• Home
• News & events
• Undergraduate
• Postgraduate
• PhD
  • Why EECS
  • Facilities
  • Research
  • Entry requirements
  • How to apply
    • Funding
    • Scholarships
  • Contact us
• Research
• For schools
• People
• About us
• Vacancies
• Brochures

In music performance studies, prosody is the musician-specific timing, stress, and sometimes intonation added when interpreting a notated score. Mid- to high-level music prosodic gestures, for example tempo trajectories, often invoke parallels in the physical world, such as a damped oscillator. This project seeks to identify and mathematically model such gestures. The mathematical descriptors will form the basis for a vocabulary of prosodic gestures for music.

In recent years, honeybees have been shown to display complex cognition, including phenomena like counting, attention, categorisation and learning the concept of ‘same’ and ‘different'. While bees might therefore be a model to understand how cognitive complexity is mediated with miniature brains , we still do not understand how these feats are generated in neural-computational terms. He we focus especially on concept learning.  In this interdisciplinary project the student will develop simulations of bee behaviour in experiments conducted in the Chittka and Fernando labs. These experiments will test 5 hypothetical mechanisms by which the same/different concepts could be learned. Further experiments will then be carried out to discriminate between these increasingly sophisticated hypotheses. This is a unique opportunity to apply sophisticated machine learning techniques to computational neuroscience and animal ethology.

Network Coding is a new type of routing protocol that originally was suggested for satellite communication, but where Microsoft later used the ideas in the file distribution system Avalanche. The project will investigate routing protocols that combine standard protocols with "lite" versions of Network Coding.

When a small group of musicians negotiate in performance (i.e. real-time) the shaping and execution of a collective interpretation, the communication is non-verbal; some of the cues can be embedded in the musical prosody, and some demonstrated through gestures. This project aims to capture, analyze, quantify, and model the cues necessary for effective and engaging ensemble performance, by studying both co-located as well as distributed (over the Internet) ensembles.

Cooperative networking allows multiple low-power terminals to create a virtual multiple-input multiple-output (MIMO) network achieving a distributed form of spatial diversity. In the presence of factors including randomly located nodes, irregularly deployed relays, and uncertain transmission range, our fundamental research questions are: “How to achieve optimal packet-level and symbol-level measurements in a cooperative network?” and “What is the impact on connectivity and delay in such a cooperative network?”.

This project will research modeling techniques and reasoning algorithms that can autonomically manage services deployed on a cloud platform. The work will correlate high-level SLA requirements with low-level management information, invoking platform management actions that can optimize the use of resources.

We will develop a framework for measuring & characterising the users' Internet traffic & digital TV viewing patterns at the home gateway, while preserving the individuals' privacy. This will enable statistical aggregation & profiling of users for provision of more appealing content (digital media, adverts or suitable deals).

A performer can take decades to learn a musical instrument. This studentship will focus on creating instruments that learn the capabilities and artistic preferences of the individual performer, with a particular focus on the relationship between physical gesture and sound production. The successful candidate will develop intelligent gesture-sound mapping strategies, which dynamically update based on feedback from the performer. User studies with professional and amateur musicians will be integral to all stages of the project, and the successful candidate will take a leading role in designing and conducting these studies. The project aims to make performance more accessible to beginning musicians while enabling new modes of expression for experts.

The project’s goal is to construct and evaluate computational models of human temporal expectation. It involves developing probabilistic models of temporal prediction, taking representational account of rhythm and metre. The models and parameters are optimised to maximise prediction performance and compared to human temporal expectations in empirical studies of listeners.

Metamaterials have led to a paradigm shift in the approach to engineering device design. However, it is now accepted that passive metamaterials suffer from unacceptable loss and narrow bandwidth. We have shown active inclusions significantly improve performance. We will further investigate theoretical limitations while continuing optimisation of electronics-based inclusions.

This study is designed to take advantage of a uniquely powerful dataset from 835 ecosystems across the UK created and managed by the Project Partners – one of the largest and most comprehensive of its type in the world – to develop novel approaches to biomonitoring the environment (thus aligning the project closely with several of NERC’s strategic funding priorities). We aim to add a completely new dimension to biomonitoring programmes by developing a novel network-approach within the existing national scheme (known as RIVPACS) that is used throughout the UK and has been exported to many other parts of the world. We believe this research proposal will produce both novel and profound insights into the structure and dynamics of ecological networks, one of the most rapidly growing fields in biology.

Ubiquitous access to digital music, and the hours of practice required to master new pieces, has led to a decline in amateur instrument playing. Cognisance of music structure can facilitate planning and sequence production, and enhance music making pleasure. Machine intelligence can help diagnose areas of difficulty and offer targeted constructive assistance. The studentship will propose/evaluate score-based visualisations of music structure and gestures that accelerate intermediate piano piece mastery for young and adult learners. Candidates should be proficient at programming, experienced with user interface design, have some background in statistics, and possess at least amateur-level piano playing ability.

Healthcare ICT systems can jeopardize patient health if misused. This PhD’s novel contribution will be to apply advanced verification techniques, more normally used in areas such as aviation software and chip design, to healthcare ICT so as to improve patient safety. In the UK about 10,000 adverse events are reported per year due to use error of healthcare ICT systems. Many are due to poor human-computer system design, either of ICT or of the socio-technical system in which ICT and users are embedded. For example, a nurse entering the wrong rate into an IV infusion pump might be the nurse’s fault but could also be poor pump design, poorly designed processes, wrong information from the pharmacy or wrong patient identification. The problem is to design resilient systems that prevent such systematic human error. This PhD will develop a systematic tool-supported methodology for uncovering critical deficiencies in the design of healthcare ICT and socio-technical systems. The innovation of the proposed approach is that models used in the analysis will take account of the information resources required by clinicians in performing an action: information that could be provided by the device being analysed or the physical environment in which the device is used. This “distributed cognition” perspective will be used to analyse how information resources constrain the activities users carry out.

Learning to identify object categories through pictures is an important area of research in Computer Vision. Most current systems aim to recognise hundreds and thousands of visual classes under changes in pose, lighting condition, and subject to occlusion. Yet the vast majority of this work tacitly assumes images in a single domain -- the photographic. Similar object class depicted in a different domain, such as painting and drawing, cannot be recognised. This project aims to bridge the semantic gap across visual domains by investigating a novel means of image representations that are independent across depictive styles and more consistent with human cognitive expectations. The development of such descriptions will have considerable impact due to the broad range of potential new applications . However, their realisation is challenging with none of the current state-of-the-art techniques shown to be effective. For instance, while “bag-of-words” models are extremely good at classifying objects and has led to many successful applications such as Google Goggles, they are often biased towards finding identical images of objects with little or no visual variations. This is because features such as SIFT only capture detailed local information of objects, but not their global structures and appearances. This project will investigate different ways of combining local object features and holistic structures, yielding descriptions that encapsulate global properties of objects and are transferrable cross different visual domains.

Research suggests that personality traits such as extraversion, agreeableness, and openness to experience, are tightly coupled with human abilities and behaviour encountered in daily lives: success in interpersonal tasks, academic ability, job performance, etc. Moreover, the problem of assessing people's personality is very important for multiple research and business domains such as computer-mediated staff assessment and training, personality profiling for personal wellness technologies, and enhancing human-computer interaction. Current personality assessments procedures are usually based on paper-and-pen tests (or their online counterparts) that require people to answer questions about their typical behaviour. However, these often contain transparent questions that people can figure out what is meant to be measured. Therefore, this project will focus on (1) developing a set of audio-visual tools that can analyse human personality traits from multiple nonverbal cues and channels (e.g., non-verbal behaviour expressed through facial, bodily and vocal cues) and (2) providing automatic personality assessment in terms of extraversion, agreeableness, openness to experience, neuroticism, and conscientiousness.

A general theory of sequential games has recently been developed (Proc. Royal Soc. A, 467:1519-1545, 2011) which extends Nash's original work on equilibrium strategies in n-player games. The work so far developed has discovered deep connections between constructions in Game Theory, Fixed Point Theory, Algorithms, and Proof Theory, via a generalisation of the technique of Backward Induction. The proposed project aims to (1) extend the theory to include simultaneous games and mixed strategies, i.e. players choose a probability distribution over their given sets of strategies , (2) understand the impact on the current state-of-the-art in Game Theory, and (3) investigate how simultaneous games could be understood from a logical (proof-theoretic) point-of-view, and what implications this has to proof theory.