Human-centred Intelligent Learning and Software Technologies (HilstLab)

Research cluster

Background

Our research is centred around humans with an aim to develop and/or study applications that can solve real-world problems. With the two technological pillars of intelligent learning and software, we develop a range of industrial applications and study their impacts on user experience and task performance. Human-centredness means humans are not just users of a technological solution; they are a central part of the solution. Our research is especially focused on 'interaction': human-AI interaction, human-software interaction, and the integration of AI with software.  In addition to the development of innovative human-centred intelligent learning and software technologies and their industrial applications such as in defence, health, and environmental sustainability, we are also interested in the potential impacts of these technologies on humans such as user experience and task performance in real-world applications.

Program of research

The cluster’s program of research is comprised of three substantive themes:

• Human-centred Artificial Intelligence (AI): human intelligence and artificial intelligence are seamlessly integrated to provide a solution that outperforms each one alone. This theme investigates better AI models and algorithms, human-AI interactive sensemaking, human-AI partnership, and AI ethics, trust, and security.
• Human-centred Software Engineering (SE): it considers the human aspects of SE such as culture, personality, collaboration, interaction and experience, as majority of software is developed by humans to be used by other humans. This theme investigates user-centred design, UI, UX, interaction design, groupware, and human aspects of software engineering.
• Integration of AI and SE: AI and SE are integrated so that the developed software allows a human to partner with an AI agent. This theme investigates intelligent software, and intelligent software engineering.

Industry project: Context-aware Explainable Human-AI Interaction by Associate Professor Haifeng Shen, Dr Kewen Liao, and Dr Theodor Wyeld

Collaborative Partners: Western Sydney University, Defence Science and Technology Group, and Total Technology Partners Pty Ltd.

This project is sponsored by Defence Industry & Innovation's Next Generation Technologies Fund to develop a collaborative human-centred AI (CHAI) system for submarine control room console operators and study its impact on the decision-making performance in a high risk, time critical environment. The system provides a dedicated CHAI assistant for each type of operator that is tailored to the specific context of each operator's task and types/sources of data to be processed in order to reduce the number of dimensions of information each console operator has to process at any given moment and assist their analysis and decision. A pilot study shows that a CHAI system helped participants achieve both faster response times for all tasks and higher accuracy. In turn, measured and perceived workload was reduced for those participants that used the system.

Industry project: Context-aware Visualisation as a Service by Associate Professor Haifeng Shen and Dr Theodor Wyeld

Collaborative Partners: University of New South Wales, Defence Science and Technology Group

This project was sponsored by Defence Industry & Innovation's Next Generation Technologies Fund to develop a framework and a concept demonstrator showing the feasibility of running high-end graphics-intensive visualisations across multiple platforms ranging from wall displays to mobile phones through CaaS (Containers as a Service) whereby cloud computing service providers can offer container-based virtualisation as a scalable online service to a range of clients that do not require high-end graphic devices. CaaS enables the users to develop, test, execute or distribute the software in the application containers and use the container services without the need to have their own high-end infrastructure. An important benefit of CaaS is that it simplifies the process of deploying applications in the cloud.

In addition to making it easier to deploy applications in the cloud, a core reason for adopting CaaS is to increase both portability and adaptability of the applications. Visualisation applications are usually resource intensive and often require specific hardware and software to run. With CaaS, visualisation applications can be deployed in the cloud that provides the specific hardware and software resources required. A variety of devices using only a Virtual Network Client (VNC) can then access the full features of these applications. Furthermore, the visualisation applications become context-aware – different containerised visualisation applications can be custom-created to service the clients' devices, such as smartphone, tablet, PC, Head-Mounted Display (HMD), dataglove, or data wall among other devices. On top of portability and adaptability, the visualisation applications can take full advantage of the cloud computing resources such as high-performance computing services (HPC) and GPU clusters. These resources can be exploited to parallel process machine learning algorithms and accelerate the graphics rendering pipeline that the visualisation applications require before passing the desktop or app to a myriad of remote devices. Implementation of this architecture using Docker containerisation provides a flexible, scalable, and secure approach that can be coupled with nVidia's CUDA kernel to improve overall performance through graphics acceleration and GPU-enabled parallel processing.

Cluster members

The cluster is composed of a core group of outstanding scholars who share common research interests. Members have a shared commitment to ACU's research intensification mission, and bring complementary skill sets and knowledge to the Cluster in terms of substantive expertise, research methods, publishing, governance, funding, and HDR supervision. The cluster members are:

• Associate Professor Haifeng Shen – haifeng.shen@acu.edu.au
• Dr Kewen Liao – kewen.liao@acu.edu.au
• Dr Maoying Qiao – maoying.qiao@acu.edu.au

Selected publications

• Haifeng Shen, Kewen Liao, Zhibin Liao, Job Doornberg, Maoying Qiao, Anton van den Hengel, Johan W. Verjans (2021). Human-AI Interactive and Continuous Sensemaking: A Case Study of Image Classification using Scribble Attention Maps. ACM Conference on Human Factors in Computing Systems Extended Abstracts, Article No.: 290, Pages 1–8. CORE: A*
• Theodor Wyeld, Peerumporn Jiranantanagorn, Haifeng Shen, Kewen Liao, Tomasz Bednarz (2021). Understanding the effects of real-time sentiment analysis and morale visualisation in backchannel systems: A case study. International Journal of Human-Computer Studies, 145:102524. CORE: A, JCR: Q1
• Lanxin Yang, He Zhang, Haifeng Shen, Xin Huang, Xin Zhou, Guoping Rong, Dong Shao (2021). Quality Assessment in Systematic Literature Reviews: A Software Engineering Perspective. Information and Software Technology, 30:106397. CORE: A, JCR: Q1
• Maoying Qiao, Jun Yu, Tongliang Liu, Xinchao Wang, Dacheng Tao (2020). Diversified Bayesian Nonnegative Matrix Factorization. The 34th AAAI Conference on Artificial Intelligence (AAAI), 5420-5427. CORE: A*
• Lu Chen, Chengfei Liu, Kewen Liao, Jianxin Li, Rui Zhou (2019). Contextual Community Search Over Large Social Networks. The 35th IEEE International Conference on Data Engineering (ICDE), 88-99. CORE: A*
• Longkun Guo, Yunyun Deng, Kewen Liao, Qiang He, Timos Sellis, Zheshan Hu (2018). A Fast Algorithm for Optimally Finding Partially Disjoint Shortest Paths. The 27th International Joint Conference on Artificial Intelligence (IJCAI), 1456-1462. CORE: A*
• Maoying Qiao, Liu Liu, Jun Yu, Chang Xu, Dacheng Tao (2017). Diversified dictionaries for multi-instance learning. Pattern Recognition, 64:407-416. CORE: A*, JCR: Q1
• Kewen Liao, Alistair Moffat, Matthias Petri, Anthony Wirth (2017). A Cost Model for Long-Term Compressed Data Retention. The 10th ACM International Conference on Web Search and Data Mining (WSDM), 241-249. CORE: A*
• Maoying Qiao, Jun Yu, Wei Bian, Qiang Li, Dacheng Tao (2017). Improving Stochastic Block Models by Incorporating Power-Law Degree Characteristic. The 26th International Joint Conference on Artificial Intelligence (IJCAI), 2620-2626. CORE: A*
• Qiang Li, Maoying Qiao, Wei Bian, Dacheng Tao (2016). Conditional Graphical Lasso for Multi-label Image Classification. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2977-2986. CORE: A*
• Kewen Liao, Matthias Petri, Alistair Moffat, Anthony Wirth (2016). Effective Construction of Relative Lempel-Ziv Dictionaries. The 25th International Conference on World Wide Web (WWW), 807-816. CORE: A*
• Maoying Qiao, Wei Bian, Richard Yi Da Xu, Dacheng Tao (2015). Diversified Hidden Markov Models for Sequential Labeling. IEEE Transactions on Knowledge and Data Engineering, 27(11): 2947-2960. CORE: A*, JCR: Q1
• Haifeng Shen, Mark D. Reilly (2012). Personalized multi-user view and content synchronization and retrieval in real-time mobile social software applications. Journal of Computer and System Sciences, 78(4):1185-1203. CORE: A*, JCR: Q1
• Chengzheng Sun, Steven Xia, David Sun, David Chen, Haifeng Shen, Wentong Cai (2006). Transparent adaptation of single-user applications for multi-user real-time collaboration. ACM Transactions on Computer-Human Interaction (TOCHI), 13(4): 531-582. CORE: A*, JCR: Q1

For more information on the research at HilstLab, please visit https://hilstlab.org.