Applied Intelligence Research Centre (AIRC)
Research in the School of Computing is mainly focused in the Applied Intelligence Research Centre (AIRC). The AIRC, a recognised research and development centre of the Dublin Institute of Technology, engages in researching the application of computational intelligence technologies to real world problems. The cores competencies of the AIRC include data analytics, machine learning, language technologies, intelligent agents and security.
Examples of real world problems the AIRC have addressed include spam filtering, sentiment analysis, dialogue management, custom search tools for language teachers, human-robot interaction, secure mobile financial transactions, engaging game characters, companion agents for mobile devices and the management and visualisation of large data collections.
The projects undertaken by the group are funded by SFI, Enterprise Ireland, the HEA and other funding schemes. The group is also involved in a range of commercialisation activities.
PhD projects available in the AIRC are available here.
CeADAR - the National Centre for Applied Data Analytics
CeADAR's mission is to create an internationally recognized, industry-led centre of excellence for innovation and applied research that accelerates the development, deployment and adoption of analytics technology and relevant innovations.
DIT and its academic partners in CeADAR (University College Dublin and University College Cork) work closely with Irish based companies to make the centre an internationally renowned hub for data analytics research. CeADAR is developing a world-class creative environment that fosters innovation and supports the development of a team of highly innovative technology professionals.
The AIRC is one of the three academic partners involved in CeADAR, the national Centre for Applied Data Analytics Research in Ireland.
Research Disciplines within the AIRC
Data analytics is the science of extracting actionable insight from large data collections to help people or organisations make better decisions. Data analytics uses techniques from machine learning, artificial intelligence, statistics, and natural language processing to find patterns in data and harnesses tools from data visualisation and human computer interaction to make these patterns understandable by users. In the AIRC we work on applications of data analytics in areas ranging from energy monitoring, to language learning, to retail analysis.
Contact: Dr Sarah Jane Delany
Text Analytics is a branch of Data Analytics that is concerned with extracting insight from textual data. The prevalence of user generated textual content available via the web, blogs, social networks, email, text messages and such like, means that this is a fast growing and important research area. Text analytics has its own particular challenges due to the unstructured nature of textual data. At the AIRC we work on applications of text analytics including document and spam filtering, sentiment analysis, helpfulness rating of reviews, the analysis of parliamentary data and the visualization of textual data.
Contact: Dr Sarah Jane Delany
Natural Language Processing
The field of natural language processing is concerned with developing computer systems that can process human language and automatically extract useful information from the text. There are a range of research topics related to natural language processing including: text summarization, text classification, machine translation, automatically identifying and extracting multi-word expressions and idioms, word-sense disambiguation, named entity recognition, co-reference resolution, semantic role-labelling, dialog systems, and generating textual descriptions of numeric data.
Contact: Dr John Kelleher
Intelligent agent research is a field of artificial intelligence that attempts to create software agents that can behave effectively in a given environment. Examples include game characters in simulated virtual worlds, robots in real environments and embodied agents that interact with users via digital screens. At the AIRC we investigate techniques in all of these areas to help create the next generation of intelligent agents.
Contacts: Dr John Kelleher
In Situated Interaction research we look at the application of artificial intelligence, machine learning and computational linguistics to the development of language based interaction that engages with users in real and virtual environments. This work involves the development of natural language understanding, dialogue management models and language generation in complete software architectures for robotic and virtual agents. The most recent projects have involved the investigation and application of activity recognition algorithms for modelling and tracking user behaviour based on support vector machines, HMMs and statistical grammars.
Contact: Dr Robert Ross
Security & Forensics Research Group
Affiliated closely with the AIRC is the Security and Forensics research group. The group’s work is mainly focused on the systems and applications of security in e-healthcare information systems. Some of the on-going projects in this area include:
- The design of algorithms for enabling sensitive data to be shared for research, decision making, and planning, i.e., increasing the utility of sensitive data (personally identifiable information). This research is also looking at efficient strategies through which organisations can share sensitive information without breaching the legal and compliance regime.
- Developing context aware authentication protocols for providing access to e-healthcare information systems in places where there is a shortage of healthcare professionals. This work has been focused on rural Tanzania. This research can be extended to other areas such as systems administration of critical infrastructure.
- It is not easy to determine the return on investment of security or the amount an organisation should invest in security. In this research we are working on improving predictive security metrics for measuring the security posture of organisations. This involves establishing tight baselines (thresholds) for determining whether an organisation is secure or not.
- Improving forensics strategies for collecting admissible evidence from mobile devices.
- The use of mobile banking has increased significantly in the last decade. Applications range from buying cinema tickets to catering for unbanked populations in developing countries. In this research we are working on developing strong local authentication methods using multi-level security.
Research in the algorithms design and development area in this group is mainly focused in the following two strands (1) the design of routing algorithms for the combinatorial optimisation problem, specifically using the Travelling Salesman Problem as a test bed for algorithms design and development of tight lower bounds; and (2) the design of routing algorithms in Mobile Ad Hoc Networks which are energy aware.
Contact: Dr Fred Mtenzi
There are a number of other areas where research is undertaken in the School of Computing. These include the following:
- Music Information Retrieval
- Wireless Sensor Networks
- Pervasive Systems
- Online Community Dynamics & Behaviour
- Universal Design & Assistive Technology
- Spatial Databases & Spatial Information Systems
- Software Defined Networks
Music information retrieval (MIR) is the interdisciplinary science of retrieving information from music. MIR includes musicology, psychoacoustics, signal processing and information systems. Specific ongoing research in the School of Computing centres on the award winning Tunepal project. Tunepal is the world’s leading search engine for traditional Irish music scores. It allows users to query a database of over 22,000 music scores by keyword or by playing a phrase of music on a traditional instrument (Query-By-Playing). There are mobile apps for iPhone/iPad and Android devices. It has around 15,000 regular users from over 30 countries and the technology has powered in excess of half a million music searches.
Contact: Dr Bryan Duggan
This research area involves investigating wireless sensor networks (WSNs), with a particular focus on security issues. Due to the severe memory and processing constraints of nodes that operate in such networks, traditional security mechanisms like public key encryption and other cryptographic algorithms are not viable. Therefore it is extremely difficult to design security protocols to operate in WSNs to provide adequate protection against various types of attack. The work has involved designing new light-weight security protocols designed specifically for WSNs that can protect against certain types of attacks. The research also involves the design of an architecture to support a WSN, upon which the security protocols could run autonomously to monitor the network and attempt to prevent and/or detect an attack and discard any node that has been attacked.
Contact: Dr Michael Collins
This research area focuses on pervasive systems and uncertainty reasoning techniques - including evidence theory and machine learning - and the applications of these to the detection of situations in a sensor based environment. In particular, work focuses on recognising people's activities in smart environments using sensors. With sensors embedded in everyday objects, people's activities can be automatically and remotely tracked. This is useful, for example, in allowing elderly residents to live independently, but with the support of a monitoring system in their home.
Contact: Dr Susan McKeever
This research investigates the behaviour and dynamics of online communities, including models of reputation and trust. It focuses specifically on the investigation of how financial online communities react to market crashes and the predictive power of such communities. This applied research makes use of a multidisciplinary set of techniques such as data- and text-mining techniques, along with econometric approaches, network analysis, agents, user modelling and trust.
Contact: Dr Pierpaolo Dondio
Since the early 1990s the School of Computing has been working on research projects that focus on the development of software and hardware to assist individuals with a range of special needs. This research work also focuses on how the design process can be augmented to consider a wider range of users at the planning stage to expand potential marketplaces for products and services.
Projects undertaken include:
- The Inclusive Learning Through Technology Project which was initiated in 2003 to study how a wider spectrum of abilities in the learning process might be recognised for students with a disability.
- The Education for Employment project which aimed to increase the employability of people from marginalised groups by training them to work as Technical Support Officers in Information and Communication Technology and Assistive Technology through the development and implementation of new innovative educational programmes.
- Working with Arthritis Ireland on their Easy to Use Commendation Programme to recognise those products and companies that design and market user friendly products and packaging.
- Representing Assistive Technology Systems using the World Health Organizations International Classification of Functioning (ICF) which looked at how systems, involving the components of Human, Activity and Assistive Technology in some context, may be modeled using the ICF. This representation scheme can be used as the basis for IT systems in the domain of Disability and has wider implications for the use of classifications in e-health systems.
- Integrating the ICF and related resources to improve Universal Design guidelines, a project which looked at the International impact of the ICF since 2008 in areas as diverse as design, data analytics, establishing eligibility for supports and participation, as well as the more traditional areas of health and medicine. The project looked at a framework where the ICF can be augmented with other nomenclatures and classifications such as SNOMED-CT and OMNICLASS for use in international standards in these domains of application.
CEN Community Workshop Agreement which involves developing a curriculum in Universal Design for ICT professionals
This research area focuses on the acquisition, analysis, storage, visualisation, manipulation and sharing of spatial content in spatial information systems. Effective exploitation of spatial information is now more relevant than ever, especially in the wake of recent natural disasters and extreme weather. Furthermore, urban planners are becoming increasingly aware of the need to create smarter cities for their citizens, where transportation, public safety and environmental urban planning are interconnected in order to provide safe and sustainable growth for the population.
In recent years, three-dimensional (3D) data has become increasingly available, and we see a clear desire to effectively apply this new information within a smarter cities context. One data source that has become increasingly more available is Light Detection and Ranging (LiDAR). LiDAR provides longitude and latitude information delivered in conjunction with a GPS device, and elevation information generatedby a pulse or phase laser scanner, which together provide an effective way of acquiring accurate 3D information of a terrestrial or manmade feature, thus providing highly accurate data in relatively little time covering a large geographical area. LiDAR scans provide a vast amount of data points that result in especially rich, complex point clouds. Spatial Information Systems (SISs) are critical to the hosting, querying, and analysing of such spatial data sets.
Other recent developments in this area relate to crowd sourcing of spatial content, whereby a myriad of publicly available sources are integrated that reference the same geographic object. Techniques and algorithms are needed in order to reference spatial structures from varying sources and allow the same semantic structures to be recognised irrespective of their origin.
In relation to database sphere itself and its ever growing effort to host big data, it is worthwhile to explore how novel technology, for example NoSQL, are able to empower the storage, manipulation and eventually sharing of large spatial datasets.
Contact: Dr Bianca Schoen-Phelan
Software Defined Networks (SDNs) will address the needs for flexibility and speed in meeting increased bandwidth requirements. SDNs are part of the evolution toward “X” as a Service (XaaS). They have the ability to provide a quick mechanism to fix security issues such as those recently seen with the Heartbleed bug. This can be achieved by making use of a feature known as “experimental protocol” support. In addition to this major technology companies (such as Oracle, Cisco and Google) have acquired SDN technology and entrusted it to manage their infrastructure.
Enterprise solutions are moving toward simplified networking and relying more on data-centres and cloud hosting. This shift in infrastructure means that individuals are relying more on data-centres and the provision of faster access speeds. On the other hand data centres are becoming quite complex and require significant maintenance. SDN can reduce this overhead through management from a central location.
Using SDN for research allows us to bridge the gap between repeatability and realism. The aim is to use SDN to solve existing problems as well as overcoming new challenges.
SDNs introduce a number of security vulnerabilities across its platform which are not present in traditional networks. The centralised control given to SDNs means that these issues can potentially have catastrophic effects. This could lead to manipulation of the SDN controller itself or actual traffic through the SDN controller. By identifying the vulnerabilities of the SDN platform a solution can be implemented to prevent security attacks. Any such implementation should be efficient so that it does not impact heavily on network performance.
Infrastructural work for backbone provisioning can be shown to be NP-Hard. Interdomain routing is sub-optimal and artificially constrains routes through the use of BGP. An SDN solution can address many of the issues with infrastructural problems and allows for further innovation.
Developing and deploying SDNs requires an understanding of the traffic flows. Big data analysis of traffic patterns has been proposed along with data mining for statistical extraction.
Contact: Dr Brian Keegan