Al-Rifaie on BBC

AISB Committee member and Research Fellow at Goldsmiths, University of London, Dr Mohammad Majid al-Rifaie was interviewed by the BBC (in Farsi) along with his colleague Mohammad Ali Javaheri Javid on the 6 November 2014. He was a...


Read More...

Rose wins the Loebne...

After 2 hours of judging at Bletchley Park, 'Rose' by Bruce Wilcox was declared the winner of the Loebner Prize 2014, held in conjunction with the AISB.  The event was well attended, film live by Sky News and the special guest jud...


Read More...

AISB Convention 2015

The AISB Convention is an annual conference covering the range of AI and Cognitive Science, organised by the Society for the Study of Artificial Intelligence and Simulation of Behaviour. The 2015 Convention will be held at the Uni...


Read More...

Yasemin Erden on BBC

AISB Committee member, and Philosophy Programme Director and Lecturer, Dr Yasemin J. Erden interviewed for the BBC on 29 October 2013. Speaking on the Today programme for BBC Radio 4, as well as the Business Report for BBC world N...


Read More...

Mark Bishop on BBC ...

Mark Bishop, Chair of the Study of Artificial Intelligence and the Simulation of Behaviour, appeared on Newsnight to discuss the ethics of ‘killer robots’. He was approached to give his view on a report raising questions on the et...


Read More...

AISB YouTube Channel

The AISB has launched a YouTube channel: http://www.youtube.com/user/AISBTube (http://www.youtube.com/user/AISBTube). The channel currently holds a number of videos from the AISB 2010 Convention. Videos include the AISB round t...


Read More...
012345

Notice

AISB opportunities Bulletin Item

PhD Scholarships, Computer Science, University of Hull

http://www.dcs.hull.ac.uk/pgr/phd_proposals_web.htm

PhD Scholarships 2008 

The Department of Computer Science at the University of Hull is offering a small number of Full Fees plus 3000 pa stipend PhD scholarships. These will be targeted at exceptional students with no other means of funding a research degree, who are interested in specific research areas. Example projects are detailed below. Note there may be room for negotiation on the exact nature of the project during the application process but contact the member of staff involved to clarify areas of mutual interest. Full details are available at: http://www.dcs.hull.ac.uk/pgr/phd_proposals_web.htm 

--- Darryl N. Davis 

Director of Research
Director of Postgraduate Research Studies 

Computer Science, University of Hull, HU6 7RX.
fax:    +44(0)1482466666
tel:    +44(0)1482466469
http://www2.dcs.hull.ac.uk/NEAT/dnd/index.htm
email:  d.n.davis@hull.ac.uk 

  

Short Descriptions of Prospective PhD Projects 

Leonardo Bottaci 

Automatic Test Data Generation (LB1) 

The automatic generation of test data for the structural unit testing of software is an active area of software engineering research. This project is directed at tackling this problem through the application of program analysis and heuristic search techniques. The program under unit test is analysed and instrumented to gather information about the extent to which any generated test case satisfies the test coverage criterion. During the search, candidate test cases are executed and evaluated on the basis of their contribution to the coverage of the test goals. A key research problem is to develop effective methods to analyse and instrument the program under test in order that the execution data collected can be used to guide the search efficiently. Simple coverage instrumentation is usually ineffective because it cannot distinguish between candidate test cases that are "near misses" and those that are "distant misses". Particular problems arise in programs that contain loops and flag variables. The prevalence of information hiding and encapsulation in object oriented programs makes them more difficult to instrument effectively. Recent work in the automatic generation of test data for programs that perform string manipulation has been published in Bottaci, L & Alshraideh, M. "Search-based software test data generation for string data using program-specific search operators" 2006 Software Testing Verification and Reliability 16 (3), pp. 175-203 

Construction of a Mutation Testing System (LB2) 

A Mutation (also known as a fault-based) testing system is a system for evaluating the fault detection effectiveness of a given set of test cases. The program under test is modified slightly with the intention of introducing a "fault". The given test set is then run on both the original program and the modified program (known as a mutant program) and the test results are compared. If the results are identical for all tests then the test set is considered inadequate with respect to detection of the given mutant. The use of this technique in practice requires a number of problems to be solved. A typical unit program under test will give rise to many thousands of mutants. An efficient method is required therefore to generate and execute these mutants. Current methods use meta-programming techniques and low level language (byte-code) manipulation, see (Yu-Seung Ma, Offutt, Y, MuJava: An Automated Class Mutation System, Software Testing, Verification & Reliability 15(2) 2005). The project goal is to optimise the use of meta-programming techniques by performing a more sophisticated analysis of the program under test. The new techniques will be implemented and evaluated in a prototype mutation system for Java programs. 

Dr Mike Brayshaw (with Brian Tompsett and Neil Gordon) 

Disability Inclusion and Accessibility to Learning Environments with Computer Technology (MB1) 

Much work has been done on applying advanced technology to learning, and with the advent of the internet this has evolved into modern virtual learning environments. Similarly, much work has been performed on using technology to assist disabled people on achieving equal access to society, and in particular with their participation in education. Many of those aids use digital technology and particularly some form of assistive software on a personal computer. 

The interaction of the two technology applications of the computer and internet for learning and as assistive technology, has in some cases, created new forms of unequal access to learning. The aim of this project, built on extensive experience at Hull in both learning and assistive technologies, is to investigate how these forms of computer support for learning can be better integrated and resulting in a unified environment that is both assistive to all users and a good learning environment. 

Paul M Chapman 

Visualisations of Krill Swarms (PMC1) 

With Dr Magnus Johnson (Centre for Coastal Studies) & Dr Geraint Tarling (British Antarctic Survey) 

We propose to develop a virtual 3D krill swarm that will allow us to develop a model where the state of individuals (stomach fullness, moult stage, reproductive stage, swimming capacity) within a swarm can interact with environmental variables (food availability, day length, temperature, swarm composition). A handicap to many biological investigations is the inability to interface with complex mathematical simulations. Collaborations between applied mathematicians and biologists lead to frustrating misunderstandings because one does not understand the problems, constraints and language of the other. The visual nature of the model developed will provide an accessible medium through which both can investigate the impact of complex mathematical individual interactions on swarm behaviour and ecology. 

The Hull Immersive Visualization Environment centre (HIVE) provides excellent facilities for this project by seemingly immersing the participants into the underwater environment. User interactions with the marine model will be through intuitive interactive interfaces that provide researchers with an almost infinite number of 'what if?' scenarios. This 'visualization' tool will provide researchers with the necessary tools to gain insight and a new understanding of the complex generated krill swarm data that would not be feasible using existing techniques. The tools that we develop will not be limited to million pound VR centres but also a cut-down version of the software would be available for schools and colleges that use standard PCs and will greatly facilitate the learning process. 

HMAP Visualization (PMC2) 

HMAP is the historical component of the Census of Marine Life (CoML), an international research program assessing and explaining the diversity, distribution, and abundance of marine organisms throughout the world's oceans. (http://www.hull.ac.uk/hmap/ ). They have a huge census database of marine life which is in pure numerical format. This research will involve the development of an interesting, possible virtual reality graphical interface for their data. This system would then be located at The Deep. A second system would do a tour around the UK. Our objective would be for the general public to have an improved understanding of things like depleting fishing stocks, whaling etc. It's possible we would also develop a web front end for anyone around the world to access. Collaborating with Scarborough, IECS (Institute for Estuarine Coastal Studies) and the Maritime Historical Studies Centre at Hull University. 

Darryl N. Davis 

Learning and Adaptation in a Cognitive Robot (DND1) 

Cognitive robots is a growing area of research interest. In bringing together many areas of interest to cognitive science, such as decision making, knowledge representation, motivation and perception, and the control of autonomous (self governing) systems, the aim is to build robots that are more adaptable and capable of working in unknown environments. Such research is at the heart of the UK Computing Research Committee Grand Challenge 5: Architecture of Brain and Mind - Integrating high level cognitive processes with brain mechanisms and functions in a working robot. The overall aim of this project is to investigate heterogeneous agent architectures suitable for the Minsky Society of Mind metaphor. Computational models of affect (and motivation) will be used to drive the organisation of these agents. It is expected that a number of implementations will ensue, and can incorporate artificial life, artificial neural network and goal-oriented processes. Although an extensive investigation of learning mechanisms will probably be outside the scope of this project, we hope to determine what other types of learning mechanism might be appropriate to different levels of processing; and in particular how these learning mechanisms can be integrated within a sophisticated (general purpose) agent. A fully equipped robot lab is available to support this project. 

Neuro-Fuzzy Data Mining in Medicine (DND2) 

Accurate models and analytical systems are urgently needed to predict outcome following vascular surgical intervention and to aid in decision making at the clinical level. Existing models, usually based on linear statistical analysis, have proved disappointing. The adoption of clinical governance in the NHS has mandated that we must develop appropriate and reliable clinical data-sets for use in comparative audit. These data-sets will be useless without the ability to interrogate and analyse them in a meaningful way. A validated model would allow us to set achievable national standards and thereby to improve quality of care through out vascular units in the UK by implementing guidelines and allowing comparative audit using local and national data-sets. 

The aim of this project is to create a neuro-fuzzy decision support system capable of meeting the above requirements and other clinical domains. The proposed system will use both linear and non-linear models and compare these against artificial neural networks (ANNs), fuzzy logic systems and genetic algorithms. We will also use hybrid models based on these techniques to develop decision support systems whose rules can be extracted and analysed by a human being. The bulk of the research will be in developing neuro-fuzzy techniques suitable for inclusion in such systems, and the use of fuzzy logic in combing classifier outcomes for practical decision making in medicine. Clinical data from this and other domains is available for use in this project. 

Neil Gordon 

Composite projections of geographic data (NG1) 

The aim of this project is to investigate and develop alternative projections to the usual orthogonal and linear perspective ones supported, creating projection matrices to perform arbitrary projective transformations, applying these to display terrain and other data and investigating the use of deformations of the underlying data. 

Previous projects at Hull have shown that an unconventional visualisation pipeline, involving an oblique parallel projection of linear deformations of the model, are more intuitive and easier to control in 2.5D, and previous work has explored the scope in 3D for going beyond linear perspective through use of model deformations, but only with parallel projections. 

The mathematics for complex projections: projective geometry techniques are ubiquitous in computer science problems involving 3-dimensional graphics. However, in general the application only makes use of a small number of well-known techniques, not far beyond the ideas used in the pre-computer development of the subject. Applying a number of projective mappings, along with techniques to preserve the topological properties, can give different and useful emphasis of content. This project would investigate mathematical projective geometry methods in novel ways, relying on the power of the computer to automate the process. If successful, this technique could be applied to full surface plots to provide an innovative yet informative view of data, and further be combined with techniques from the CISRG to provide distinctive views of data. 

Applying deformations of the model along with the above projections would then be investigated, along with the potential to make use of the true 3D visualisation technologies in HIVE. 

Empowering learning through technology (Neil Gordon & Mike Brayshaw) (NG2) 

The aim of this project is to investigate the effectiveness of technological solutions to leaning in higher education, and their potential impact on society. One particular case study would be that of the use of learning technologies to support the transition for students from school and college into HE (one of the foci of the University Learning and teaching themes). 

The project would look at evaluating approaches and building flexible teaching environments within the context of reusable learning objects and standards for these. This is likely to require studying of different models of education how these are implemented through models of learning systems. 

This work would build on previous work in the department - linking research on learning styles, computer based learning, and flexible learning environments that have been undertaken with the DRIS research group (Brayshaw and Gordon). The potential for these learning objects to provide support for students with special educational needs offers other links with existing departmental projects (Tompsett). 

Chandrasekhar Kambhampati 

Networked Embedded Systems (CK1) 

Modern systems are characterized by three key features (a) the ubiquitous presence of embedded systems and (b) the networking of these embedded systems and (c) their resultant complexity.  These features have lead to computing (in all its forms) being pervasive, and the decision making systems becoming more complex. This pervasiveness and the interconnectedness of the systems have increased the need for more and better fault tolerance, and also for increased efficiency in their performances. Networking allows for true self-organization, in that the connections which are useful for overall performance are retained, whilst others are discarded.  

In critical applications, where networked embedded systems are present, it is important to maintain, control and monitor the performance of the system.  Often the control and monitoring take on two forms (a) coordinating the efforts (performance) of the system various components (distributed control)and (b) ensuring that the system maintains a desired performance in event of a fault. Once a fault has been diagnosed, the system should have the ability to reconfigure itself in order to either maintain performance or to ensure that the fault does not lead to a total failure in the system.   A system which is able to coordinate the efforts, diagnose a fault, isolate it and reconfigure itself would improve dependability, provide greater autonomy. At the same time it would enable the system to have nomadic components (plug-and-play), increase modularity, incorporate redundancy. 

There are two projects in this area: (a) Pervasive Embedded Systems: Network of reconfigurable multi-agent systems and (b) Reconfiguration and Fault-Tolerance in Plug-and-Play Network of Embedded Systems 

Real Neurons and Quantum Networks (CK2) 

A key factor in introducing real neurons into the feed loop when interfaced with prosthesis is the ability to distinguish between different stimuli. Similarity between two spike trains is generally estimated using a 'coincidence factor'. This factor relies on counting coincidences of firing-times for spikes in a given time window. The research group has developed a computational framework for this . The next stage of the project is to introduce a real neuron and close the loop. 

The research group is also investigating the development of algorithms using Quantum Princinples for Artificial Neural networks. Filter for signal processing based on Quantum Neural Network has received a lot of interest in the recent past. Implementation of this filter has a number of design parameters which sometime leads to numerical inefficiencies. Solution procedures employed in that the evolution of the time varying functions had to be controlled which often leads to numerical instabilities. This project will investigate procedures for the (a) stability (b) numerical efficiency and (c) accuracy of Quantum Recurrent Neural Network filter. Results will be compared with the filters available in the literature. In the first instance a one dimensional filtering example will be used to illustrate the principles employed. The procedure is then to be extended to the two dimensional motion detection and filtering problem 

Thus there are two possible immediate projects in this area, viz (a) Closing the loop with real neurons (b) Filtering using Quantum Recurrent Neural networks 

Qingde Li 

GPU based implicit surface rendering (QL1) 

Implicit surface has been a well known form for representing geometric shapes. Despite its various advantages over parametric and mesh-based shapes, it has not been used as popularly as parametric and mesh-based shapes in geometric shape modeling. This is mainly due to the fact that with conventional graphics acceleration hardware it is time consuming to render an implicitly represented shape. However, with the advent of programmable graphics hardware, this situation has been changed rapidly. In this research, we aim to develop techniques for rendering implicit surfaces implemented entirely as a set of shaders running on modern programmable GPUs. The tasks involved in this research will involve the comparison of different approaches for rendering implicit objects like implicit surface polygonizing based and the ray tracing based approaches. In addition, various techniques to accelerate the rendering process will also be investigated, such as the building of a hierarchical structure of implicit surfaces used for LOD rendering. The developed techniques in this research will be implemented in the later stage to develop an immersive sculpting system with the use of the motion tracking facilities housed in HIVE. 

GPU-accelerated Medical Data Visualization (QL2) 

The process of reconstructing 3D shapes of a human organ from data acquired from a medical information scanner is a time-consuming task due to the nature of complexity of the scanned data. Despite the ever increasing power of modern graphics acceleration hardware, real-time reconstruction and rendering of 3D shapes from the scanned data remains to be investigated. The focus of this research is to study how to implement 3D shape reconstruction and rendering algorithms completely on GPU by taking advantage of the storage and processing power and the programmability of modern graphics hardware. 

Yiannis I Papadopoulos 

Automated Safety Analysis and Intelligent Fault Diagnosis (YP1) 

Hull is currently pioneering the development of a new method that automates the safety and reliability analysis of computer-based systems, such as those used in aircraft, cars and various types of engineering plants. The method is known as Hierarchically Performed Hazard Origin and Propagation Studies (HiP-HOPS) and its contributions so far include fast algorithms for automatic synthesis of models that predict how systems can fail such as fault trees and Failure Modes and Effects Analyses (FMEAs) as well as radical extensions to Boolean logic that facilitate temporal analysis of sequences of faults in a system. This work has influenced int'l research in this area and received numerous awards which include distinction among best papers in SAFECOMP'99, SAFECOMP'02, COMPSAC'03 and a best paper award in INCOM'06. It has developed with support from the British government, the European Commission, Volvo Cars and Jaguar Landrover. Recently, Hull has been awarded a "Yorkshire Concept" fund for the development of a commercialisation strategy for dissemination and exploitation of this technology. Opportunities for doctoral work in this area include development of concepts for representation and re-use of component failure patterns, synthesis with emerging Architectural Description Languages, development of algorithms for real-time and on-line fault diagnosis using intelligent agents, and techniques for probabilistic evaluation of temporal safety analyses. 

Genetic Algorithms for Automatic Design of Computer-based Systems (YP2) 

The aim of this work is to provide support in the design of complex computer-based systems by enabling optimisation of initial designs with respect to criteria that include reliability, safety and cost. To achieve this aim, genetic algorithms have been developed to enable the automatic "evolution" of initial designs drafted by humans to improved equivalents that employ hardware & software fault tolerance to satisfy safety and reliability requirements with minimal costs. This work is highly innovative and leads both to theoretical contributions and useful industrial applications. Publications of initial results have received a distinction among best papers in COMPSAC'03, INCOM'04 and a nomination for the "Future of Automotive technology Award" in FISITA'04, the automotive world congress. Currently, this work develops in SAFEDOR - the largest project on technological safety ever funded by the EU - where Hull is the main technology provider in subprojects concerned with safety analysis and this type of design optimisation. PhD research projects in this area will extend current work by expanding the criteria of optimisation, via application of parallel genetic algorithms and by adaptation of this concept to engineering processes used in automotive and aerospace industries. 

Eur Ing Brian C Tompsett 

Rapid Evaluation of Electronic Digital Evidence (BCT1) 

The investigation of electronic devices to gather evidence for criminal prosecution is becoming increasingly important as such devices play a larger and larger role in our everyday lives. However, the range of devices and the volume of information they contain is also increasing, as is the criminals deviousness at obscuring the evidence from investigators. This has resulted in proposals to change the law to permit linger detention of suspects to allow more time for the digital investigations. 

Current digital forensic practise requires a manual examination of the entire contents of a digital device using the knowledge experience and expertise of the investigator. Pressure of time means that only lines of enquiry related to the suspected offence under consideration are followed. This research proposes some form of automated assessment of digital evidence sources, which can be deployed by non-experts in the field, such as law enforcement officers, and can permit rapid evaluation of the evidence available and an indicator of the types and nature of possible offences that could be linked to the evidence source. 

Applicability and Interpretation of Cyberprofiling in Real Environments (BCT2) 

Cyberprofiling is a project that has applied the techniques of Geographic and Criminal Profiling to internet crimes. It has developed methods for the gathering and relation of data relating to internet crimes from a large variety of data sources. 

One of the areas that remains to be investigated is the visualisation of such a large amount of data relating to a criminal investigation by unfamiliar with the technology involved. This project aims to investigate such data visualisation techniques as might be valuable in exploring internet crime related data. The project will also have to encompass data anonymisation techniques that will permit patterns and trends in crime data to be shown, but without detailed individual identities being compromised, and thus retaining data protection integrity. 

Bing Wang 

Structured Mark-up Languages and XML Database Design (BW1) 

Structured mark-up languages use special tags to define either layout or contents of documents. They provide a simple, very flexible text format to structurally construct different types of documents. Because of this advantage, they can easily meet those challenges of large-scale electronic publishing. Among these languages, XML is a widely accepted and used industrial standard mark-up language. Database technologies have proved their power to structurally maintain un-structured information. They are focusing on three major aspects: database query languages for XML data, data models for semi-structured data, and C# based XML database machine design. We aim to fully implement a true XML DBMS which can be used both as a storage for XML documents and as a powerful server suitable for web-based applications. 

Database Modelling Languages and Tools Design (BW2) 

Current data-modelling support tools are designed for specific purposes. This makes it difficult for users to employ existing tools to model and define their application environments using their own methods. In this research area, we aim to investigate existing modelling languages and define a new general-purpose data modelling language which allows users to create objects and define semantics among objects on the basis of the needs of applications. In particular, we will study the theoretical aspect of how to define universal object and relationship types which can be used to define real-world enterprise in an object-oriented way. We also investigate and implement the corresponding database interfaces for supporting this modelling language. 

James W Ward 

Improved Stereoscopic Viewing for Immersive Displays (JWW1) 

There have been vast improvements in 3D display technology in recent years, leading to a renewed interest in their use for scientific, engineering and entertainment applications. However, there remain a number of technical challenges to overcome before these displays can be wholly convincing. These include the accomodation-vergence conflict, shear distortion and incorrect scaling during head tracked stereo, and unlimited depth of field. 

Some of these result from fundamental limitations of the technology, while others are due to inadequate modelling of the human visual system. This project will focus on improving the accuracy and realism of a 3D display with head tracking (i.e. the head position of the viewer is considered when calculating the 3D projection). Through experimental work with 3D displays and an investigation of existing literature on human vision, an improved stereo projection model will be developed and evaluated in practice. 

There are a number of 3D displays available to support this project, including a large (5.3x2.4m) workwall, equipped with a Vicon optical tracking system. 

Helen Wright 

HW1-Novel interaction modalities for visual analytics 

Visual analytics encompasses the science and technology of using the human visual sense to search for patterns, links and dependencies in large, abstract data sets. In addition to understanding the status quo of a particular scenario, its aim is often also to make underpinning predictions that justify and quantify future strategy decisions. A key feature of visual analytics is the ability to interact rapidly and freely with the data, thereby utilising to the full the high-bandwidth channel that is our visual sense. Drawing on the state-of-the-art display and interaction facilities available in the Hull Immersive Visualization Environment (immersive stereoscopic visualization, gesture input and force-feedback rendering), this project will investigate the impact of these novel interaction modalities on visual analytics.