National Cyber Security Centre “Security for Artificial Intelligence”: Call for Proposals

483198795

Closing date: Friday 2nd August 2019, 16:00

Summary

The Nation Cyber Security Centre (NCSC) is inviting proposal from academic researchers for research into the topic of advancing the understanding of the security challenges that face Artificial Intelligence (AI).

Research Challenges

The research proposals should clearly address one aspect of these principle challenge areas.

  • Assessing the behaviour of intelligent systems, understanding how to detect degradation in behaviour due to malicious activity. Identifying effective mechanisms for correcting or compensating for malicious behaviour.
  • Sourcing and handling of training data. An understanding of how data is gathered and utilised in training and operating intelligent systems, Highlighting the problems and risks that can occur when data provenance can’t be guaranteed, and guidelines on managing, mitigating and remediating these risks.
  • Best practices for building secure intelligent systems. Understanding which algorithms and decision models are most suitable for a range problem classes and use cases, and guidance into selection criteria. This additionally considers requirements for the design of user interfaces and system architecture for providing secure AI.
  • Uncertainty in outputs, a view to understanding how the outputs of intelligent systems can be interpreted, and how to make decisions given the uncertainty in these outputs.
  • Creating robust AI for security. Looking at different approaches for applying artificial intelligence techniques to security, with an emphasis on how the design decisions support the security of the end product.
  • Robustness of algorithms to adversarial attacks. Defining the principles and techniques that are needed to design algorithms such that they are robust against malicious adversarial actors.

The importance and impact of secure artificial intelligence is wider than just those of the security application.  Any context where there is a viable need to secure the intelligent system would be considered a valid use case for this research.

Available Funding

Funding will be available for 2 to 3 short projects over the duration of 3 months.  Each application can be for a maximum of £30,000.  The proposal should be defined such that the research can be completed by the 31st March 2020.

The funding and contract will be under the NCSC’s standard terms and conditions: a draft copy of the contract can be made available on request.  The research will be funded at Full Economic Cost.  Budgets for attendance to publicise and disseminate the work should be included within the research proposal.  Funding should be allocated to allow for travel to the NCSC office in London for a workshop in March 2020. 

The funders are committed to full and open publication of the research outputs in line with academic practices.

The NCSC believes that this topic is of significance across all applications.  We will be campaigning for more attention to be given to the topic at a national scale, and seeking additional funding for research from both government and industry partners.

How to Apply

Applications should be sent to Andrew F, Senior Security Researcher via email: Andrew.f2@NCSC.gov.uk.  We must receive your application by 1600 on Friday 2nd August 2019

To find out more, please click here.