How Do Large Language Models Capture the Ever-changing World Knowledge? A Review of Recent Advances

Speaker: Ling Chen, University of Technology Sydney, Australia
Abstract: Large language models (LLMs) have shown impressive performance in solving various tasks. However, as the world is constantly changing and new information is being generated every day, the trained LLM can be quickly outdated without re-training, which can be a time-consuming and resource-intensive process. Consequently, maintaining LLMs up-to-date status is a pressing concern in the current era. In this talk, I will provide a review of recent advances in aligning LLMs with the ever-changing world knowledge without re-training from scratch, by categorizing research works and providing in-depth comparisons and discussion. I will also discuss existing challenges and highlight future directions to facilitate research in this field.

Bio: Ling Chen is a Professor in the School of Computer Science at the University of Technology Sydney. She leads the Data Science and Knowledge Discovery Laboratory (The DSKD Lab) within the Australian Artificial Intelligence Institute (AAII) at UTS. Ling has been persistently working in the area of machine learning and data mining for 20 years, dedicated to undertaking innovative research to produce high quality results and attracting and leading research and industry projects to initiate and investigate new research areas and create real impacts.

Ling’s recent research interests include anomaly detection, data representation learning, and dialogues and interactive systems. Her research has gained recognition from both government agencies, receiving competitive grants such as ARC DP/LP/LIEF, and industry partners, with contracted research support from entities like Facebook Research and TPG Telecom. Ling serves as an Editorial Board member for journals including the IEEE Journal of Social Computing, the Elsevier Journal of Data and Knowledge Engineering and the Computer Standards and Interfaces.

On Extreme Classification & Large Language Models for Search & Recommendation

Speaker: Manik Varma, Microsoft Research India, India
Abstract: Large language models have generated a lot of excitement in the community and have delivered many successes. They have also raised many research challenges in terms of their accuracy, efficiency, reliability, safety and applicability. In this talk, I will address some of these challenges for closed-domain search and recommendation applications. I will discuss how such applications can be reformulated as classification tasks by treating each item to be ranked or recommended as a separate label in a multi-label classification framework. This reformulation allows us to leverage solutions based on extreme classification where the objective is to tackle classification problems with millions to billions of categories/labels. I will demonstrate how to combine extreme classifiers with language models to develop accurate search and recommendation systems with low training and inference costs that can benefit many people worldwide.

Bio: Manik Varma is a Distinguished Scientist and Vice President at Microsoft Research India. He is best known for having started the research area of extreme classification in large-scale machine learning which has opened a new paradigm for search and recommendation applications. As a result, his algorithms have been deployed extensively on the Microsoft Search and Advertising platform including in the new Bing Chat experience. They have made billions of predictions a day, have improved the experience and productivity of hundreds of millions of people and increased the reach and revenue of many businesses worldwide. Manik is also known for having developed tiny classifiers which can fit in 2-16 KB of RAM and which can be deployed on microcontrollers smaller than a grain of rice. These classifiers have protected hundreds of millions of devices from unknown viruses and malware. Manik has served as an Associate Editor-in-Chief of the IEEE TPAMI journal and as a Senior Area Chair (or equivalent) at most of the premiere conferences on machine learning, artificial intelligence and computer vision. Manik is a recipient of the Government of India’s Bhatnagar Prize, is a Fellow of the Indian Academies of Science and Engineering, has been a Visiting Miller Professor at UC Berkeley and a Rhodes Scholar at Oxford.

The Dermatology Domain

Speaker: Peter Soyer, University of Queensland, Australia
Abstract: Artificial Intelligence (AI) has made significant advancements in the field of dermatology, offering a wide range of applications and benefits. Here are some key areas in which AI is being used in the dermatology domain and on which my presentation will focus:

1. Skin Cancer and Skin Disease Diagnosis:
   AI-powered systems, particularly deep learning algorithms, can analyze images of skin conditions such as moles & melanoma and rashes. These systems can often provide accurate and quick preliminary diagnoses, helping dermatologists in their decision-making process.
2. Telemedicine and Teledermatology:
   AI-powered apps and platforms enable patients to take images of their skin conditions and share them with dermatologists remotely. This is particularly useful for individuals in rural or underserved areas and has become more important during the COVID-19 pandemic.
3. Clinical Desicion Support:
   AI systems can provide dermatologists with additional information and insights when making complex diagnostic and treatment decisions, improving overall clinical decision-making.
4. Research and Drug Development:
   AI can analyze large datasets to identify patterns, potential drug candidates, and research areas for various skin conditions. This can accelerate drug discovery and research in dermatology.
5. Wearable Devices and Monitoring:
   AI-powered wearable devices can continuously monitor skin conditions, track changes over time, and alert patients and healthcare providers to any significant developments.

Bio: Inaugural Chair in Dermatology at UQ and Director of the Princess Alexandra Hospital Dermatology Department (2008 – 2022). More than 200 publications in the last five years. He initiated and was co-leader of the Australian Skin and Skin Cancer Research Centre, a collaboration between UQ and QIMR Berghofer Medical Research Institute. Co-president of the 9th World Congress of Melanoma in Brisbane 2017. Leading the Australian Centre of Excellence in Melanoma Imaging and Diagnosis (ACEMID), a consortium of The Universities of Queensland, The University of Sydney and Monash University. Fellow of the Australian Academy of Health and Medial Sciences.

Beyond optimal solutions for real-world decision-support problems

Speaker: Maria Garcia De La Banda, Monash University, Australia
Abstract: Combinatorial optimisation technology has come a long way. We now have mature high-level modelling languages in which to specify a model of the particular problem of interest; robust complete solvers in each major constraint paradigm including Constraint Programming (CP), MaxSAT, and Mixed Integer Programming (MIP); effective incomplete search techniques that can easily be combined with complete solvers to speed up the search such as Large Neighbourhood Search; and enough general knowledge about modelling techniques to understand the need for our models to incorporate components such as global constraints, symmetry constraints, and more. All this has significantly reduced the amount of knowledge required to apply this technology successfully to the many different combinatorial optimisation problems that permeate our society.

And yet, not many organisations use such advanced optimisation technology; instead, they often rely on the solutions provided by problem-specific algorithms that are implemented in traditional imperative languages and lack any of the above advances. Further, while advanced combinatorial optimisation technology is particularly suitable for the kind of complex human-in-the-loop decision-making problems that occur in critical sectors of our society, including health, transport, energy, disaster management, environment and finance, these decisions are often still made by people with little or no technological support. In this talk I argue that to change this state of affairs, our research focus needs to change from improving the technology on its own, to improving it so that users can better trust, use, and maintain the systems we develop with it.

Bio: Maria is a Professor at the Faculty of Information Technology at Monash University with more than 25 years of experience as an academic. She is currently a member of the ARC College of Experts, the Co-Chair of the Monash-Woodside FutureLab and until July 2022 the Deputy Dean Research of the Faculty. Prior to that she was overall Deputy Dean of the Faculty (2013-2016) and the Head of the Caulfield School of Information Technology (2009-2011). Her research interests include Combinatorial Optimisation, Program Analysis and Transformation, Programming Languages and Bioinformatics. Since 2010 she has been Area Editor of the Journal of Theory and Practice of Logic Programming and, since 2019, member of the Editorial Board of the Constraints journal. She has been Chief Investigator in 11 ARC grants (2 cross-Faculty), and Principal Investigator in an NHMRC program.

Maria’s PhD won the Universidad Politecnica de Madrid’s Best PhD Award. In 1997 she was awarded the first and only prestigious Logan Fellowship in the Faculty of Information Technology, which she held for six years. In 2005 she won, with Peter Stuckey, the International Constraint Modelling Challenge. She was an elected member of the Executive Committee of the Association of Logic Programming (2005-08) and of the Executive Committee of the Association of Constraint Programming (2017-20), of which she was also elected as President (2019-20). In 2021 she was inducted into the Monash Honour Roll.

Evolutionary Machine Learning: Research, Applications and Challenges

Speaker: Mengjie Zhang, Victoria University of Wellington, New Zealand
Abstract: Since the 1990s, evolutionary computation techniques have been widely used to solve machine learning tasks. In this talk, I will firstly provide a brief overview of machine learning and evolutionary computation, then provide a narrow view and a broad view of evolutionary machine learning. After discussing the state-of-the-art research and applications of the main paradigms of evolutionary machine learning and their success in classification, feature selection, regression, clustering, computer vision and image analysis, scheduling and combinatorial optimisation, and evolutionary deep learning, the main challenges and lessons will be discussed. If time allows, I will provide an overview of our recent developments and discuss potential opportunities.

Bio: Professor Mengjie Zhang is a Fellow of Royal Society of New Zealand, a Fellow of Engineering New Zealand, a Fellow of IEEE, an IEEE Distinguished Lecturer, currently Professor of Computer Science at Victoria University of Wellington, where he heads the interdisciplinary Evolutionary Computation and Machine Learning Research Group. He is also the founding Director of the Centre for Data Science and Artificial Intelligence at the University.

His research is mainly focused on AI, machine learning and big data, particularly in evolutionary learning and optimisation, feature selection/construction and big dimensionality reduction, computer vision and image analysis, scheduling and combinatorial optimisation, classification with unbalanced data and missing data, and evolutionary deep learning and transfer learning. Prof Zhang has published over 800 research papers in refereed international journals and conferences. He has been serving as an associated editor for over ten international journals including IEEE Transactions on Evolutionary Computation, IEEE Transactions on Cybernetics, the Evolutionary Computation Journal (MIT Press), and involving major AI and EC conferences as a chair. He received the “EvoStar/SPECIES Award for Outstanding Contribution to Evolutionary Computation in Europe” in 2023. Since 2007, he has been listed as a top five (currently No. 4) world genetic programming researchers by the GP bibliography (http://www.cs.bham.ac.uk/~wbl/biblio/gp-html/index.html).

Prof Zhang is a past Chair of the IEEE CIS Intelligent Systems Applications Technical Committee, the IEEE CIS Emergent Technologies Technical Committee and the IEEE CIS Evolutionary Computation Technical Committee, a vice-chair of the IEEE CIS Task Force on Evolutionary Feature Selection and Construction, a vice-chair of the IEEE CIS Task Force on Evolutionary Computer Vision and Image Processing, and the founding chair of the IEEE Computational Intelligence Chapter in New Zealand.

Unleashing the Power of AI: Revolutionising Industries and Driving Innovation

Speaker: Dadong Wang, Data61, Australia
Abstract: Artificial Intelligence (AI) has emerged as a transformative technology, revolutionizing industries across the world. With its ability to process vast amounts of data, learn from patterns, and make informed decisions, AI is enabling unprecedented levels of automation, efficiency, and innovation. As we stand at the precipice of a technological revolution, it is paramount to explore the immense potential and industry applications of AI, ensuring a balanced perspective that both inspires and informs.

This presentation will explore the diverse applications of AI across industries such as healthcare, agriculture, environment, and more. We will delve into specific use cases, showcasing how AI is enhancing decision-making processes, augmenting human capabilities, and optimizing operational efficiency across diverse sectors. Additionally, we will discuss the ethical considerations surrounding AI, ensuring that this powerful technology is developed and deployed responsibly.

This keynote speech aims to ignite a forward-thinking conversation wherein attendees will gain a comprehensive understanding of AI's industry applications, exploring ways to harness its potential while addressing the challenges lying ahead. Join us as we uncover the potential of AI and explore how AI can shape the future of industries and drive sustainable growth in the ever-evolving landscape of technological innovation.

Bio: Dr Dadong Wang is a Principal Research Scientist & the leader of the CSIRO Quantitative Imaging Research Team, part of the CSIRO Data61. He is recognised as a leader in developing revenue generating IP which is exemplified by the team’s license deal with a world leading imaging company, ranked Data61's No. 1 license in the top 10 licenses of CSIRO Data16 since 2017. He has been actively engaged with Australia's industry and is recognised as a CSIRO SME (Small and Medium Enterprises) champion. He is also recognised externally for his contributions to the industry. The joint team of Data61 (led by him) and an industry partner is the winner of NSW iAwards 2023 in Business and Industry Solution category, and the merit recipient of Technology Platform Solution category. The team is also the merit recipients of National iAwards 2023 in both the Business and Industry Solution and Technology Platform Solution categories.

He is also a Conjoint Professor at the University of New South Wales (UNSW) and an Adjunct Professor at the University of Technology, Sydney (UTS). Through these universities and the University of Sydney, he is a primary/joint PhD supervisor. Prior to joining the Commonwealth Scientific and Industrial Research Organisation (CSIRO) in 2005, he had worked for two multinational companies for six years, developing large intelligent systems for infrastructure monitoring and control.