level without the intervention of humans. In this talk, Dr Yan will introduce intelligent surveillance from practice to theory, and then applying the theory to practical applications, especially the applications of AI technologies in surveillance computing at object and event levels, the talk will emphasize the computing on observation, learning, presentation and reasoning.
- Seminar: SAS in New Zealand: Applications and What the Market Looks Like for Future Graduates by Andrew Cathie
4pm - 5.30pm, 11 March 2013, WF202, Business Building, Corner of Wakefield Street and Mayoral Drive, Auckland CBD
SAS has operated in NZ for over thirty years, and has built up a diverse customer base. Recent trends in industry have emphasized the importance of analyzing data and it’s fair to say that this trend is real. However, what does this mean for people contemplating being part of this trend? What skills do they need to develop? What are the near term developments that are likely to affect graduates in the NZ market? This talk will address and highlight some of the issues for those contemplating a future in analytics.
Andrew Cathie is an analytics solutions consultant who works in the SAS Auckland office. He specializes in analytic applications of SAS, such as forecasting, predictive modeling and optimization. He has many years of experience variously working for software vendors, consulting companies and industry. Originally graduating in Computer Science and Operations Research, he is continuing as a student and is currently studying for a Masters in Applied Statistics. Andrew Cathie is a member of the Advisory Board for the School of Computing and Mathematical Sciences
- An Efficient, Secure, and Scalable Key-agreement Scheme for Sensor networks by Mee Loong (Bobby) Yang
12noon - 1pm, 27 February 2013, WS101, Level 1, WS Building, St. Paul St. Auckland CBD
This talk presents a new key agreement scheme for use among low resourced devices such as wireless sensor nodes. It is fast, secure and scalable for large networks. Unlike schemes based on public key cryptographic which rely on complicated operations on large integers of a thousand or more bits, this new scheme uses simple arithmetic operations on a large number of small integers, each of 5 bits or less.
This gives rise to an important side-effect - the obscurity of the private keys stored in the nodes. Even if an attacker obtains all the keys from captured nodes, it will be shown that the keys cannot be used directly, and considerable effort and resources would be require to first break this obscurity. Some attack strategies and procedures to do this are analysed and demonstrated. The results of computer simulation attacks showed that by choosing suitable keying parameters, the attacker would either require an excessive amount of computing resources, or would need to capture a very large number of nodes to break the system.
In spite of the small-integer operations, suitable choice of the keying parameters makes it usable for large networks with thousands of nodes.
- Research Seminar: Spatial-Temporal Data Representation and Processing in Ontology-Based Systems for Personalised Decision Support by Muhaini Binti Othman
10am - 12noon, 27 February 2013, WT608, AUT Tower
This research focuses on designing a novel framework for personalized decision support system and subsequently develops an application that employs discrete and spatial-temporal data for personalized knowledge representation and prognosis. The framework aims to unify machine learning approach and ontology knowledge representation approach for a better decision support. The main elements are the ontology-based system and personalized modeling engine. The relation between these two elements performs as an evolving system that interacts actively and harnesses the knowledge from each other hence enriching the knowledge based by discovering new knowledge. The ontology-based system in this study aimed to model space and time contextually together with the entities and its behaviors for better comprehension of the whole problem. The personalized modeling engine is a process of model creation for a single input vector in a problem space based on nearest neighbor spatial-temporal data and information available in the ontology-based system. The first part of this study will focus on the development of the personalized modeling engine. Several methods for personalized modeling will be investigated including conventional methods, such as KNN and WKNN. Also new methods in SNN could be utilized to process spatial-temporal dataset such as ESNN. Throughout this study the learning capability of the proposed framework will be improved by varying the reservoir computing techniques that is incorporated with ESNN methods such as EESNN, reSNN and NeuCube module. The case studies chosen for the proposed framework implementation focuses on stroke related problems for prevention and rehabilitation purposes. Result from preliminary experiments show promising potential to move forward as contribution to bio-informatics field.
- Seminar: A Review of Methodologies and Empirical Results from Field Studies Conducted by Software Engineering Global interAction research Lab (SEGAL) by Daniela Damian
12.30pm, 26 February 2013, WY315
In this talk Daniela will present methodologies and empirical results from field studies at large software organizations in which they investigated the communication and coordination in projects beyond the technical dependencies found in software code. Most of the existing knowledge they have about coordination in software projects is about developers managing their work as informed by dependencies in the code that they produce. However, software projects are driven by many other dependencies that are not of technical nature, and most often guided by the project requirements. This complex process involves parties other than developers, such as clients, analysts, designers and testers, and their effective coordination is crucial to project success.The multi-method research approaches combine methods of network analysis, social network surveys, qualitative observations in software projects, data mining as well as automated classification of online communication in large software repositories. Their work reveals patterns of requirements-driven communication in globally distributed software teams, the role of brokers in project communication, as well as the application of automated support to identify problematic requirements from online project communication.
Daniela Damian is an Associate Professor in the Department of Computer Science, where she leads research in the Software Engineering Global interAction Laboratory (SEGAL, segal.uvic.ca). Her research interests include Software Engineering, Requirements Engineering, Computer-Supported Cooperative Work and Empirical Software Engineering. In Requirements Engineering she has studied aspects of the communication and collaboration involved in the definition, negotiation and management of requirements throughout the project lifecycle. Specifically, recent studies investigated the information flow between the various stakeholders involved in the requirements definition and management, and modeled their interaction using techniques from social networks analysis. Daniela has also studied the interplay of the social and technical aspects of developers’ coordination in large, geographically distributed software projects, in an attempt to model the relationship between socio-technical congruence and project success. Daniela’s research methodologies involve extensive field work and studies of developers in-situ through collaborations with a number of industrial partners such as IBM, General Motors, Siemens and Dell. Her work includes both qualitative and quantitative methods, as well as techniques for data mining large software project repositories. She is funded by IBM, General Motors and the Canadian funding agency NSERC.
Daniela has served on the program committee boards of several conferences including the International Conference on Requirements Engineering, Software Engineering and Foundations of Software Engineering, as well as the CHASE workshop at ICSE conferences. She was the program co-chair for the First International Conference on Global Software Engineering (ICGSE06), and a guest editor of the IEEE Software Special Issue on Global Software Engineering (Sep/Oct 2006). She had served on the Steering Committee for the Conference on Global Software Engineering. She is currently on the editorial boards of the Journals of Requirements Engineering, is the Requirements Engineering Area Editor for the Journal of Empirical Software Engineering, and the Human Aspects Area Editor for the Journal of Software and Systems (starting 2013).
Learn more about Artificial Intelligence from a number of researchers from AUT, University of Auckland, University of New South Wales and University of Wollogong.
AUT is proud to be hosting the Association for Computing Machinery's (ACM) 8th International Computing Education Research Conference. The conference is preceded by a Doctoral Consortium on Sunday, 9th September.
Our team is a strong contender for the cup! They are also in the running for the People's Choice Awards so lend them your support by voting
for them now!
AUT-led Team Mobile Eye has won the 2012 Microsoft Imagine Cup.
The delicate art of connecting theory and practice: A variation theory study in programming education
Dr Anna Eckerdal, Uppsala University
Click here for more information
Eliciting Some of the Different Types of Reasoning Skills Used by the Novice Programmer
Dr Raymond Lister, University of Technology Sydney
Click here for more information
Addressing a decade-old cosmic mystery: observation and analysis of radio recombination lines in spectra of galactic nebulae.
Read abstract and biography
Dr George Allan
Collaborative software development and social networks in action: A framework and experiences from studies of collaboration and project outcomes in global software teams
Dr. Daniela Damian
Software Engineering Global interAction Laboratory, Department of Computer Science at the University of Victoria, BC, Canada (Email: firstname.lastname@example.org )
Read abstract and biography