Meta-Interpretive Learning from textual data studentship

The main objective of this PhD project is to develop a new machine learning algorithm that uses a combination of meta-interpretive learning (MIL), a novel machine learning approach, and natural language processing (NLP) for automated text classification from small number of training examples.

Start date
1 October 2020
Duration
48 months
Application deadline
Funding information

This PhD studentship has an annual stipend of £20,000 plus full coverage of tuition fees, for the duration of the study.

Funding source
Coaching Principles - £57,870 and EIT Digital IVZW - £57,870
Supervised by

About

Machine learning, in particular Deep Neural Networks (DNNs), have demonstrated impressive results in many real-world problems such as image and text classification. However, existing learning algorithms require a large number of training examples, e.g. we need hundreds or thousands of examples to train DNNs for image or text classification. Nevertheless, in some applications we only have a small number of training examples. For example, for an automated software tool to be efficiently customised with new user requirements it should be able to learn from a small number of interactions and examples provided by the user.

This research project will be conducted in collaboration with Coaching Principles Ltd, the creator of GoalShaper, an Enterprise level goal setting solution that is the first to link user actions to company goals with real-time feedback. However, its capability is limited because it requires manual input and doesn’t integrate fully with company servers. The PhD student will work with the academic supervisor and the software developers at Coaching Principles to introduce a new machine learning module called Meta-Goal Learner (MGL), enhancing the next generations of GoalShaper, e.g. enabling GoalShaper 2.0, for Enterprise clients, to automatically classify individuals productivity tasks into pre-defined or new categories and make concrete connections to their overall goals. GoalShaper 2.0 will then be able to interactively learn the classification rules from a small number of examples provided by the user and authorised background knowledge from other sources (e.g. company databases, calendars, e-mails, project management software etc). This classification of tasks, and connections to goals, is currently done manually in the existing implementation of GoalShaper.

The Coaching Principles team is based in the UK and lead by several leading global technicians and includes Dr Marc Teerlink who was head strategist at IBM Watson and is now head of global AI at SAP.

Supervisors:

Dr Alireza Tamaddoni-Nezhad, Lecturer in Machine Learning and Computational Intelligence, Dept. of Computer Science, University of Surrey, 

Dr Marc Teerlink &  Mr Vijay Mitra (CTO), Coaching Principles Ltd

Related links
Department of Computer science EIT Digital Goalshaper

Applicants would be expected to engage with the EIT Digital Centre in London during the project.

Eligibility criteria

Candidate's specification:

Essential

  • BSc in Computer Science, Mathematics, Physics or related discipline (UK equivalent of 2:1 classification or above)
  • Keen interest in AI, Machine Learning and related research topics
  • Solid programming experience with at least one of the following programming languages: C/C++, Python, Java, Prolog or Haskell
  • Analytical and formal skills: knowledge of foundations of computer science and of relevant mathematical background.
  • Ability to think independently and self-learning capabilities
  • Strong verbal and written communication skills, both in plain English and scientific language for publication in relevant journals and presentation at conferences. Fluent written and verbal communication skills in English

Desirable

  • Master’s degree (UK equivalent of Merit classification or above)
  • Experience of applying Machine Learning and data analysis techniques and have relevant skills (e.g. data pre-processing and visualisation, setting up cross-validation, data labelling and version control, etc)
  • Experience of AI programming (e.g. Logic Programming & Prolog), Natural Language Processing (NLP), Rule-based Machine Learning (RBML), and Deep Learning (supervised & unsupervised) would be very desirable
  • Flexible, able to work collaboratively
  • A strong team player with good interpersonal skills able to build and sustain effective working relationships with the research group
  • Self-motivated researcher, with a hands-on approach, willing to develop their technical and analytical skills and contribute to the overall aims of the research project in innovative ways

IELTS requirements: 7 or above.

How to apply

Applications, consisting of a CV, a Personal Statement, and documents showing your academic track records, should be submitted to the University of Surrey via the Computer Science PhD programme page. 

Information on how to apply can be found by clicking on the ‘Apply Online’ button.


Application deadline

Contact details

Alireza Tamaddoni Nezhad
34 BB 02
Telephone: +44 (0)1483 682650
E-mail: a.tamaddoni-nezhad@surrey.ac.uk

Studentships at Surrey

We have a wide range of studentship opportunities available.