Shane Feng (UQ), Zhongkui Ma (UQ), Zihan Wang (UQ), Eu Joe Chegne (UQ), Mengyao Ma (UQ), Sharif Abuadbba (Data61), Guangdong Bai (UQ), “Uncovering Gradient Inversion Risks in Practical Language Model Training” accepted at ACM CCS’24 (Data61 target venue). The significance of this paper lies in its identification of a potential vulnerability in language models, specifically their susceptibility to inversion attacks, which can lead to the exposure of private data used in their training.
Akib Karim (Data61), Shaobo Zhang (UniMelb), Muhammad Usman (Data61), “Low depth virtual distillation of quantum circuits by deterministic circuit decomposition”, was published in Physical Review Research on August 27, 2024. The paper introduces a novel method of performing duplicate circuit virtual distillation at a significantly lower depth than previously possible and for arbitrary observables, and which was demonstrated on real quantum devices to mitigate real experimental noise. The paper is accessible here: https://journals.aps.org/prresearch/abstract/10.1103/PhysRevResearch.6.033223.
Fatemeh Jalalvand, Mohan Baruwal Chhetri, Surya Nepal, Cécile Paris, “Alert Prioritisation in Security Operations Centres: A Systematic Survey on Criteria and Methods” accepted at ACM Computing Surveys journal (Impact Factor: 23.8, ranked 1/143 in Computer Science Theory & Methods). Effective alert prioritisation (AP) is crucial for Security Operations Centres to address the problem of the increasing volume of security alerts, causing security analysts fatigue, and increasing the risk of security incidents. This paper provides a comprehensive review of the criteria and methods for AP in Security Operations Centres. It analyses the advantages and disadvantages of the different categories of AP criteria and methods based on human-AI teaming, specifically considering automation, augmentation, and collaboration.
Binh M. Le, Shahroz Tariq, Simon S. Woo, “Bridging Optimal Transport and Jacobian Regularization by Optimal Trajectory for Enhanced Adversarial Defense”, accepted at the 17th Asian Conference on Computer Vision (ACCV 2024) for Oral presentation (5.6%). We explore adversarial training and Jacobian regularization, two key defenses in robust classifier design. Our work is the first to thoroughly analyze and compare these methods, both theoretically and empirically, to assess their impact on robust learning. We then introduce OTJR, a novel method that combines input Jacobian regularization with output representation alignment using optimal transport theory. [ePublish: EP2024-2986]
Yansong Gao, Seyit A Camtepe, Nazatul Haque Sultan, Hang Thanh Bui, Arash Mahboubi, Hamed Aboutorab, Michael Bewong, Rafiqul Islam, Md Zahidul Islam, Aufeef Chauhan, Praveen Gauravaram, Dineshkumar Singh, “Security Threats to Agricultural Artificial Intelligence: Position and Perspective”, Computers and Electronics in Agriculture, EP2024-1719, This position paper endeavors to amplify the security consciousness among stakeholders (e.g., end-users such as farmers and governmental bodies) engaged in the implementation of AI systems within the agricultural sector.
Amena Khatun & Muhammad Usman, “Quantum Transfer Learning with Adversarial Robustness for Classification of High-Resolution Image Datasets”, has been published in Advanced Quantum Technologies. In this work, we propose a quantum transfer learning (QTL) architecture that integrates quantum variational circuits with a classical machine learning network pre-trained on ImageNet dataset. Through a systematic set of simulations over a variety of image datasets, we demonstrate the superior performance of our QTL approach over classical transfer learning. Furthermore, we evaluate the adversarial robustness of QTL architecture with and without adversarial training, confirming that our QTL method is adversarially robust against data manipulation attacks and outperforms classical methods. [arXiv: https://arxiv.org/abs/2401.17009]
Tariq, Razaib, Tariq, Shahroz and Woo, Simon S., “Exploring the Impact of Moire Pattern on Deepfake Detectors”, IEEE International Conference on Image Processing (ICIP 2024), This paper shows how Moire Pattern in recaptured videos can impact the performance of state-of-the-art deepfake detectors. [ePublish ID: EP2024-0578]
Diksha Goel, Kristen Moore, Mingyu Guo, Derek Wang, Minjune Kim, Seyit Camtepe, “Optimizing Cyber Defense in Dynamic Active Directories through Reinforcement Learning,” 29th European Symposium on Research in Computer Security (ESORICS 2024). This paper proposes dual reinforcement learning-based policies for advancing cyber defense in dynamic and large-scale environments. Our results demonstrate that the proposed approach generates highly effective defenses and accurately models attacker behavior in large-scale dynamic environments. [ePublish ID: EP2024-3916]
Chen C-J, Xu R, Wei Shao, Zhang J, Tu Z. OpenCDA-Loop: A Closed-loop Benchmarking Platform for End-to-end Evaluation of Cooperative Perception. Advances in Neural Information Processing Systems.(CORE A*) 2024. This paper provides a pioneering closed-loop benchmarking platform that bridges perception and behavioral planning, offering new insights into system-level evaluations for enhancing safety and efficiency in V2V cooperative perception systems.
Nam Tran, The Anh Ta, Debmalya Mandal, Long Tran-Thanh, “Symmetric Linear Bandits with Hidden Symmetry”, The Thirty-eighth Annual Conference on Neural Information Processing Systems (NeurIPS 2024) CORE Rank A*. We develop an efficient algorithm for scenarios, like fast adaptation user’s preference in online commercial platforms, where there are many similar users.
Nam Tran, The Anh Ta, Debmalya Mandal, Shuqing Shi, Yali Du, Long Tran-Thanh, “Learning the Expected Core of Strictly Convex Stochastic Cooperative Games”, The Thirty-eighth Annual Conference on Neural Information Processing Systems (NeurIPS 2024) CORE Rank A*. We develop an efficient algorithm for sustainably grouping participants in a cooperative task, like planning to set up and maintain a shared computing lab, in an optimal manner so that no participant will benefit by leaving his group.
Patents
Hevish Cowlessur, Chandra Thapa, Tansu Alpcan, Seyit Camtepe filed a patent under Ref No. P10334/AU/PRV, titled “A Hybrid Quantum Neural Network for Split Learning,”
Nazatul Sultan, Josef Pieprzyk, Wei Ni, Sharif Abuadbba, Hajime Suzuki filed a provisional patent under Ref. No. 801105PRV, titled “Attack-Resistant Authentication.” This patent is an outcome of our 6G security and privacy program funded by the Department of Home Affairs. It introduces an active attack-resistant AKA protocol, solving. We plan to submit input to the worldwide standardisation body named 3GPP.
Ejaz Ahmed filed a provisional patent under the title “Function Identity Prediction”, CSIRO Ref. P10348/AU/PRV. It introduces an AI-powered method to predict function names in machine code, revealing the intent of the software’s behaviour when attackers have obfuscated the original function names.
Amena Khatun won the 2024 Women in Technology Awards in the Emerging Tech Star category for her contributions to Quantum Machine Learning. The winners will be announced on September 27th at the WiT Awards night.
(Distinguished Paper Award) Shaofeng Li, Xinyu Wang, Jason Xue, Haojin Zhu, Zhi Zhang, Yansong Gao, Wen Wu, Xuemin (Sherman) Shen, Yes, One-Bit-Flip Matters! Universal DNN Model Inference Depletion with Runtime Code Fault Injection, USENIX Security Symposium, 2024 , EP2024-3767 . The paper proposes, FrameFlip, a novel attack for depleting DNN model inference with runtime code fault injections using Rowhammer, which is a security vulnerability, enabling an attacker to change the data stored within memory cells. FrameFlip operates independently of the DNN models deployed and can succeed with only a single bit-flip injection.
Sid Chau won the IEEE Transactions on Automation Science and Engineering Best New Application Paper Award for a paper about autonomously recharging drones published in 2023
Wei Shao was awarded the Best Paper Reviewer at the SIGKDD Conference (CORE A*).
Staff and Students
Dika Sukaedi from the National Cyber and Crypto Agency, Indonesia visited the CQS Group for 3 months. She was based with the Human-Centric Security Team in Clayton. Mohan Baruwal Chhetri was her mentor and M. A. P. Chamikara was her buddy. During her stay, Dika conducted a thorough analysis of existing cybersecurity indices and performed a gap analysis in relation to Indonesia’s National Cyber Security Strategy. Dika was supported through CSIRO’s Data for Development Fellowship Program.
Norbert Henseler commenced his industry PhD scholarship at Deakin University with BUPA as the industry partner, and Marthie Grobler and Tina Wu as Data61 supervisors.
Samuel Karumba, Garrison Gao, and Xiangyu Hui left our group
Mina Khan, who was an iPhD student at Deakin University, finished up with our group.
Event Organisation
Human and AI in Cybersecurity Learning (HACL) workshop was held at Deakin Downtown on 13 September 2024 to highlight several outputs of the CSCRC Executive education project portfolio, including CL@BEL, CyberBuddy and ACDC (Organisation Committee: Lauren Ferro, Marthie Grobler, Diksha Goel, Regine Richelle, Chehara Pathmabandu, Georgia Psaroulis, with participants Kristen Moore, Mohan Baruwal Chhetri, Tooba Aamir and Yue Huang)
The first workshop on Security and Resiliency of Critical Infrastructure and Space Technologies (SR-CIST) was successfully held in conjunction with the Sixth IEEE International Conference on Trust, Privacy and Security in Intelligent Systems and Applications and the 10th IEEE International Conference on Collaboration and Internet Computing (28-30 October 2024). Mohan Baruwal Chhetri and Mai Abdelhakim (University of Pittsburgh) were the workshop co-chairs.
The first workshop on Large Language Models and Cybersecurity (LLM-CyberSec) was successfully held in conjunction with the Sixth IEEE International Conference on Trust, Privacy and Security in Intelligent Systems and Applications (28-30 October 2024). Sharif Abuadbba, Kristen Morre and Yansong Gao were the workshop organisers.
Marthie Grobler is the General Co-Chair for the 28th International Symposium on Research in Attacks, Intrusions and Defenses (RAID 2025), Gold Coast, Australia, on 19-22 October, 2025.
Sharif Abuadbba, Nazatul Islam, and Raymond Zhao attended the 21st Annual International Conference on Privacy, Security and Trust 2024. Sharif Abuadbba was the local organisation chair of the event.
Shigang Liu and Sharif Abuadbba attended Usenix Security 2024 in Philadelphia, USE from 14-16 August 2024 and presented papers. As part of the same trip, Sharif Abuadbba gave talks at New York University, IBM Research Watson, and Princeton University.
Lauren Ferro gave a keynote talk at REEACH (Representation Advances And CHallenges association) titled XR’s challenges and solutions for cybersecurity (Oct 8-9, 2024, Hybrid: Padua, Italy / Online, https://www.reaach.eu/symposium-2024/)
Lauren Ferro chaired the Video and Design Exhibition Tracks at Ubicomp/ISWC 2024 (5 – 9 October in Melbourne)
Lauren Ferro and Sharif Abuadbba, were invited, along with Liming Zhu and Jamie Rossato, also of CSIRO, to participate in CSIRO’s AMA: cybersecurity (31 October, Online)
Chehara Pathmabandu, Lauren Ferro, and Tooba Aamir attended Driving Cybersecurity Change at Deakin Downtown on 18 October 2024. Chehara presented a talk on Security Software Supply Chain Management.
Diksha Goel attended the 29th European Symposium on Research in Computer Security (ESORICS 2024, a Data61 target venue) and presented the research paper “Optimizing Cyber Defense in Dynamic Active Directories through Reinforcement Learning”.
Sebastian Kish, Muhammad Usman & Eromanga Adermann were invited speakers at the CSIRO Quantum Conference 2024 held on Oct 2-3 at CSIRO Clayton.
Mohan Baruwal Chhetri gave a talk on IoT, Digital Transformation and Cybersecurity at the DataX Blue Pacific event in Sydney on 25 October organised by Asian Development Bank
Mohan Baruwal Chhetri gave an invited talk titled “Australia’s Quantum Opportunity: What does national investment mean for industry?” at Something Digital 2024 on 28 August 2024 (https://my.csiro.au/News/Missions/2024/August/Something-Digital)
Marthie Grobler and Alex Blauensteiner represented CIPR (Critical Infrastructure Protection and Resilience), with Guy Barnett and Richard Matear from Environment RU, at the NATO- and Department of Homeland Security-funded CPSICC (Cyber-Physical-Social Infrastructure in Climate Change) nexus workshop that we co-hosted with Purdue University and Sandia National Laboratories in Washington in July/August 2024, https://my.csiro.au/News/Missions/2024/September/CIPR-Washington.
Marthie Grobler participated in a panel session on ‘Turning Insights into Impact’ at the CIPR (Critical Infrastructure Protection and Resilience) co-sponsored AI4C symposium (17 October 2024) exploring the multifaceted role of AI in responsibly adapting to and mitigating against climate change through innovative talks and keynotes.
Media Interviews
Marthie Grobler was interviewed by Acuity Magazine on the role of human-centric security in cyber defence.
The Google/ CIPR (Critical Infrastructure Protection and Resilience) software supply chain project was featured in CSIRO News and CyberExpress.com The project is led by Dr Ejaz Ahmed from the CQS Group.
Marthie Grobler wrote a through leadership piece on the work that Critical Infrastructure Protection and Resilience mission is doing in the annual 2023/2024 WA Defence Review.
Funding Success
The Quantum Systems team participated in three successful proposals that were successfully funded by the Queensland State Government to work on Olympics 2032 Challenge, Quantum Computing Testbed, and Quantum Decarbonisation Mission.
The Human Centric Security team was successful in securing funding from the Australia-India Cyber and Critical Technology Partnership (AICCTP) Grant 2024 to develop Foundational Principles for Responsible Development and Use of Quantum Technologies in the Indo-Pacific Region
