Project Summary:
Streaming data analytics in large sensor networks is increasingly important due to both the volume of data collected and the real-time decision making nature. Cyber-physical systems, social networks/social media, smart grids, smart transport systems are just few examples of large, distributed sensor networks that would benefit from real-time streaming data analytics for achieving situation awareness.
This project continues the existing work developed in KDE on graph-based data mining for both historic data and online streaming data. The primary areas to investigate are anomaly detection and concept drifting in a single sensor data stream and across multiple sensor data streams. Anomalies represent unusual situations that may need further investigation whilst concept drifting research is about how to discover gradual situation changes that occur over time, which may appear as anomalies in the first instance.
Contact Details:
Weiru Liu
Email: w.liu@qub.ac.uk
Telephone:+44 (0)28 9097 4896
2. DEVELOPING ADAPTIVE CAPABILITIES FOR AUTONOMOUS AGENTS IN UNCERTAIN ENVIRONMENT
Project Summary:
Future autonomous agents/robots will help humans to achieve many tasks ranging from repetitive trivial ones to significant ones conducted in inaccessible and hostile environments for humans (e.g., disaster responses, search and rescue, military engagements). Such robots are increasingly sophisticated and incorporate powerful autonomous capabilities. They are capable of operating individually or working alongside humans to achieve their goals [A glimpse at our robotic future: 235 start-ups reviewed (World Robotics Service Robots 2013, pp. 191-198)].
For this project, we want to develop novel reasoning algorithms for a robot to have situation
awareness capabilities for decision making. To achieve this, a robot or more generally an autonomous agent,
must possess at least three inter-related intrinsic abilities: continuously perceive and fuse uncertain,
inconsistent, or erroneous information to recognize situation changes; and dynamic online planning (and taking actions)
to respond to its new beliefs about the environment and about other agents.
Contact Details:
Weiru Liu
Email: w.liu@qub.ac.uk
Telephone:+44 (0)28 9097 4896
Research Interests
1. Theories of Reasoning under Uncertainty, Uncertain Information Fusion, Multi-criteria Decision Making under Uncertainty:
This research is concerned with modelling, reasoning, and merging uncertain information from heterogeneous sources
in any intelligent systems (e.g., large sensor networks). We particularly focus on the Dempster-Shafer theory of Evidence
(belief function theory), possibility theory and possibilistic logic, and probability theory.
We research into
the appropriateness of modelling uncertain information using these formalisms, aggregation
approaches offered by them, and conflict/inconsistency analysis among multiple
pieces of uncertain information within these theories. Recent work has advanced to
handling ambiguous evidence in game theory for security,
and multi-criteria decision making under uncertainty in complex systems.
2. Data mining, large scale data analytics, anomaly/threats detection
Knowledge acquisition is expensive and often there are no experts around from whom
to elicit the knowledge. We develop Machine Learning and Data Mining algorithms
to discover
knowledge from data that are easily comprehensible to humans.
Our earliest work was on developing algorithms for constructing Bayesian
Networks from data.
Recent work has been focusing on graph-based approaches for
both historic and streaming data analytics, with numerous applications.
Design and develop anomaly detection algorithms for detecting abnormal
behaviors (anomalies) in physical access control environment under the context of security
within CSIT.
Develop graph-theory based algorithms for identifying exercise
patterns and influences among participants in events.
Discover social connection patterns from social networks with streaming data.
Design and develop various data analytical approaches, in collaboration with
Belfast City Council, for analyzing data on Pollution, Waste disposal/treatment, Recycling; Anti-Social Behaviors, etc.
Design and develop real-time threats and anomaly prediction
algorithms with missing values in datasets, using knowledge discovered above,
to provide real-time situation awareness for decision support.
3. Intelligent Autonomous Systems and Event-reasoning based Situation Awareness:
Our theoretical research includes uncertain information modelling and fusion, and planning under uncertainty. This research has led to:
Design and develop a multi-agent based event reasoning framework for correlating dispersed events detected from heterogeneous sources
in a distributed complex environment for achieving situation awareness. Applications include intelligent surveillance in cyber-physical systems,
smart homes, and intelligent energy and transport management in smart cities.
Design and develop intelligent autonomous systems
using multi-agent techniques for complex control problems and for designing collaborative (software) agents, or mixed teams of
multi-robots and human for working
together in complex environment.
Applications include search and rescue, services, complex industrial control problems,
and games for entertainment or education.
4. Theoretical aspects of Merging/Revising Uncertain and Inconsistent Knowledgebases:
Our research includes developing fusion methods (merging operators) and algorithms for merging
multiple knowledgebases (maybe with constrains), especially, propositional and possibilistic knowledgebases,
stratified knowledgebases, imprecise probabilistic logic based knowledge bases,
and heterogeneous uncertain information. We also develop
revision strategies/operators for revising such knowledge/belief bases.
Recent research has progressed to providing a toolkit for identifying minimal
inconsistent subsets and calculating
inconsistency values of knowledgbases or individual formulae in large scale knowledge bases. This research has also been extended to developing approaches for
detecting inconsistencies in probabilistic knowledge bases (learned by other machine learning systems) and
repairing such inconsistencies.
(NEW)
Anomaly Detection System for Insurance Claim and Point of Sale (Allstate and Invest NI, 2016-2019), PI.
(NEW)
Leverhulme Interdisciplinary Network on Cybersecurity and Societal (LINCS) (Leverhulme Trust. 2015-2019), Co-I.
(NEW)
CSIT 2: EPSRC / TSB, (2015-2019), Co-I.
(NEW)
DEVELOP: Developing Careers through Social Networks and Transversal Skills (EU-Horizon 2020, March 2016- Feb. 2019), Co-I.
LNiK: Northern Ireland Administrative Data Research Centre (ESRC: Ref ES/L007509/1, 2014-2019), Co-I.
PAuSE: Personalising Autonomous Systems (MoD/DSTL 2014) Co-I.
WAVeTrack: Graph-based Trajectory Mining for Knowledge Discovery of Vehicular Movements (MoD/DSTL 2014) Co-I.
PACES Providing Autonomous Capabilities for Evolving SCADA (EPSRC 2012-2015), PI
ARIES Accelerated Real-Time Information Extraction System (EPSRC 2012-2013),
Co-I
TEI@I Framework for Forecasting and Decision Making
(The Royal Academy of Engineering 2012-2013), PI
(In collaboration with Prof Gang Xie and Prof Xiaoguang Yang at
Chinese Academy of Sciences, China)
Weiru Liu and Kedian Mu (Guest Editors): Special Issue of International Journal of Approximate Reasoning
2016.
Weiru Liu and Henri Prade (Guest Editors): Special Issue of Fuzzy Sets and Systems 2013.
Weiru Liu (Guest Editor): Special Issue of International Journal of Approximate Reasoning, 2013.
Conference Chairs/Steering Committee member:
Program Co-Chair of the 7th International Conference on Scalable Uncertainty Management
(SUM'13), 16-18 September, 2013. Washington DC Area, USA.
Conference/Program Chair of the Eleventh European Conference on Symbolic and Quantitative Approaches to
Reasoning under Uncertainty
(ECSQARU'11). June 29the - July 1st, 2011.
Belfast, Northern Ireland, UK.
