Department of Informatics

Funding and Scholarships

School of Engineering and Informatics, PhD Scholarships (2016)

When applying for a PhD Scholarship applicants may choose to apply with their own research proposal or apply under a topic specific project proposed by a member of the department's faculty. Projects proposed by applicants must fit within the areas of research in the School; applicants with interests that span both Departments within the School are particularly welcome.

Research groups in Engineering, Design, and Informatics

The research groups in the Department of Engineering and Design are:

  • Dynamics, Control and Vehicle Research Group
  • Industrial Informatics and Signal Processing Research Group
  • Sensor Technology Research Centre
  • Thermo-Fluid Mechanics Research Centre

The research groups in the Department of Informatics are:

  • Cognitive and Language Processing Group (Data Science)
  • Evolutionary and Adaptive Systems
  • Foundations of Software Systems
  • Creative Technology (formerly Interactive Systems)

For further details see the research web pages for the respective Departments:

http://www.sussex.ac.uk/engineering/research

http://www.sussex.ac.uk/informatics/research

Type of award and award amount

The amount of the award can vary depending on who funds the award:

School funded Scholarships normally include a three year stipend at a standard rate (currently £14,057 per annum) and, in addition, fees as follows: (a) for Home/EU applicants, full fees; (b) overseas applicants, a contribution of up to £12,000 towards overseas fees, depending on qualifications.

EPSRC funded Scholarships normally include a three and a half year stipend at a standard rate (currently £14,057 per annum) plus a fee waiver to the UK/EU amount for 3.5 years. Due to funding restrictions, the EPSRC scholarship is open to UK and EU resident students only. Not all EU students may be eligible. Before applying, EU students should check the EPSRC eligibility criteria here: https://www.epsrc.ac.uk/skills/students/help/eligibility/

For further information on Univesrity fees visit: http://www.sussex.ac.uk/study/money/fees/pg2016

Eligibility

Applicants should have at least a good 2:1 bachelors degree (or equivalent) relevant to their area of study.

For further information on entry requirements visit:

http://www.sussex.ac.uk/study/pg/applying/2016entry/qualifications

Application procedure

Applicants must make an online application for a PhD place and state clearly on the form that they wish to apply for a School Scholarship:

http://www.sussex.ac.uk/study/pg/2016/research

Applicants for topic-specific Scholarship projects (see projects below) must include a file with a statement that explains their particular interest, knowledge and skills in relation to the chosen project. The filename should follow the format 'surname_personal_statement'.

Applicants submitting their own Scholarship project proposal should read the advice about finding a potential supervisor and how to write a proposal on these web pages:

http://www.sussex.ac.uk/informatics/pgstudy/doctoral/applying

The closing date for applications has been extended to: 18/01/2016

Contact details 

For general information about the Department visit:

http://www.sussex.ac.uk/informatics/

Or for enquires via email contact: phd@informatics.sussex.ac.uk

Applying with your own research proposal

If you choose to apply with your own research proposal visit our applying webpages which cover how to write your research proposal and how to identify, and contact, potential supervisors. When you are ready to apply follow the procedure as detailed above.

Applying for a topic specfic project

You will find summaries of the topic specific projects listed below along with contact details for the project supervisor so that you may request full details of the project, or discuss the project further. When you are ready to apply follow the procedure as detailed above.

You may only apply for one of these projects:

Foundations of Meta-Programming
Project Supervisor - Martin Berger

Summary:

Meta-programming (MP) is the ability of programming languages to treat programs as data. MP enables the generation of programs by other programs, either at compile-time or at run-time. A simple example of MP is:

   printf ( "printf ( \"Hello World!\" );" );

which outputs a C program that, when executed, prints "Hello World!".

Many programming languages, going back at least as far as Lisp, have explicit MP features that go beyond using strings to represent programs. These features can be classified in various ways such as: generative (program creation), intensional (program analysis), compile-time (happening while programs are compiled), run-time (taking place as part of program execution), heterogeneous (where the system generating or analysing the program is different from the system being generated or analysed), homogeneous (where the systems involved are the same), and lexical (working on simple strings) or syntactical (working on abstract syntax trees). Compilers use MP to compile programs; web system languages such as PHP use MP to produce web pages containing Javascript; Javascript (in common with some other languages) performs MP by dynamically generating strings and then executing them using its eval function. MP is often used for embedding domain-specific languages (DSLs) in a host language. In short, MP is a mainstream activity.

Perhaps surprisingly, given its long history, there is a near complete absence of a sound theoretical understanding of MP in general and explicit language support for MP in particular. (The only exception is existing work on "hygiene" such as [1].) Instead, MP's semantics have largely been defined by implementations, often created by individuals unaware of other approaches to MP. This has several negative consequences.

  • Terminological confusion, making communication about MP difficult.
  • Implementors reinventing the wheel, often badly.
  • Considerable confusion in the community about the relationship between different approaches to MP.
  • Near complete lack of theories and tools for the specification and verification of meta-programs.

This is an undesirable situation given the prominence of MP in mainstream computing. The purpose of the proposed research is to put MP on a firm basis. This involves several distinct but related tasks:

  • Providing a firm and convenient formal basis for MP in form of a foundational calculus that explains all key MP features.
  • Providing an algorithm that extends any programming language with MP constructs in a uniform and mechanical way.
  • Building program logics that can specify and verify meta-programs, based on [2].

The candidate. The candidate should have all or most of the following skills.

  • Familiarity with modern programming languages such as Haskell or Scala, Ocaml, F# or Rust.
  • Familiarity with MP.
  • Mathematical skills, including familiarity with formal proof, as for example found in [3].

Familiarity with proof assistants like Coq, Agda or Isabelle is desirable but not required. Candidates with a pure maths background would also be suitable.

For more information, please contact Martin Berger.


Literature

  1. M. D. Adams. Towards the essence of hygiene.
  2. M. Berger, L. Tratt, Program Logics for Homogeneous Meta-Programming.
  3. G. Winskel. The Formal Semantics of Programming Languages: An Introduction.  
Machine Learning-Driven Generation of Congestion Control and Flow Scheduling Algorithms for Improving Data Centre Performance
Project Supervisor - George Parisis

Second Supervisor: Dr Novi Quadrianto

Funding Note: This project is EPSRC funded. Please see the above 'Type of award and award amount' section for further details.

Summary:

Data centres consist of racks of commodity servers interconnected by multiple, high bandwidth links. The Transmission Control Protocol (TCP), the de-facto protocol for reliably transmitting data, is not appropriate for data centres; it is single-path and its congestion control algorithm, which regulates senders so that network congestion is minimised, while resources are utilised efficiently and fairly, follows a one-size-fits-all approach. On the other hand, data centres support high-speed, low-latency links and multiple paths among servers as well as applications with diverse data transport requirements and network workloads which constantly change.

This project aims at utilising machine learning techniques, such Bayesian optimisation and Gaussian processes, the power of parallelisation offered by modern Graphics Processing Units (GPUs), and Software Defined Networking (SDN) to automate the process of controlling congestion and scheduling flows so that data centre network resources are utilised efficiently even under dynamically changing network traffic.

Mind the body: Interoceptive inference and the neural basis of emotion and self
Project Supervisor: Prof Anil Seth

Second Supervisor: Prof Hugo Critchley

Summary:

Emotional feeling states are an essential part of what it is to be conscious self. A long tradition associates emotions with perceptions of the internal state of the body (interoception). Recently, this this tradition has been extended by proposing that emotions emerge from actively generated topdown predictions of the causes of changes in internal physiological state (Seth, 2013, Trends Cogn Sci, 17:565-573). The challenge now, and the focus of this Ph.D project, is to test experimentally this process of ‘interoceptive inference’. Finding evidence for, or against, this idea will have a large impact in how we understand emotion and selfhood, in health and in disease. This exciting project will combine advanced virtual reality methods with psychophysics, psychophysiology and neuroimaging to look for evidence for ‘interoceptive predictions’ in the brain, and to explore how such predictions shape experiences of emotion and self. The project will also bring significant opportunities to translate experimental paradigms to clinical populations (e.g., anxiety, psychosis, autism) which are already being studied within the Sackler Centre. The project will suit a student with a strong background in quantitative aspects of psychology, cognitive science, or neuroscience, who has a deep interest in consciousness, emotion, and self.

Characterizing functional connectivity in the brain, in health and in disease
Project Supervisor: Prof Anil Seth

Second Supervisor: Dr Lionel Barnett

Summary:

A key challenge in neuroscience is to understand how different parts of the brain communicate with each other, and how the brain’s flexible functional networks underlie and account for cognition, behavior, and consciousness. Research at the Sackler Centre has, for many years, pioneered the development analysis methods for identifying ‘causal’ networks based on neuroimaging data. We have now reached the exciting point where different prominent methods are converging within a coherent theoretical framework, and multimodal neuroimaging data are becoming available that allow the rigorous application of these methods. This exciting Ph.D project will involve development and application of new ‘causal’ analysis methods based on state space and autoregressive modeling, validation of these methods in detailed neuronal simulations, and application to neural data collected both in-house and via a rich networks of collaborators. These applications will shed new light on the brain basis of conscious perception in health and in psychiatric and neurological disease. The project will suit a student with a strong background in mathematics, statistics, physics, and/or computational neuroscience. Reference: Seth, A.K., Barrett, A.B., Barnett, L.C. (2015) J Neurosci 35: 3293-3297.

Verifying Software-Defined Networks
Project Supervisor: Dr Bernhard Reus

Second Supervisor: Dr George Parisis

Summary:

Modern computer networks are very complex, consisting of hundreds or thousands of network devices, such as switches, routers, firewalls and middleboxes1, which are manufactured by different vendors and support diverse functionality. Configuring, managing and operating modern networks are difficult and error-prone tasks, which require a tremendous effort from the network operator to ensure correctness of network-wide properties (e.g. absence of loops), security (e.g. network flows of different customers are fully isolated) and fault tolerance (e.g. spare network paths are enabled upon failures). Software-Defined Networking (SDN) is a new paradigm for operating and managing computer network. With OpenFlow, a standardised instantiation of SDN, a network operator can program event-driven network functionality in a logically centralised software entity to control the whole network, thus departing from a model where only network equipment manufacturers could develop software that ran on their own devices. Given the complexity and asynchronous nature of computer networks, controller code may easily be buggy, which can bring large networks down causing significant cost to the network provider. This project is about utilising cutting edge software verification techniques (and hybrids of these) to design and implement a framework for verifying safety properties of Software-Defined Networks. It aims at extending already existing methods to cover more controllers and properties, using synergies of various verification techniques.

Imperative Programs out of Proofs
Project Supervisor: Dr Bernhard Reus
Summary:

Program extraction is a powerful method of generating a program and a certificate of its correctness from a single constructive proof that states the existence of a solution to the problem specification at hand. It constitutes one of the most amazing properties of constructive logic and is based on the proofs-as-programs paradigm (also known as Curry-Howard isomorphism) and on realizability semantics. Traditionally, extraction has been carried out in various formulations of constructive mathematics using terms of computable functionals or partial combinatory algebras, which can be naturally interpreted as functional programs. Therefore, program extraction has long been associated with functional programs and many tools are available that can extract functional programs from proofs (Coq, Minlog, Agda, Epigram, NuPRL, Isabelle). Relatively little research and tool development, however, addresses the problem of extracting imperative programs from proof, although the vast majority of programs is written in an imperative language. Imperative programs are renowned to be difficult to verify due to dynamic memory and pointer aliasing. It is therefore desirable to extend the program extraction paradigm to the realm of imperative programs (with dynamic memory), which is the aim of the proposed project.

Deep Convolutional Neural Networks for Activity Detection in a Network of Surveillance Cameras
Project Supervisor: Dr Novi Quadrianto

Second Supervisor: Dr Viktoriia Sharmanska

Summary:

Action detection in videos means action recognition plus localization of the actors in spatial and temporal dimensions. Localizing becomes a key component in the “complex real-life scenario” involving multiple people in actions. Action detection is typically solved by recognition on candidate regions within the frame in contrast to recognition on the whole region of the frame. The success of detection methods depends on striking a balance between precision and recall of informative regions localizing the actors. We propose to solve the activity detection in surveillance videos from two perspectives, coarse and fine-grained. In the coarse perspective, we take a human-centric approach and generate candidate regions for each actor separately. We will use the most recent available models for human detection followed by the action detection models for each actor. In the fine-grained perspective, we will localize the action in more detail and detect characteristic body poses for each actor in action. As an illustration, a sneezing action can be visually seen as movement of the upper body, taking a deep breath, closing the eyes while possibly covering the mouth with a hand, and similarly with a typical coughing action. The temporal pose configuration of an upper body and hands gives a meaningful visual cue for a sneezing or coughing action. Discriminating two different actions, such as sneezing and hand waving, would be, in essence, to differentiate two pose configurations.

Insect-inspired Robotic Navigation
Project Supervisor: Dr Andrew Philippides

Second Supervisor: Dr Paul Graham

Summary:

Navigation is a vital task for animals and autonomous robots. Ants are particularly impressive navigators and, being practical to study and ‘robotic’ in their behaviour, provide excellent inspiration for biomimetic engineers designing computationally simple (and so, low-power and low-weight) navigation algorithms. We have developed a new ant-inspired method for route navigation with robust performance in complex but simulated environments. This PhD is an exciting interdisciplinary opportunity to develop such navigation algorithms for real world applications including: terrestrial vehicles for space exploration and agriculture; flying robots in GPS-denied environments; human guidance through wearable sensors. The first application area: autonomous navigation in terrestrial vehicles is supported by collaborations with RAL Space and Harper Adams University who develop autonomous vehicles for space exploration and agricultural applications, respectively.

Development of True3D displays using Non-Solid Diffusers
Project Supervisor: Dr Diego Martinez Plasencia

Funding Note: This project is EPSRC funded. Please see the above 'Type of award and award amount' section for further details.

Summary:

Recent market studies foresee VR will become mainstream, reaching a $7 billion market from 2018 and $62 billion by 2025. However, users are still forced to carry a head-mounted display, becoming unaware of people around them (e.g. colleagues), and encumbering use of other technologies or instruments.

Other solutions have similar limitations. Planar displays only allow hands and 3D content to be simultaneously visible in close proximity to the display. Bandwidth requirements limit holographic technologies to small sizes, limited resolution or non real-time behaviour.

In practice, either technologies only allow limited access to the benefits of 3D interaction (e.g. no direct manipulation, fixed positions and form factors) or they require aids (e.g. glasses, gloves) that make it cumbersome to interact with the world around us (e.g. typing, using a smartphone, watching a LCD screen).

Our main goal is to 1) develop systems that allow uninstrumented users to directly see and interact with 3D content; 2) identify interaction principles to support users around such systems. This main goal is enabled by our previous research on shape-changing, 3D displays using NSDs (Non-Solid Diffusers, i.e. made of fog, mist, dust).