Developments in Statistical Learning Theory from a Perturbation Analysis Perspective

Lead Research Organisation: Imperial College London
Department Name: Mathematics

Abstract

Statistical Learning Theory aims to provide guarantees on the sample complexity and accuracy of algorithms. The theory for independent identically distributed samples is well established but less developed in comparison when the restrictive i.i.d. assumption is lifted.

We will extend results in Statistical Learning Theory for Stochastic Processes using in particular the algorithmic stability and algorithmic robustness frameworks.

The directions of research are the extension of existing results for mixing processes to the more general alpha-mixing case, development of further results for unbounded loss functions as well as results for data-dependent hypothesis sets (in the context of non i.i.d. data).

Planned Impact

The current DTC in Financial Computing is acknowledged by the Department of Business Innovation & Skills as having had a major impact on our financial industry partners and on our academic partners. We and our Industry partners are also central to the forthcoming investments in Big Data from EPSRC and ESRC (e.g. Business Datasafe). The Pathways to Impact Attachment provides a comprehensive description of impact.

INDUSTRY
The Centre will continue to actively promote the placement of PhD and Masters students.
* PhD student placements- many of the Banks now have established formal PhD Internship programmes, in part due to the current DTC.
* Masters student placements - the Centre is actively involved in placing UCL Masters students in our partner companies, in part to address the shortage of PhD students.
* Utilising Industry Lecturers on Courses - UCL, LSE and Imperial College increasingly use industry professionals to enhance their finance-related courses and is popular with students.

REGULATORS AND GOVERNMENT
Collaboration with the financial Regulators and Government is a priority for the Centre (See letters of support from BoE, PRA and FCA).
*Financial regulators - currently we have 2 PhD students in the Bank of England, and we will be seeking to place more. In addition, with the new CDT we will be seeking to establish one or more major collaborative projects with the Regulators.
* Government - we have a PhD student collaborating with the Cabinet Office on their major cross-government benefit fraud programme, and we also support the MiData initiative.

SOCIETAL
The existing DTC and the proposed CDT's societal contribution focusses on supporting entrepreneurship.
* Entrepreneurship & start-ups - we will continue to encourage and support our PhD students in launching their own start-up.
* Professional (part-time) PhD students - we will continue to recruit young professionals who wish to pursue part-time PhDs.

ECONOMIC
* UK Services Sector - already the existing Centre is building research collaboration with companies such as Tesco, Sainsbury's, Alliance Boots, BUPA, Unilever, P&G, dunnhumby and SAS.
* Other sectors - we are promoting, but not funding, analytics in other sectors, a notable example being Sports Analytics.

Publications

10 25 50