Computing & Mathematical Sciences Events

Mathematics Seminar: New Classes of Priors based on Stochastic Orders: Theory and Applications in Reliability 3rd Mar 2023

In the context of robust Bayesian analysis, we first introduce a new class of univariate prior distributions based on stochastic orders and distortion functions. Then we introduce a new class of multivariate priors based on stochastic orders, Multivariate Total Positivity of order 2 (MTP2) and weighted distributions.

We provide the new definitions, their interpretation and the main properties and we also study the relationship with other classical classes of prior beliefs. We also consider metrics (Kolmogorov and Kantorovich in the former case, Hellinger and Kullback-Leibler in the latter) to measure the uncertainty induced by such classes.

Finally, we present the application of the former class in the context of fault tree analysis for a spacecraft re-entry example, whereas the latter will be illustrated with an example about train door reliability.

SE Seminar: Proving Ethereum Smart Contracts correct using Vale, F* and Z3 21st Feb 2023

We explore the use of Microsoft’s Vale framework to guarantee low level correctness of Ethereum Virtual Machine (EVM) bytecode, while affording smart contract developers higher-level language and reasoning features. We encode EVM-R (a subset of EVM semantics and instruction set) into F*, and raise the EVM-R into Vale design-by-contract components in an intermediate language supporting conditional logic. The specifications of Vale procedures constructed from these verified EVM bytecodes carry integrity to the bytecode level, which cannot be guaranteed by current EVM compilers. Furthermore, raising the instruction set to Vale allows opportunity for refinement of the instructions, which we did ensuring safety properties of overflow protection, invalid memory access protection, and functional correctness. We demonstrate our contributions through two case study smart contracts, a simple casino, and a subcurrency coin.

Mathematics Seminar: Mutually Orthogonal Frequency Rectangles 1st Feb 2023

A frequency rectangle of type FR(m, n; q) is an m × n matrix such that each symbol from a set of size q appears n/q times in each row and m/q times in each column. Two frequency rectangles of the same type are said to be orthogonal if, upon superimposition, each possible ordered pair of symbols appear the same number of times. A set of k frequency rectangles in which every pair is orthogonal is called a set of mutually orthogonal frequency rectangles, denoted by k–MOFR(m, n; q).

In this talk, I will discuss the application of frequency rectangles in experimental designs and their connection with orthogonal arrays and Hadamard matrices. I will also describe a stronger form of orthogonality for a set of frequency rectangles. A k–MOFR(m, n; q) is said to be t–orthogonal, if each subset of size t, when superimposed, contains each of the q^t possible ordered t-tuples of entries exactly mn/q^t times. A set of vectors over a finite field F_q is said to be t-independent if each subset of size t is linearly independent.
I will discuss a method to obtain a set of t–orthogonal k–MOFR(q^M, q^N, q) corresponding to a set of t–independent vectors in (F_q)^M+N. Then I will talk about some bounds and known values in the literature on the size of a set of t–independent vectors.

A frequency rectangle of type FR(n, n; q) is called a frequency square. For p an odd prime, I will describe a construction for a set of (p − 1) binary MOFS of order 2p. This improves the lower bounds in (Britz et al. 2020) for the previously known values for p ≥ 19.

Mathematics Seminar: Interactions in oscillatory systems far from equilibrium 24th Nov 2022
Living systems produce their energy by themselves. In this process they depend on energy and matter delivered form the environment and their ability to remove the waste products of their metabolism, which makes them thermodynamically open systems. This means that they are characterised by multiple interactions - between themselves and with the environment. Their dynamics is also nonlinear. However, most of the models we use to model living systems are linear and treat them as isolated systems. In this talk I will argue that it is much simpler to treat them as non-autonomous, nonlinear and open systems than to try to adapt our current models and extend them to fit the main characteristics of living systems. Practical examples of living systems – related to cell dynamics and circadian rhythms, as well as condensed matter physics – will be used to illustrate how the multiscale dynamics of non-autonomous systems can be rescued from the common current perception of being noisy or chaotic.

PhD Students' Conference 2022 18th Nov 2022
CMS PhD students present their current research to the School.

Industry Talk: Secrets to Scoring your Dream Job 23rd Sep 2022
Applying for jobs and interviewing after graduating can be a daunting task. What are hiring managers looking for in your CV and cover letter (here’s a tip always include a cover letter) and how can you stand out from the crowd? How can you apply for roles when you have little or no relevant experience in the industry? New Zealand Herald Development Manager Tracey Bond will explain what makes a successful candidate and share some tips on how to make sure you land an interview as well as explain some of the classic mistakes to avoid.

In this session we will cover

• Why a cover letter is important and which phrases you should avoid
• Basic CVs dos and don’ts
• The skills you need to ace an interview

Student Research Conference 2022 30th Aug 2022
Computer Science students will present their final year honours projects. Family, friends and fellow students are welcome. 3rd year students considering honours are encouraged to attend the conference.

Industry Talk: Cyber Security - Ask us Anything! 26th Aug 2022

We are honoured to host a panel of Cyber Security experts in this week's Industry Talk series. Entitled "Cyber Security: Ask us anything!". Each panelist will begin with a short intro, before opening up the Zoom audience for questions.

Panelists: 
Stephen Coates - Senior Security Consultant (Aura's Advisory Team, Auckland)
Tyale Rodrigues - Security Consultant (Aura's Advisory Team, Auckland)
Srikar Vema - Security Consultant (Aura's Advisory Team, Auckland)
Piyush Sharma - Junior Security Consultant (Aura's Advisory Team, Wellington)
Lachlan Davidson - Security Consultant (Aura's Penetration Test Team, Auckland)

Moderator:
Dr Vimal Kuman -- Head, Cyber Security Researchers at University of Waikato

Industry Talk: Quality Assurance in an Agile Environment 27th May 2022
It is my belief that everyone owns quality in an organisation—project managers, product owners, developers and quality assurance (and a whole many more people). The key is open communication between stakeholders in the product. Organisations that facilitate this communication will more likely have reduced instances of issues in production causing embarrassment. The cost of remediating a defect found increases the further along the line that it is found; by the time a bug has hit production, remediation is expensive not only from a financial but also from a reputational standpoint. The earlier the defect is found, the better the outcome. The quality assurance staff are not the only and last line of defence. I like to think that everyone on the squad is on attack and defence!

The presentation discusses the day to day of test analysts and software engineers in test in an Agile environment and what expectations are in today’s Agile squads. Today’s quality assurance staff are critical not just in performing test responsibilities but they also steer and develop process to improve the product. Responsible quality assurance staff are consistently feeding back information and opinions to development and product/project management that might decide whether a release might take place or that further development might need to be done.

Industry Talk: Modern Aspects of Industrial Cyber Security / First Watch Approach 13th May 2022
The goal of hackers is to take control of an organisation’s resources. Depending on their motivation, they will either sell acquired credentials on the Dark Web or use them to encrypt computers and extort money, as in case of Waikato District Health Board in 2021. Critical infrastructure organisations attract special interest from hacking groups due to the scale of the potential negative outcomes. State-sponsored teams actively try to conduct reconnaissance attacks to gather sensitive information and use it to achieve control over a critical resource. Causing harm to the population of a town, city, or large geographic area by poisoning the drinking water supply is a high impact example of a critical infrastructure attack. In the case of a Florida Treatment Plant in 2021, an outdated version of Windows (operating system) and a weak cyber security network allowed access to the treatment plant’s computer system and the manipulation of chemical levels in the drinking water which if not detected would have been catastrophic for the local population.

Cyber attacks have reportedly increased across OT/ICS organisations by 80% in 2021 (Claroty report). Recovery from a cyber-attack is estimated to cost between USD3.2m and USD4m (IBM 2020 estimates) on top of any ransomware costs paid and the risk of national infrastructure shutdowns, economic production and risk to life and wellbeing.

In the light of recent significant cyber-attacks, it’s obvious that cyber security must be addressed in any organisation. The question is where to start?

Industry Talk: Empowering people and business through data science in Genesis Energy 17th Sep 2021
Data can help business gain valuable insights such as better predicting demand, more accurate sales forecasts, predicting maintenance time for power plant equipment, and so on. But there are limits to what numbers and charts can do on their own; Opportunities or answers to business problems don’t just spring from algorithms, they need people to translate the information into compelling stories. So in order to empower people and the business through data science, I propose a number of steps to unlock the little data scientist living inside each of us.

ORKA Seminar: Making a Space Train - A talk about adaptive systems in interactive media 14th Sep 2021
The field of interactive media is defined in part by its multidisciplinary requirements. From audio to character art, gameplay systems to narrative, games depend on a range of people with a wide range of skills. The work of many of these disciplines can benefit, either now or in future, from adaptive systems.

In this talk, I’ll discuss an overview of game development from an adaptive systems standpoint. We’ll cover the problems, solutions, and opportunities in the spaces of AI, gameplay, and real-time graphics, as well as random parts of the pipeline that I happen to find interesting.

ORKA Seminar: Automated Retrofit of Heat Exchange Networks 7th Sep 2021
With increasing legislative and societal pressures placed on industry in response to climate change, increased efficiency of energy usage in the form of process integration is an area of growing interest and active research. One specific area wherein potential for increases in efficiency and consequent energy savings has been identified is that of Heat Exchange Networks. Such networks are common through industrial settings, and represent a significant portion of energy usage. While methods such as Pinch Analysis have been applied to such systems for the purpose of retrofitting for reduced energy wastage, there remains a lack of general tools and automated solutions that can be easily applied to derive solutions across different operational contexts. As such, the exploration of existing approaches and the development of accessible tools is of great relevance for industries that find themselves facing necessary change.

ORKA Seminar: Deep Learning for Intelligent Surgical Tool Management 17th Aug 2021
We develop a multi-level Deep Learning prediction framework which provides a viable solution for intelligent management of surgical tools in a hospital. We present a new hierarchical organised surgical tool dataset, HospiTools, for classification and retrieval of surgical tool images. In addition to 360 tool classes, we create a four level hierarchical structure for our dataset defined by 2 specialities, 12 packs and 35 sets. We employ different convolutional neural network training strategies to evaluate performance on this dataset, including the provision of prior information in the form of a taxonomic hierarchy tree structure. We demonstrate that the hierarchical nature of surgical tool classes could be used to make improved predictions. Our solution can potentially lead to greater patient safety, significant cost savings and increased efficiencies across New Zealand DHBs.

Statistics Seminar: Revisiting the Sins of My Past 13th Aug 2021
In this talk, I will discuss how some of the work that I did as a PhD student, has eventually (or in some cases continually), been revisited by other researchers. My PhD was centred on problems arising from the statistical interpretation of forensic glass evidence. This is not something that most people are familiar with, so I will start with an actual case to introduce the general methodology, and then highlight some issues that have arisen from my research on that methodology.

A friend of mine once said, "Anything new in this talk is a typo." I add to this my corollary "But you might not have heard it before."

ORCA Seminar: Model Predictive Control (MPC) 9th Feb 2021
In spite of its long history of success as a powerful and versatile control technique, the model predictive control (MPC) has continued attracting strong interest from both industry and academia, and being utilized across a broad range of application domains. The attention has been mainly focused on computational efficiency, closed-loop performance, and the time required to develop an MPC solution by constructing suitable prediction models. This presentation aims at providing a foundational treatment of the theory and design of MPC. This includes the theoretical background that leads to successful implementation of MPC in practice, the main elements that appear in any MPC formulation, the state-of-the-art MPC algorithms, as well as a summary of key advantages and potential challenges. The last part of the talk will be dedicated to the application of MPC in diverse fields where several use cases will be presented including some from the speaker’s current research in nanomedicines.

ORCA Seminar: Multi-Agent Deep Reinforcement Learning for Request Dispatching in Software-Defined Networks 26th Jan 2021
Recently, distributed controller architectures have been quickly gaining popularity in Software-Defined Network (SDN). However, the use of distributed controllers introduces a new and important Request Dispatching Problem with the goal for every SDN switch to properly dispatch their requests among all controllers so as to optimize network performance. This goal can be fulfilled by designing a dispatching policy to guide request distribution at each switch. Existing policies are designed under the assumption of a single agent and fully observable environment which cannot always be satisfied in practice. In this talk, we present a Multi-Agent Deep Reinforcement Learning (MA-DRL) approach to automatically design policies with high adaptability and performance. This is achieved through a new problem formulation in the form of a Multi-Agent Markov Decision Process, a new adaptive policy design, and a new MA-DRL algorithm. Extensive simulation studies show that our MA-DRL technique can effectively train policies to significantly outperform man-made policies, model-based policies, as well as policies learned via single-agent DRL algorithms.

Seminar: Your app is tracking you: zero-permission user tracking 11th Dec 2020
Mobile apps collect information from users' mobile phones. They use such information to build profiles of their users, analyze their behavior and preferences so as to serve targeted advertisements and derive greater profit, regardless whether the user is willing or not.

Android OS adopts a permission model for the information that an app can obtain. Explicit permission from the user must be obtained before the app can obtain that information. However, we discovered a covert channel where an app can obtain users’ location information without users’ permission or even notice. Our findings show that there are flaws in the Android’s permission model that should be revised or be replaced to better protect users’ privacy.

Seminar: AI and the Future of Finance 30th Nov 2020
Join us as we bring together lead researchers, data practitioners and business leaders to talk about not just these opportunities but also the need for increasing awareness of the potential and for driving greater adoption of machine learning and artificial intelligence in the financial services industry.

Seminar: Reconsidering Sustainability for HCI 11th Feb 2020
For the past 10 years there has been a burgeoning discourse on sustainability concerns within HCI: including themes around how to design interventions that can support us in preserving energy, preserving our natural environment, and making our buildings and food chain more efficient. While these sustainability concerns are very much still at the forefront of research and policy, an equally important sustainability endeavour is starting to emerge concerning the role of technologies in preserving our communities and especially those amongst us that are at risk the most.

Seminar: T-codes, T-complexity, T-information and T-entropy - theory and applications 1st Nov 2019
This talk is my "theory turf", so to speak. It looks at T-codes, a family of statistically self-synchronising variable-length codes with a highly recursive hierarchical construction. The number of (weighted) construction steps for the code lends itself as a natural measure of code complexity: It is literally the number of bits required to address an arbitrary internal node in the tree. But T-codes also have a number of other rather intriguing properties. For one, it is possible to transform any finite string into a unique T-code and vice versa. This allows us to define a string complexity by simply using the aforementioned complexity for the code that the string transforms into. This string complexity estimate, T-complexity, and its linearised version, the T-information, whose gradient is known as the T-entropy, have been shown to align well empirically with both Shannon entropy and the Lempel-Ziv production complexity. We have used them, among others, for network event detection and in steganalysis. Another property of T-codes is their relationship with cyclic equivalence classes: Every T-code codeword belongs to its own cyclic equivalence class, and no T-code can contain two codewords from the same cyclic equivalence class. Moreover, the construction process of any T-code guarantees that any given non-periodic cyclic equivalence class is either represented by a codeword, or was represented by a codeword until that codeword was used during the construction, or can be represented by a codeword produced through further extension of the code. There is also a clear distinction on how codewords from periodic and non-periodic cyclic equivalence are generated. This links cyclic equivalence classes more generally to code synchronisation than was the case beforehand, when they were associated with fixed-length comma-free codes only. For our work, the link to cyclic equivalence classes permitted us to solve a number of problems in the asymptotics of T-complexity, lending an asymptotic rationale to the linearisation of T-complexity into T-information.

Seminar: Learning to Reason: from Question Answering to Problem Solving 17th Oct 2019
Recent advances in Machine Learning applied to Natural Language Processing have resulted in systems with quite impressive scores on Question-Answering tests in text and simple visual domains. Similarly, Google Assistant and other similar systems are performing quite well in answering real questions by answer extraction. This progress is real, but it is limited in some important respects: the systems typically have worked by identifying a passage span to serve as an answer; more recently, in "Multi-hop" QA, they have started to work by forming a short linear chain of extracted relations from question to answer. This falls significantly short of human-problem solving, including question-answering: it does not recursively decompose problems for solution, it does not follow that decomposition to assemble answers, and it does not store and apply salient background knowledge for decomposition, partial solution, or answer composition. Although it's at the early stage, the aim of our Broad AI Lab at the University of Auckland School of Computer Science is to learn from the far-from general capabilities of symbolic AI to extend our reach, especially in problem solving over text, by applying learning to bridge these current short-falls. In this seminar, I'll characterise the problem-solving problem and the baseline state-of-the-art, describe some preliminary work done with the Learning to Reason team at IBM Research, and sketch a programme towards broader, learning-based, quasi-symbolic AI. We hope this programme will extend the reach of AI-based problem solving, and especially question-answering.

Industry Talk: All suggestions welcome 11th Oct 2019
The talk specifically covers the process Origins used to design and roll out the ApiTrak system, a cloud-based software for the honey industry. It also covers an overview of the ApiTrak system, the reasons why particular hardware and software were chosen, as well as, reasons why it is essential for senior management who may not have a technical background to involve technical staff in all aspects of the design.

Seminar: Evaluation Evaluation 10th Oct 2019
Poor choices of evaluation measures is holding back AI and leads to biased easily-spoofed classifiers. I am interested in biasing toward recognizing the “right” kind of feature and ignoring the “wrong” kind of artefact - and understanding the nature of bias and in particular, when a human-like bias is desirable, or a different bias is more appropriate. In the 1990s a student and I published a CoNLL paper demonstrating of how tuning to optimize accuracy and F-measure was making things worse not better in the NLP context, following which I derived an empirical multiclass evaluation method for estimating Informedness, the probability that you are making an informed decision rather than guessing.

In the dichotomous case informedness turns out to be equivalent to Pearce’s I from the 1890s and Youden’s J from the 1950s, and has also been reinvented as DeltaP' and deskewed WRAcc, the former related to Matthews Correlation, the latter derived from ROC.

In the balanced multiclass case, Informedness corresponds to the “correct” way of marking multiple choice exams, so that guessing gives you zero.

My subsequent work has gone in two directions, modifying learning/boosting algorithms to optimize informedness, and looking at the relationship between informedness and other popular measures (including F-measure, Kappa and ROC AUC). Informedness has now been demonstrated to give a better and more intuitive idea of how good a system is across a wide range of AI and CI technologies and applications.

Industry Talk: Farmax – Moving from 1990s to 2020s 27th Sep 2019
FARMAX, built by Farmax Limited, is an application that uses science-based system modelling to help farmers see the likely outcomes of various farm planning decisions. It was built to maximise profits in a simpler time, but it has found fresh relevance in a world where farmers must stay profitable while managing a slew of environmental responsibilities. FARMAX has enormous potential to contribute to kaitiakitanga – the responsible stewardship of land and other resources. But in its present form it will struggle to respond to all the new opportunities and challenges. FARMAX was built as a single user Win32 application in the mid 1990s. A team of just 3 is charged with taking this aging and brittle value and transforming it into a scalable cloud service with modern integration points and user experience.

Design Seminar: How to improve character design through costume thinking 20th Sep 2019
Heli Salomaa is a Costume Artist specialized into costumes in games and other digital fields. She works as a speaker and a guest lecturer at Universities, conferences and events presenting the benefits of acknowledging the narrative potential of costumes in character design. Her latest project was Remedy Entertainment's upcoming game Control on which she worked for two years, creating concept art and manufacturing the characters' costumes. She has designed costumes for physical bodies in 35 productions of theatre, dance, circus and performances, and digital bodies for animation, VR and video games. Some productions are displayed on her website www.helisalomaa.com

Industry Talk: Healthcare - A Changing World 13th Sep 2019
Healthcare is an extremely complex industry, the art of trying to combine the human element of healthcare with the latest technology has always been challenging. In this talk Sam Mackenzie will discuss some of the human elements that are affecting implementation of IT for the benefit of general health outcomes while also being a prudent use of the taxpayer dollar. Sam will also discuss the ever-increasing importance of data and the pitfalls of collecting incorrect or corrupt data for analysis. There is often a need for a clinical understanding at the base level of the data, knowledge of multiple computer languages in order to accumulate the data correctly and then an understanding of what information is required at the other end. Often, we find ourselves trying to solve answers to questions our clients didn’t even know they had. If you have a passion for health and want to see how ICT is changing the sector for good, then join Sam in an interactive discussion highlighting the direction this industry is taking.

Industry Talk: Tackling Real-World Complex Problems - A Systems Thinking Approach 6th Sep 2019
The world is faced with many complex problems (climate change, world poverty, terrorism, drug abuse); problems that are very difficult to solve, and can only be addressed using a multidisciplinary approach. In this talk, Nick Pickering will present a recent real-world disaster case study, and introduce the holistic systems thinking approach employed in the solution; systems thinking is a tool which can be applied in multiple domains of knowledge.

Seminar: Efficient Parametric Model Checking Techniques Using Domain Knowledge 23rd Aug 2019
In this talk I present an efficient parametric model checking (ePMC) method for analysing reliability, performance and other QoS properties of service-based systems (SBSs). ePMC speeds up the analysis of parametric Markov chains modelling SBSs by exploiting domain-specific service invocation patterns of operations using functionally-equivalent services. ePMC precomputes closed-form expressions for QoS properties of such patterns to use in the analysis of whole-system models, substantially reducing analysis time compared to state of the art parametric model checking methods.

Industry Talk: Robotic Process Automation (RPA) in businesses, and beyond 16th Aug 2019
The talk will cover what Robotic Process Automation (RPA) is, and how it is employed in businesses. The VexeBot has been developed to harness the functionality of Alexa, and takes RPA a step further. VexeBot automatically runs tasks and reports like any other person, or seamlessly connects to APIs – eliminating the grunt work usually involved. It then teams up with Alexa and presents the results back to the users with thoughtful analysis and commentary. Results which would help businesses make better decisions, more quickly and with less effort.

CISSP (ISC)² CBK Training Workshop 5th Aug 2019
The University of Waikato is the NZ official training provider for the prestigious (ISC)2 certification training program.

The next workshop training dates have been finalised for 5-9 August 2019 at the Novotel Lakeside Rotorua.

This training course is intended for professionals who have at least 5 years of recent full-time professional work experience in 2 or more of the 8 domains of the CISSP CBK and are pursuing CISSP training and certification to acquire the credibility and mobility to advance within their current information security careers. The training seminar is ideal for those working in positions such as, but not limited to:

• Security Consultant
• Security Manager
• IT Director/Manager
• Security Auditor
• Security Architect
• Security Analyst
• Security Systems Engineer
• Chief Information Security Officer
• Director of Security
• Network Architect

Industry Talk: Pivoting - or how we managed to make money with research outputs 26th Jul 2019
The ADAMS workflow system was never planned, but merely a by-product of a research project. Though the research project never saw the light of day, other projects built on top of it were commercially successful and turned ADAMS into a modular open-source Java framework. Being able to quickly create and deploy machine learning applications (no coding required!) was the key to success in bringing predictive models for spectral and image data to the agricultural sector.

Industry Talk: Autonomous Viticultural Tractor: R & D Opportunities in the Vineyard 31st May 2019
The Smart Machine Company is developing an autonomous vehicle for use in the vineyard environment for undertaking tasks such as mowing, spraying, and trimming, that would currently be performed by a tractor and driver. The intention is that a number of these vehicles can now be monitored by a single operator who is responsible for day to day duties such as refuelling, loading mission plans, responding to any mid mission warnings such as obstacle detections or mechanical faults. The vehicle also provides a unique opportunity to start collecting specific and potentially large amounts of information about the vineyard all year round and store this information with GPS tags. This talk will introduce the current projects as well as provide information on interesting challenges and opportunities for R & D in the ICT space (in the vineyard).

Industry Talk: Ethics and Professionalism in Information Technology 17th May 2019
Following the terrible events in Christchurch on March 15, when a terrorist attacked two mosques and shot dead 50 people, there has been a great deal of debate about the role of social media in fostering and encouraging extremism and terrorism. In particular, the discussion is around how social media providers can be forced to quickly detect, shut down and remove illegal content.

We also live in a time when the most technologically-advanced airplanes are literally falling out of the sky, apparently due to software issues, causing hundreds of deaths in different parts of the world.

The rise of Artificial Intelligence (AI) and machine learning also present a number of new issues around safety, control and how to prevent unintended consequences.

There are also dozens of examples, many local, where a lack of professionalism – or in some cases the refusal to respect what an IT professional has recommended – has caused millions of dollars in losses, disruption and inconvenience – Novopay, anyone?

This presentation will define what we mean by professionalism and why it has never been more important in IT. It will address some of the requirements for being a professional, including signing up to and adhering to a Code of Ethics, and how to achieve professional certification with IT Professional New Zealand. It will also cover the Seoul Accord and ITPNZ’s role in accrediting IT degrees.

Industry Talk: The Essence of Project Management 10th May 2019
An introduction to project management learning from real life project failures and understanding the why behind the need for project management to be successful in teams (even if you don’t want to be a project manager).

Industry Talk: Untethering the World 3rd May 2019
Wireless Technologies has completely revolutionised the way that we interact with the world in recent times. The most observable of these interactions is our use of smartphones, but there are many less obvious applications including everything from the use of sensors to monitor vital signs in real time through to delivering high-speed broadband to remote communities. This talk will cover my journey working with wireless technologies over the past twenty years to demonstrate some of the applications that wireless technologies are being used and to challenges that have presented themselves along the way.

Industry Talk: Best Practice - Moving from the Desktop to the Cloud 5th Apr 2019
Best Practice Software is an Australian company that started out producing practice management software for GPs. It added similar products aimed at allied health professionals and specialists through acquisition of two companies based here in Hamilton. In the position of now having a range of legacy products in diverse technologies, the company faces the challenge of consolidating its products into a single software-as-a-service offering called Titanium. Titanium has been designed to be a service-oriented architecture, deployed to Azure, with a progressive web application front end. This has been a significant departure from the architectural patterns used by the company in the past to develop locally hosted client/server applications. The presentation will walk through the architecture at a high level and discuss some of the changes and challenges faced in developing it. We will also discuss the continuous deployment pipeline and specific development disciplines needed to ensure quality is baked into the product as it is staged from development into production. Beyond technology, the shift from the relatively separate desktop applications developed initially by each of our constituent companies to a single large web application has introduced process and organisational challenges. One challenge has been adapting development methodologies to coordinate across multiple distributed Agile development teams. Another has been how to replace three well-established legacy products with a new “start-up” cloud product. This presentation will outline some of the unique challenges around merging and replacing existing products with an entirely new technology stack.

Seminar: Change Detection in High Dimensional Datastreams 2nd Apr 2019
Change detection problems are ubiquitous in science and engineering: promptly detecting changes is often key to understand the dynamics of a monitored process and for activating suitable countermeasures.

During this talk, I will address the problem of detecting distribution changes in high-dimensional datastreams and present QuantTree, a recursive binary splitting scheme that yields histograms for change-detection purposes. In fact, we theoretically prove that in QuantTree the bin probabilities do not depend on the distribution of stationary data, and the same holds for any test statistics based on bin counts, like the Pearson's one. Therefore, when using QuantTree it is possible to numerically compute the detection thresholds on univariate and synthetically generated data, yet guaranteeing a controlled false positive rate in any dimension and for any data distribution. Experiments show that QuantTree can effectively detect changes and control the false positive rate in high dimensional datastreams, even when the number of training samples is relatively small.

Our extensive experiments also indicate that all the considered techniques suffer of the detectability loss problem, namely that detecting a change of a fixed magnitude becomes increasingly more difficult when the data dimension scales. This is an intrinsic difficulty of change-detection methods, which we further investigate and analytically demonstrate to occur when monitoring the log-likelihood of a Gaussian datastream.

Industry Talk: Mind the Gap 29th Mar 2019
Technologists enjoy technology. We intuitively understand its benefits and how technology works. Business people don’t always see technology in the same way. This interactive talk addresses how executives, operational leaders, and other non-technical business professionals see technology and how those different perspectives can create barriers to effective communication with technologists. Examples of communication challenges will be provided. Attendees will learn about Simon Sinek’s golden circle and how to apply his concepts related to “how,” “what” and “why” to improve communication with nontechnical leaders. This includes getting the audience's attention, retaining their attention, and ensuring their understanding of the message. Attendees will learn how to read their audience and regularly assess the effectiveness of their communication. Approaches for how to adjust or correct the communication will be provided, with specific tactics and real-world examples from the presenter’s experiences in healthcare technology. Throughout the talk, concrete strategies will be presented for technical staff to overcome common barriers, connect their technology with business imperatives and overcome biases, so that they may improve their ability to talk and relate with non-technical business and operational leaders and staff.

Industry Talk: Tales from the Frontier of Genomics 22nd Mar 2019
In this talk, Sean will give some background on past projects he has been involved with, leading up to his current work in DNA sequence analysis at Real Time Genomics. The main part of the discussion will concern the challenges in processing human DNA sequence data. In particular, the talk will step through the computational steps involved in taking the output of a genomic sequencing machine through to generating outputs suitable for clinical interpretation. These steps include the mapping of sequence data to a reference human genome, the identification of germline and somatic variants in a sample, and clinical interpretation. Current problems in genomics including evaluation and bench-marking of results, dealing with pedigrees, and the detection of copy number and structural variants will be mentioned. Lessons learned and success stories from dealing with data from real customers will be presented. Time permitting, a brief introduction to metagenomic analysis and strain detection will be given. The talk will concentrate on computational aspects of dealing with large datasets and accurately determining variants in individual samples including the ranking of variants using a machine learning approach. There will be some discussion of the best practice software engineering techniques used by Real Time Genomics. In particular, the continuous integration environment, testing framework, and Jumble system for measuring code and test quality in Java will be shown. Brief mentions of projects involving entity extraction and text classification, leading to SureChEMBL system for identification of chemicals in patent documents. Only a minimal knowledge of genomics will be assumed.

Seminar: Advanced Infrastructure: Is Artificial Intelligence going to be a game changer? 19th Mar 2019
Computational methods have gained increasing attention in engineering and materials science applications, as they allow for the prediction of the failure of systems that would be prohibitive to test experimentally (either for their size or the external actions to simulate). In order to describe complex phenomena such as crack nucleation and propagation, computational tools must be able to accurately describe the material at the length scale at which they occur (the so-called ‘local’ scale). Meaningful predictions, however, rely on the description of the response of entire engineering systems (‘global’ scale). The exponential increase in computational power recorded in the last decade will inevitably lead to the possibility of simulating every atom or molecule in a bridge or a medical implant in a not-so-distant future. Until that day arrives, current practices to take into account physical phenomena that happen at different length scales revolve around the concept of multiscaling: we can solve the problem in exam at different observational levels (i.e. the local and global scales) and bridge the information between them by means of appropriate mathematical theories. In my recent works, I am investigating the possibility to embed classical multiscale theories in the framework of Artificial Intelligence, with the ultimate goal of increasing accuracy and efficiency of the model without compromising on the detailed physical description of the numerical models. We will discuss possible applications of Artificial Intelligence techniques to the regional-scale simulation of a flood protection infrastructure.

Industry Talk: Advances in Self-Managed Clouds and Language Runtimes 15th Mar 2019
Cloud Computing abstracts computing resources and offers them in a pay-as-you go manner to its tenant clients. This is a strong benefit for SMEs and non-technical companies who do not have to carry the burden of managing their own servers, with all the nuances that come with it (such as keeping them performant and secure). The current state-of-the-art cloud approach is that of Platform-as-a-Service (PaaS), which abstracts large parts of the software/hardware stack balancing development flexibility with reduced toil (i.e., repeating and tedious tasks). Crucially, cloud users rely on high-level languages—such as Node.js and Java—that are executed by Language Runtimes and provide a plethora of autonomic features, such as memory management and hardware-specific optimization. From a theoretical perspective, both Clouds and Language Runtimes can be explained through the theory of Self-Adaptive Systems, a cutting-edge field that aims to unify AI and Systems in its quest of explaining everything from cruise control in cars to autoscaling on the Cloud to autonomous exploration rovers. In this talk, applied research on the field of Self-Adaptive Clouds and Language Runtimes that was conducted in collaboration with IBM will be presented. The technical focus was on improving the satisfaction of nonfunctional requirements of the IBM Bluemix PaaS features and the J9 JVM (IBM’s Java Language Runtime). More specifically, the overarching theme was the overcoming of performance interference in clouds that co-locate multiple runtimes. Besides numerous academic publications, this project resulted in 2 patents and 4 anti-patents with IBM as well as was awarded the international “Project of the Year” distinction by IBM CAS in 2016. In Waikato, this work has culminated with the new Oceania Researchers in Cloud and Adaptive-systems (ORCA) lab, which currently hosts 9 research students, supervised by 7 faculty members.

Seminar: Towards Deep Continuous-Discrete Machine Learning 12th Mar 2019
Since the beginnings of machine learning - and indeed already mentioned in Alan Turing's groundbreaking 1950 paper "Computing machinery and intelligence" - two opposing approaches have been pursued: On the one hand, approaches that relate learning to knowledge and mostly use "discrete" formalisms of formal logic. On the other hand, approaches which, mostly motivated by biological models, investigate learning in artificial neural networks and predominantly use "continuous" methods from numerical optimization and statistics. The recent successes of deep learning can be attributed to the latter, the "continuous" approach, and are currently opening up new opportunities for computers to "perceive" the world and to act, with far-reaching consequences for industry, science and society. The massive success in recognizing "continuous" patterns is the catalyst for a new enthusiasm for artificial intelligence methods. However, today's artificial neural networks are hardly suitable for learning and understanding "discrete" logical structures, and this is one of the major hurdles to further progress.

Accordingly, one of the biggest open problems is to clarify the connection between these two learning approaches (logical-discrete, neural-continuous). In particular, the role and benefits of prior knowledge need to be reassessed and clarified. The role of formal logic in ensuring sound reasoning must be related to perception through deep networks. Further, the question of how to use prior knowledge to make the results of deep learning more stable, and to explain and justify them, is to be discussed. The extraction of symbolic knowledge from networks is a topic that needs to be re-examined against the background of the successes of deep learning. Finally, it is an open question if and how the principles responsible for the success of deep learning methods can be transferred to symbolic learning. In talk, I will discuss these topics and give examples of various recent approaches.

Seminar: Facets of Fairness and Transparency in Algorithmic Decision Making 5th Mar 2019
Modern predictive analytics and machine learning techniques contribute to the massive automation of the data-driven decision making and decision support. It becomes better understood and accepted, in particular due to the new General Data Protection Regulation (GDPR), that employed predictive models may need to be audited. Disregarding whether we deal with so-called black-box models (e.g. deep learning) or more interpretable models (e.g. decision trees), answering even basic questions like “why is this model giving these answer?” and “how do particular features affect the model output?” is nontrivial. In reality, auditors need tools not just to explain the decision logic of an algorithm, but also to uncover and characterize undesired or unlawful biases in predictive model performance, e.g. by law hiring decisions cannot be influenced by race or gender. In this talk I will give a brief overview of the different facets of comprehensibility of predictive analytics and reflect on the current state-of-the-art and further research needed for gaining a deeper understanding of what it means for predictive analytics to be truly transparent and accountable.

Seminar: Self-adaptive Cloud - Predictive Autoscaling 26th Feb 2019
Self-adaptive systems are all around us. Moreover, we, as human beings, are the most advanced self-adaptive systems ever. What does it mean to be self-adaptive? Essentially, self-adaptive systems have notion of environment that they constantly observe as well as raison d'etre, i.e. why do they exist, what are their goals. Self-adaptive system utilizes observations to derive intrinsic “adaptation algorithms” that compute system’s “settings” that allow it to meet its goal (to fulfill its raison d'etre) within the corresponding environment conditions. This general biological concept was borrowed by computer science to enable automation in complex multi-parameter systems that are highly stochastic like cloud.

An example of a self-adaptivity in the cloud is the predictive autoscaling technology. This technology aims to provide enough cloud resources and container replicas to serve the anticipated user demand according to some service-level objectives. Such technology uses advanced AI techniques and time series-related models to derive the capacity/performance model of application and to forecast the demand on this application. This AI-powered technology was recently introduced by AWS to EC2 service that offers various types of virtual machines.

In his talk, Vladimir Podolskiy, a Ph.D. student from the Technical University of Munich (Germany) will introduce an ongoing project in developing the predictive autoscaling engine SCALENDAR at TUM. Vladimir will give an overview of the SCALENDAR’s architecture, and will dive deeper into its components implementing various functions such as: workload forecasting, microservice capacity/performance modeling, application structural modeling, scaling actions scheduling, and scaling actions timely execution. The aim of the talk is to introduce the research project and to invite the colleagues from the University of Waikato to collaborate on corresponding research topics.

Seminar: Cloud Computing and Self-adaptive Management of Cloud 19th Feb 2019
With the introduction of cloud computing paradigm, it became possible to rent the compute and memory resources as well as storage capacity instead of owning them. The usage of these remote resources via Internet is billed on pay-as-you-go basis, i.e. you pay only for what you really use. Both cloud services providers and the users of such services came to understanding that they need to optimize the cloud. Cloud services providers are interested in minimizing the free hardware capacity and in increasing the reliability of the cloud. Cloud users, on the other hand, are interested in getting high quality of service and in minimizing the cost of the cloud services. Difficulties of cloud and cloud applications management require a high degree of automation, which, in turn, relies on how accurate models of cloud and cloud applications are. By utilizing the vast tracing, monitoring, logs and load tests/performance data, one can approach the task of automated cloud management with the concept of self-adaptive systems. The methods and models utilized by self-adaptive systems vary a lot, they span from formal and statistical models to cutting-edge AI techniques such as long short-term memory artificial neural networks and deep learning.

In his talk, Vladimir Podolskiy, a Ph.D. student from the Technical University of Munich (Germany) will give a background about cloud computing, cloud applications, and the concepts of self-adaptive systems as they are used for managing the cloud. The aim of the talk is to highlight the opportunities that cloud provides to its users and to point out how the modeling and methods from AI can be employed to automate the management of cloud.

Industry Talk: Commercial Applications of Machine Learning in Agriculture 8th Feb 2019
Using Machine Learning in an agricultural context for building predictive models on spectral, thermal and image data. Rapid application development is achieved with the help of workflows, by writing Java components only once and applying them without any additional code.

Industry Talk: Starting and growing a subscription software business from the Waikato — learnings from 10 years in business 1st Feb 2019
When Rocketspark was started 10 years ago in a Hamilton flat, they knew that the build costs of traditional web development were unrealistic for most small businesses, putting a beautiful and effective website out of reach. They could also see that traditional content management systems were too complicated for business owners and they couldn't afford to pay someone to manage the site themselves. Rocketspark is a website builder and ecommerce software platform where clients can easily build their own beautiful website without needing to know code—or if they’d rather pay someone else to build it for them, they can connect with a Rocketspark design partner. Over the last ten years Rocketspark’s founders have learned a lot about growing a subscription software business, finding their first customers, collecting customer feedback and interviewing customers, learning how to delight customers with hands-on customer support, releasing new features to customers, managing infrastructure and growing a team from just the founders to a team of 14 — almost all of them graduating from Waikato. Rocketspark recently secured a Callaghan Innovation Growth Grant and has employed 4 summer interns from the University of Waikato.

Seminar: Machine-Learning For Security Analysis: Opportunities and Challenges 8th Jan 2019
Machine learning (ML) is widely being used worldwide to solve problems in many areas including image recognition, natural language processing, anomaly detection, and more. Its success has also resulted in a lot of hype. While there is no doubt that many of ML techniques are very sophisticated, there are a number of challenges remaining.

In this talk I will summarise our work on using ML in the area of vulnerability detection using off-the-shelf ML techniques. I will compare results from the ML approach to those from static program analysis techniques. Labelled data used in this work comes from our earlier static analysis work. I will point to challenges and open questions that remain open in order for ML techniques to be useful for security analysis purposes, in a reliable way.

Seminar: Cyber Security Modeling and Analysis of Internet of Things 19th Dec 2018
A lot of Internet of Things (IoT) devices are vulnerable to cyber attacks. It is important to assess and enhance the security of IoT devices and IoT networks and service. The Cybersecurity Lab at the University of Canterbury (UC) along with other universities have received an MBIE grant on “advanced security technologies for the Internet of Things” (http://iotresearch.org/) as international collaborations between NZ universities (UC, Auckland University of Technology, Massey University, and Victoria University of Wellington) and Universities in Korea (Korea University and Sungkyunkwan University). In this talk, I will introduce on-going research topics under the MBIE grant and (1) a cybersecurity modeling and evaluation framework will be presented, (2) security assessment of IoT networks with non-patchable IoT nodes and mobile IoT nodes and (3) network level security defense techniques including moving target defenses and deception techniques. Finally, future work research revenues will be discussed.

Industry Talk: Real-world agile software development 14th Dec 2018
Agile is a work flow system that is rapidly gaining traction as a way organize and run projects within companies and organizations. Through the use of tribes, scrums, chapters, squads, sprints and missions small groups of people from differing backgrounds and skill sets collaborate to solve problems by sharing knowledge, ideas and effort. Agile breaks down barriers in traditional management structures by sharing knowledge and decision making among team members, giving the potential to greatly speed up workflows. There is no single best way to implement the Agile process as its performance and success varies based on the organization, team makeup, skill sets, project requirements and customer expectations. From my experiences as an employee in small to medium sized businesses working on various software projects I will give my thoughts and views on the challenges of real world software development, how we have applied components of the Agile Software Development process. I will describe some of the tools we use and others we have built to aid in our software development processes.

Australasian Conference on Combinatorial Mathematics and Combinatorial Computing 2018 10th Dec 2018
41ACCMCC will follow a similar format to previous conferences in the ACCMCC annual series, which is overseen by the Combinatorial Mathematics Society of Australasia (CMSA) and began in 1972. The conference programme will include keynote speakers, contributed talks in parallel sessions, a conference dinner and presentation of the CMSA student prize, and the CMSA Annual General Meeting.

Cyber Security Seminar: Australian Cybersecurity Policy at Federal Level (and its Relationship to both Cybercrime and Cyberconflict) 23rd Nov 2018

David Sanger, the national security correspondent for the New York Times and author of the recent book “The Perfect Weapon: War, Sabotage and Fear in the Cyber Age” stated in a recent interview that “..in cyber conflict, the advantage goes to the least-wired society attacking the most-wired society.”  In this regard governments worldwide, and at various levels, have rapidly published statements and policies related to appropriate responses to cyber-security and cyber-defence, now often referred to simply as “cyber”. Australia’s Federal Government released “Australia’s Cyber Security Strategy: Enabling Innovation, Growth & Prosperity” in 2016 with a “First Annual Update” in 2017. The strategy identified five major “themes” as requiring action over the years to 2020. These were based upon:

1. A national cyber partnership;
2. Strong cyber defences;
3. Global responsibility and influence;
4. Growth and innovation; and
5. A cyber smart nation.

These need to be assessed against national military and defence policy overall, particularly in regard to funding and development as well as against similar policies and programs developing internationally, particularly in the Asia/Pacific (APAC) region. Emphasis needs to be given to necessary support for training, education and research in the area, now deemed to be of critical national security importance.

International Conference On Asia-Pacific Digital Libraries 2018 19th Nov 2018
Since its beginnings in Hong Kong in 1998, ICADL has become one of the premiere international conferences for digital library research. ICADL 2018 at the University of Waikato in New Zealand offers a valuable opportunity for researchers, educators, and practitioners to share their experiences and innovative developments. The main theme of ICADL 2018 is “Maturity and Innovation in Digital Libraries”.

Industry Talk: Statistics & Alchemy 16th Nov 2018
Imagimation is a technology start-up incubator based in Frankton, which currently houses several electricity retailers, a game development studio, and a technology solutions provider doing “affordable intelligence”. Within the umbrella of Imagimation, we work with full-stack web development, native ML/AI applications, and video games, alongside many smaller areas of interest. Being a group of technology companies, almost everything we do is tech-related at some level. Our primary technology stack uses Microsoft products at every level: VS IDE; Azure storage, hosting and compute; Razor Pages and ASP .NET Core frameworks. We also use Python, C++, Java, Julia, and Node/React/Redux in varying capacities.

I started at Imagimation as purely a machine learning developer, but quickly found myself involved in many other projects and areas. I’ve worked in almost every role I had skills in, and many I hadn’t, including project management, web development, and business strategy. Currently, I am a technology lead for one start-up and the owner/CEO of another.

The key skills needed to establish yourself in a technology start-up are adaptability and the willingness to learn, comfort with ambiguity, communication, and a positive attitude. Everything else can be learned on the job.

Cyber Security Seminar: SafeToOpen: Fighting newly made phishing website 9th Nov 2018
SafeToOpen has emerged out of a huge need to protect against phishing attacks which require its own defence strategy. Phishing attacks are not easy to detect as they grow smarter. In combination with social engineering techniques, the attack vector is very difficult to detect using conventional security controls. SafeToOpen has been designed using advanced strategies uniquely for detecting and eliminating phishing attacks. It uses security intelligence feeds as well as specially crafted algorithms to detect and respond to phishin.