Data Dynamics

Current project

VIP Snapshot

Diving into the data big bang - do cool things that help us manage massive data and support data-driven science, healthy living and global mobility!

Data Dynamics

VIP ChallENG research goals

The massive spread of false information on social media has become a global risk, implicitly influencing public opinion and threatening social/political development. False information detection (FID) has thus become a surging research topic in recent years.

As a promising and rapidly developing research field, we will explore the new research problems and approaches of FID, including early detection, detection by multimodal data fusion, and explanatory detection, model adaptivity/generality to new events, embracing of novel machine learning models, aggregation of crowd wisdom, adversarial attacks and defence in detection models.

Our research includes:

  • Data acquisition and storage, data labelling and augmentation algorithms and data representation and association learning.
  • Exploring new machine learning models of analysing multimedia data such as videos, audios and texts.
  • Descriptive, predictive and prescriptive modelling.
  • Data visualisation and model explanation.

Research Areas:

  • Machine Learning
  • Artificial Intelligence
  • Data Mining
  • Human-Computer Interactions
  • Transport System Engineering
  • Large-scale Control and Optimisation
  • Natural Language Processing
  • Computer Vision
  • Computer Science / Information Technology
  • Machine Learning / Algorithm designing
  • Web programming / Mobile app development / Software Engineering
  • Embedded systems
  • Data Visualisation
  • Mathematics / Statistics

2021 VIP Consortium Innovation Competition - Project Video

Data Dynamics – Invisible Vision Team

Late medical diagnosis is among the leading causes of death across the world. With the vision that “Every second matters”, the Invisible Vision Team has developed a smart-phone app that can detect various diseases based off medical images. With an accuracy of over 80%, it can successfully detect covid-19, brain cancer and breast cancer!

Explore the Data Dynamics Projects

Below are the various aspects of Data Dynamics you can choose to explore.

The increasingly digital nature of urban infrastructures and services is radically changing the way we work, live and travel, and bringing about novel data availability for urban systems. This project, by utilising multiple sources of big data, will quantify human activity and mobility patterns in urban context, and develop integrated approaches for complex human flows. Datasets to be utilised include, e.g., Opal card data. This project will involve collaboration with researchers from the School of Civil and Environmental Engineering.

The Smart Cities and Suburbs Program is a major Government initiative to improve the liveability, productivity and sustainability of cities and towns across Australia. Along with aims and objectives of this program is improving mobility and increasing accessibility in metropolitan areas. This research topic is focused on extracting meaningful patterns from various sources of data (e.g., social media, sensor and trajectory data) and reconcile these identified patterns to come up with plans and strategies to improve traffic efficiency and mobility. Students with strong interests in e.g., big data analytics, programming, and mathematical modelling.

Human mobility analysis plays an important role in many smart city applications, such as traffic congestion control, ride-sharing dispatch and hotspot detection. However, accurately understanding human mobility is a challenging task due to various complex impact factors from different domains, including dynamic spatial and temporal correlations, as well as the properties of each location (e.g. Point of Interests). In this project, we propose to integrate satellite imagery data, which is ready to collect, to characterize each region of the city and further enhance the deep learning models for analyzing human mobility patterns in a city scale.

Spatial-temporal pattern analysis is a fundamental problem for enabling smart city. However, mining spatial-temporal patterns normally requires a large amount of data, which is hard to collect and may be not available (e.g. when companies start business in a new city). In this project, we propose to utilize deep transfer learning technique to solve this problem. The spatial-temporal patterns learned from data abundant cities will be transferred to the target data insufficient city, which solves the data scarcity problem and helps understanding the spatial-temporal patterns in the target city.

Connected and autonomous vehicles (CAVs) are increasingly ready for at scale deployment and are expected to improve existing mobility systems significantly. It will be many years before CAVs adoption is widespread despite recent developments in technology Deployment of CAVs in the existing network create a mixed traffic environment which is a cooperative-uncooperative game with human drivers seeking to minimise their own travel time (following user equilibrium (UE) behaviour) while CAVs may be centrally controlled to minimise the total system travel time (following system optimum (SO) behaviour). CAVs can unlock the potential for system optimal traffic assignment and thus, it is of importance to investigate how CAVs may be leveraged to better manage and operate our transport networks. In particular, the question remains on what the minimum control ratio and operational model of CAVs in road networks in order to ensure traffic efficiency.

Many urban areas experience significant traffic congestion from day to day. In recent years, there is a growing and unprecedented trend in applying reinforcement learning (RL) for traffic signal control. The main challenge is to apply these techniques to optimize traffic control and coordination to account for real-time operation, large-scale networks, and traffic stochasticity. We will integrate RL techniques with transport and control theories for traffic control problems.

Team Academic Lead