Melbourne Program – 23 June 2021

David Rawlinson, Lead data scientist, WSP
From POC to Production: Successful delivery of AI/ML projects
AI/ML projects have a high failure rate, which can be particularly damaging to SMEs. Companies often face significant challenges operationalising new technology, especially when working to bring change to large client organisations and government. Successful adoption of AI/ML technology almost always involves changing business processes, integration with a variety of existing systems, management of competing stakeholders, and often depends on a solid foundation of digital maturity in client organizations. This talk will describe some of our experiences integrating AI/ML into existing businesses, and some of the strategies that we have found successful.

Dr Maggie Zhang, Deep Learning Engineer, NVIDIA
Accelerating AI applications using NVIDIA GPUs and AI software stack
Accelerating AI applications demands great computational resources when datasets increase in size and models in complexity. This talk will introduce NVIDIA’s latest Ampere architecture and AI software stack to help you maximize the performance of your AI workloads on GPUs. NVIDIA AI software stack includes libraries, deep learning frameworks optimized for GPUs, and multiple tools for deep learning training and inference, to help you get the most out of GPUs. We will also share some customer stories to show you how they successfully accelerated their AI workloads in different domains using GPUs.

Prof John Grundy, Department of Software Systems & Cybersecurity, Monash Data Futures Institute
Impact of developer and end user human issues on AI and cybersecurity
Humans are a key part of software development, including customers, designers, coders, testers and end users. This talk will discuss several examples from recent work on handling human-centric issues when engineering software systems. This includes personality impact on aspects of software development; understanding interpersonal issues in agile practices; incorporating end user emotions into software requirements engineering; providing proactive design critics in software tools to augment human decision making; modelling diverse human users of software systems; human-centric defect reporting; and the use of human-centric, domain-specific visual models for non-technical experts to specify and generate systems, without the need for software engineers at all.

Prof Svetha Venkatesh, Alfred Deakin Professor and co-Director of Applied Artificial Intelligence Institute (A²I²) at Deakin University
SMEs: what are the AI opportunities?
This talk will outline promising directions for the use of AI in SMEs. It will introduce basic technologies that underpin areas of prediction, recommendation, anomaly detection and optimization.

Dr Paul Beinat, Principal at Finity Consulting, Adjunct professor at UTS
Using an artificial immune system to detect malicious intrusions
Can people who know almost nothing about cyber attacks and malicious activity make any headway in this field? Using a novel artificial neural network to model the joint probability distribution of data, as the self in an artificial immune system, has enabled a robust type of anomaly detection. Anomalies can be detected individually or by comparing the joint distribution of live data to the reference distribution, leading to collective anomalies. We will examine its performance on a real web traffic data set.

AI and Cyber for Culture and Society

Dr Shiri Krebs, Associate Professor of Law, Deakin Law School
Co-Lead, Law and Policy Theme, Cyber Security Cooperative Research Centre
Predictive Technologies: New Challenges in a Future-Focused World
Businesses have been increasingly relying on AI-based applications to assist decision-making processes and risk assessments in various organisational contexts. AI-based technologies provide sensitive, future-focused, data to assist decision-makers in evaluating risk. While adding a large amount of time-sensitive and relevant data that informs decision-makers in real time, AI-generated data also place additional burdens on decision-makers, including by triggering cognitive biases, including invisible blind spots, and creating a persuasive virtual representation of reality that is difficult to refute. In this talk, A/Prof Krebs will discuss the challenges stemming from human-machine interaction in the context of technology-assisted decision-making processes, and suggest ways for improving technology-assisted decision-making processes.

Dr Catherine Lopes, AI, Data & Analytics + Intelligent Automation Networks
AI and people analytics: Risks and rewards

Dr Omaru Maruatona, CEO at Aiculus
Innovating with Cyber-AI: Challenges and opportunities
Aiculus was founded on the grand ambition and opportunity of Cyber-AI in 2017. In 2020, Aiculus secured significant seed funding from a venture capital firm and was also conferred the biggest innovation award in Singapore.  Dr Omaru Maruatona shares his experiences and thoughts on innovating in the Cyber-AI space, including the challenges and opportunities that Aiculus has faced along the journey.

Dr Sue Keay, Chair: Robotics Roadmap for Australia, Chair: Robotics Australia Group
Transforming business operations with AI and Digital Twins
Improved computer processing power, access to the internet and the proliferation of robust, low cost sensors is paving the way for more and more businesses to take advantage of AI and create digital twins. The technology is now becoming accessible to SMEs across a range of new sectors. Examples of such applications will be discussed.