At UQ, we specialise in Data-Centric AI—advancing AI performance by improving data quality, consistency, and relevance through research in multimodal data management, preparation, governance, privacy, and security.
Capabilities
Data-Centric AI Foundations
Enhancing AI performance requires shifting from model-centric optimisation to strategic data curation that prioritises quality, consistency, and relevance, while actively addressing real-world challenges such as data scarcity, annotation errors, noise, and bias.
Advanced Data Management
Advancing robust, transparent, and socially responsible AI systems requires research in scalable and target-driven multimodal data lake management and preparation techniques, underpinned by strong expertise in data governance, privacy preservation, and information security to ensure trustworthy data practices.
Diverse Data Modalities
By working with semantically structured data, time series, multimedia, and spatio-temporal datasets, AI systems can become more adaptive, scalable, and better aligned with complex operational realities.
Ethical and Trustworthy AI
Ensuring AI systems are built on reliable foundations involves tackling misinformation and data bias, while promoting transparency and trust throughout data pipelines and decision-making processes.
Impact
Scalable Machine Learning with Data-Centric AI — Developed a system to discover, enrich, and select multimodal data for targeted tasks, enabling faster, scalable, and cost-effective preparation that supports adaptive, data-efficient enterprise AI aligned with real-world needs.
In partnership with Health and Wellbeing Queensland, we are advancing causal AI for diabetes prevention by developing active learning algorithms that enable high-quality inference from limited, ethically sourced observational data—achieving up to 11 percent accuracy improvements while minimising data acquisition costs.