Bytes

A big shout-out to Team Bumblebee from NUS Computing and College of Design and Engineering for their incredible performance at the 2024 Maritime RobotX Challenge! Held from November 3-9 at Nathan Benderson Park in Sarasota, Florida, this biennial competition brought together 15 amazing teams from Australia, Chile, Singapore, South Korea, the UK, and the US. Team Bumblebee truly stood out, taking home multiple awards:

• 1st Place in the Autonomy Challenge
• 1st Place in Design Documentation
• Best Website
• Best Team Introduction Video
• Best Systems Assessment
• Best Presentation

The Maritime RobotX Challenge is all about pushing the boundaries of autonomous robotics in the maritime world. Team Bumblebee’s achievements are a testament to their hard work, creativity, and passion for shaping the future of robotics. Congratulations to the entire team! 

At the 2024 Critical Infrastructure Security Showdown (CISS), NUS Greyhats' Team Sesame secured first place in a demanding 48-hour Capture The Flag (CTF) format. The event, focused on operational technologies (OT) and industrial control systems (ICS), challenged participants to solve complex problems using datasets provided by iTrust. Competing with seasoned professionals like Ensign Infosecurity and GovTech, Team Sesame's proficiency in analysing ICS artifacts and navigating both physical testbeds and cyber twins highlighted their technical expertise and adaptability.

Despite their limited experience in OT, the team's solid foundation in various cybersecurity competitions proved invaluable. Preparation was key to their success; they researched past CISS events and identified critical areas in OT security. Dividing tasks and regularly sharing knowledge, they deepened their understanding leading up to the event. The teamwork and communication enabled them to remain agile and focused, allowing them to adapt quickly to evolving challenges.

One of the competition's highlights came when the team exploited a vulnerability in a firmware upload system, gaining remote access to the machine. Initially trapped in a Docker container, they discovered a misconfiguration that allowed them to escape and gain root access. This unexpected success underscored how real-world vulnerabilities often arise from simple misconfigurations.

Throughout the competition, Team Sesame demonstrated exceptional teamwork and problem-solving, effectively identifying and mitigating potential threats. Their approach combined analytical thinking with practical application, showcasing their ability to operate under pressure and make informed decisions in real time.

Reflecting on the win, team leader Lim Jin Kai noted, "Competing in CISS has deepened our understanding of OT security. It has shown us how these challenges mirror real-world vulnerabilities, helping us become better cyber defenders of tomorrow." NUS Greyhats plans to continue participating in cybersecurity events, sharing their knowledge with the broader community, and inspiring the next generation of cybersecurity enthusiasts. This victory not only reinforces their status as leaders in the field but also positions them for future success in the evolving landscape of cybersecurity.

CISS 2024, recognised as a premier international cyber exercise, was sponsored by the Cyber Security Agency of Singapore and co-organised with the Singapore Armed Forces' Digital Intelligence Service, bringing together top teams from around the globe.

Associate Professor Liang Zhenkai and his PhD students, Ruan Bonan, Liu Jiahao, and Zhang Chuqi, have been honoured with the prestigious Best Practical Paper Award at the RAID 2024 Conference for their groundbreaking work titled “KernJC: Automated Vulnerable Environment Generation for Linux Kernel Vulnerabilities.” The paper, which is part of a larger effort at the National Cybersecurity R&D Lab (NCL) at NUS, introduces a novel framework, KernJC, designed to automate the generation of vulnerable environments, offering a powerful tool that greatly improves the process of assessing Linux kernel vulnerabilities.

In cybersecurity, reproducing vulnerabilities for testing and analysis has always been a challenging and time-consuming task. KernJC aims to streamline this process by automating the environment creation required to accurately reflect specific Linux kernel vulnerabilities. The framework addresses two significant hurdles faced by researchers and cybersecurity professionals: the first is identifying the correct kernel versions that contain the vulnerabilities, and the second is configuring the often complex, non-default system settings necessary to trigger those vulnerabilities in practice.

KernJC’s innovation lies in its patch-based methodology, which works in tandem with a graph-based analysis to ensure that the proper vulnerable kernel versions are selected. It also automates the configuration of system settings, which can vary widely depending on the specific vulnerability being tested. This combination allows cybersecurity teams to replicate vulnerabilities in a much more efficient and accurate manner than ever before.

The impact of KernJC on the field of cybersecurity is profound. By automating these tasks, the framework significantly reduces the manual effort required to reproduce vulnerable environments, thus speeding up the security assessment process. This advancement is critical, as it enables security professionals to more efficiently evaluate the risks posed by Linux kernel vulnerabilities, ultimately helping to safeguard systems from potential threats.

Moreover, the ability to reliably recreate vulnerable environments with KernJC will help enhance the overall understanding of how these vulnerabilities operate, supporting the development of more effective mitigation strategies. This framework represents a major step forward in the continuous effort to improve Linux system security and defend against evolving cyber threats.

The recognition of Associate Professor Liang Zhenkai and his team at RAID 2024 highlights the real-world implications of their research, demonstrating how KernJC can play a vital role in the cybersecurity landscape by improving the ability to manage and address Linux kernel vulnerabilities.

PhD students Gregory Lau Kang Ruey and Apivich Hemachandra, along with faculty members Professor Ng See-Kiong and Associate Professor Bryan Low Kian Hsiang, won the Best Paper Award at the 2024 International Conference on Machine Learning (ICML) Workshop on AI for Science. Their paper, "PINNACLE: PINN Adaptive ColLocation and Experimental Points Selection", introduces an innovative algorithm that optimises the training of Physics-Informed Neural Networks (PINNs). By leveraging Neural Tangent Kernel (NTK) analysis, PINNACLE dynamically selects and adjusts collocation and experimental points, leading to enhanced efficiency and accuracy in solving forward, inverse, and transfer learning problems. This breakthrough offers a promising new approach to tackling complex scientific and engineering challenges, showcasing significant performance improvements over current methods.

PhD alumnus Zhu Cungen and Assistant Professor Jin Chen received the Outstanding Paper Award at the 18th International Conference on Operations and Supply Chain Management (ICOSCM). Their paper, “Right to Repair and Competition in Durable Goods Markets,” uses game theory to investigate the economic and environmental effects of right-to-repair legislation. The study shows that while these laws can increase competition and benefit consumers, they may also generate additional waste. Their findings offer critical insights for policymakers looking to balance consumer rights and sustainability in such regulations.

PhD student Yang Jiong, Post-doctoral fellow Mate Soos and Associate Professor Kuldeep S. Meel along with researchers from other universities (Tan Yong Kiam and Magnus O. Myreen), received the Distinguished Paper Award at the 36th International Conference on Computer Aided Verification (CAV) 2024. Their paper, “Formally Certified Approximate Model Counting”, introduces a certification framework for approximate model counting with verified guarantees, marking a significant advancement in formal methods. 

Approximate model counting, a crucial task in various areas of computer science and artificial intelligence, involves estimating the number of solutions to Boolean formulas. While existing tools like ApproxMC provide efficient approximations, the validity of their results relies heavily on complex algorithms and optimised implementations. This new approach ensures the correctness of approximate model counting with minimal overhead.

This is Yang Jiong’s second win at CAV, having won in 2023 for his work on “Rounding Meets Approximate Model Counting”.

Team Absinthe's silver medal win at the 2024 International Collegiate Programming Contest (ICPC) World Finals marks a historic achievement. Comprising Bui Hong Duc, Rama Aryasuta Pangestu and Nyamdavaa Amar—all medallists from the International Olympiad in Informatics (IOI) in 2020-2021—the team's success highlights their exceptional programming and problem-solving abilities. This victory also underscores the commitment of their coach, Associate Professor Steven Halim, whose 24 years of unwavering dedication (as an ICPC contestant in 2001, as coach since 2008 and as ICPC contest director numerous times) have been instrumental in achieving this milestone.

Additionally, National University of Singapore (NUS) celebrates another significant milestone. Not only did the university secure its first medal at the ICPC World Finals, but it also became home to the President of the IOI for the first time. Associate Professor Tan Sun Teck was recently elected President of the IOI at IOI 2024. These achievements and recognitions reflect NUS Computing’s role as a global leader in computing education and its success in attracting top global talent through initiatives such as the SEA Scholarship and international programming competitions.

This dual celebration positions NUS Computing as a key player on the global stage in nurturing future innovators, driving research excellence, and shaping the future of computing. The university’s involvement in prestigious events like ICPC and IOI continues to reinforce its leadership position and solidify its reputation as a hub for top computing talent.

In a proud moment for NUS Computing, Associate Professor Tan Sun Teck was recently elected President of the International Olympiad in Informatics (IOI) at the IOI 2024. His new leadership role in this prestigious global competition reflects both his personal commitment to nurturing young talent and NUS Computing’s unwavering dedication to computing education.

As one of the most challenging international Science Olympiads, the IOI brings together exceptionally talented students from across the world, encouraging them to engage in complex algorithmic problem solving. Associate Professor Tan’s election to this role highlights his dedication to advancing informatics and fostering collaboration across borders.

NUS continues to solidify its place among the top institutions for nurturing talent in informatics. NUS currently ranks 3rd globally for attracting IOI medallists to enrol in its computing programmes, following closely behind MIT and Cambridge. This is a testament to the university’s ongoing commitment to excellence and its ability to attract some of the brightest minds in the field.

Researchers from the NUS Automated Program Repair research programme win the Distinguished Paper Award at the 2024 ACM Conference on the Foundations of Software Engineering (FSE) for their paper "ProveNFix: Temporal Property-Guided Program Repair". Authored by Song Yahui (Post-doctoral Fellow), Gao Xiang (former NUS PhD Student and Post-doc), Li Wenhua (PhD Student), Chin Wei-Ngan (Associate Professor) and Abhik Roychoudhury (Provost’s Chair Professor), the paper introduces ProveNFix, a tool using temporal logic to automatically detect and fix bugs. ProveNFix outperforms existing tools, quickly identifying bugs across large open-source C projects, including critical systems like OpenSSL, improving software security and reliability. ProveNFix is particularly valuable for maintaining large, complex software systems.

Professor Ooi Beng Chin, Lee Kong Chian Centennial Professor at NUS Computing, has been honoured as Fellow of the China Computer Federation (CCF).

Awarded by CCF, this distinction represents the highest honour which supports the academic and professional advancement of computing professionals worldwide.

With only 180 CCF Fellows selected globally, Professor Ooi's achievement demonstrates his outstanding contribution to the field of computing. This recognition not only reflects his dedication to pushing the frontiers of technology but also reinforces our leadership in research and innovation on the global stage.

Bui Hong Duc and Shashwat Chandra have won the Best Student Paper Award at the 2024 ACM Symposium on Principles of Distributed Computing (PODC) in Nantes, France. Their award-winning paper, "Improved All-Pairs Approximate Shortest Paths in Congested Clique", is a joint work with Chang Yi-Jun, Michal Dory (Haifa University), and Dean Leitersdorf (former Research Fellow at NUS Computing).

The All-Pairs Shortest Paths (APSP) problem, which asks for the shortest paths between all pairs of nodes in a network, is a classical problem in graph theory and computer science. Efficiently solving APSP, particularly in large-scale networks, has numerous applications, including routing, network analysis, and transportation planning. This paper showed a major improvement in reducing the number of communication rounds required to compute approximate APSP in the Congested Clique model, a fundamental computational model in parallel computation. The improvement is achieved through new techniques for analysing and constructing hopsets, which add new connections to shorten the number of hops in shortest paths, and skeleton graphs, which are smaller graphs that approximately preserve distances.

On August 21, 2024, some of the most exciting trends on AI were unfolded at NUS Computing as Google Chief Scientist Jeff Dean captivated an audience with his latest insights. Co-hosted with the NUS AI Institute (NAII), the Distinguished Speaker Seminar explored the Gemini family of multimodal models and their transformative potential across science, engineering, health, and sustainability. Prateek Jain, Director of Google DeepMind, set the stage with an introduction on the Efficiency of Large Language Models. The event concluded with a social mixer, sparking deeper dialogue among students, faculty and speakers.

PhD student Yuang Shi and Assoc Prof Wei Tsang Ooi have been recognized with the best paper award at the ACM Special Interest Group on Data Communication workshop on Emerging Multimedia Systems 2024. Their paper, "Multi-frame Bitrate Allocation of Dynamic 3D Gaussian Splatting Streaming Over Dynamic Networks”, offers innovative solutions for streaming 3D content over varying network conditions.

They collaborated with Taiwan’s National Tsing Hua University and National Yang Ming Chiao Tung University to create algorithms that efficiently allocate bitrate across multiple 3DGS frames.

This research improves adaptive streaming of complex 3D content, with potential uses in VR, remote visualization, and interactive 3D experiences.

The Department of Information Systems and Analytics at the NUS School of Computing successfully hosted the NUS MUFG DELL HumanITy Challenge from 18 May to 17 August 2024, with the generous support of two platinum sponsors – MUFG Bank and Dell Technologies. The objective of the Challenge is to provide a platform to inspire and develop young talents in designing technology-driven digital innovations to solve real-world sustainability issues for humanity. 

Targeting pre-university students from all over Singapore, the Challenge attracted 168 participants in 42 teams from 16 different high schools or junior colleges (i.e., ACS Independent, NUS High, Anderson Serangoon, Anglo-Chinese, Eunoia, Hwa Chong, Nanyang, National, Saint Andrews, Temasek, and Victoria JCs) and polytechnics (i.e., Nanyang, Ngee Ann, Republic, Singapore, and Temasek Polytechnics).

Sustainability is one of the key challenges that humanity faces today. In the Challenge, participants were tasked to identify a pressing sustainability issue within or across different pillars of sustainability (e.g., economic, social, and environmental) and develop a comprehensive and IT-driven innovative solution that addresses the problem. The solution must be sustainable, feasible, scalable, and implementable. 

The Challenge kicked off on 18 May 2024 with several academic and industry speakers sharing their knowledge and insights with the participants on the complexity of tackling sustainability issues and encouraging them to think “out of the box” when proposing their sustainability solutions. Several webinars were also conducted to provide the participants with greater exposure to various research and industry initiatives involving technology to address sustainability challenges. Based on the preliminary submissions received, 10 finalist teams were shortlisted to present and showcase their novel ideas, creative problem-solving skills, prototype mobile apps and tech gadgets that leverage the power of technology and digital innovations to tackle society’s pressing sustainability challenges for humanity.

The panel of judges for the finals were Mr. Terence Chia (Assistant Chief Executive, IMDA), who was also the guest of honour, Mr. Tan Bee Teck (Assistant Chief Executive, Admin and Social Cluster and CIO for MOE), Mr. Colin Chen (Head of ESG Finance, MUFG Bank), Mr. Nigel Chng (Director, Dell Technologies Select (Asia Pacific and Japanese Inbound)) and Mr. Richard Wilkins (CEO and Co-founder, Wilkins Technologies). They were most impressed by the finalist teams’ and students’ tech ideas and business proposals that employed the potentials and capabilities of digital technologies such as mobile apps, data analytics, AI/ML solutions and sensors, IoT devices, etc., to tackle sustainability challenges and problems that spanned from e-waste recycling, water conservation, public transport adoption, AI-driven recycling bins, light pollution reduction to ESG investment analysis. 

The winning team was “IT Girls” from Temasek Polytechnic, and the winning project was called “H20 Saver”. Their idea was to develop a small smart device that is attached to any tap to measure and control water output precisely. Further, the device learns and predicts the optimal water usage for a task over time.

The runners-up team was “ACEnders” from Anderson Serangoon Junior College with their “E-Waste Sustainability” project. Their idea was to develop a mobile app that can help people locate e-waste bins and to create a reward system to incentivize recycling. They also proposed a novel app feature to call for disposal services to dispose of large appliances.

The third-placed team was “Mickey’s Clubhouse” from Temasek Polytechnic with their submission called “EcoPulse”. With the goal of cutting down CO2 emissions from automobiles, they proposed to encourage people to use public transport by awarding points that can be converted to rewards for every train and bus ride taken. Additionally, they proposed creating sustainability challenges whose completion can be shared with the community to motivate others to take steps towards environmental sustainability. 

As a testament to our commitment to sustainability in the HumanITy Challenge, all food and beverages in the final event were sourced from and provided in sustainable sources and utensils from Treatsure and the winners’ medals were made from recycled plastic materials!

Zeyong Li, a PhD student at the NUS Centre for Quantum Technologies, has a paper accepted at the ACM Symposium on Theory of Computing, held earlier this year in Vancouver. 

The single-authored paper “Symmetric Exponential Time Requires Near-Maximum Circuit Size: Simplified, Truly Uniform” represents a significant advancement in computational complexity theory. 

By pushing the boundaries of what is understood about circuit complexity, Li’s work showcases ongoing contributions to resolve fundamental questions in computer science. These findings lay crucial groundwork for continuous development of more powerful algorithms and secure cryptographic systems to enhance computational efficiency.