Privacy Enhancing Technologies Challenge Prize

Innovate UK, part of UK Research and Innovation, will work with the Centre for Data Ethics and Innovation (part of the Department for Digital, Culture, Media and Sport) to run a Privacy Enhancing Technologies Challenge. This is part of an aligned programme with the US National Science Foundation, the White House Office of Science and Technology Policy, and the National Institute of Standards and Technology.

The aim of this Challenge is to accelerate the development and adoption of privacy-preserving federated learning approaches, and build trust in their adoption.

To achieve this, participants are asked to develop approaches that:

• leverage a combination of input and outputs privacy techniques
• deliver strong privacy guarantees against a set of common threats and privacy attacks
• develop privacy technologies that are capable of supporting machine learning tasks in one or two predefined use cases in the financial crime and public health sectors

The challenge is split into 3 phases:

• Phase 1: Your approach. You will develop a technical white paper which will describe your proposed approach
• Phase 2: Solution development. If successful in phase 1, you will be invited to develop your solution
• Phase 3: Testing. The top solutions will be tested by dedicated Red Teams

Up to £700,000 is available in funding across the three phases of the competition.

Applicants successful in Phase 1 and invited to Phase 2 will be able to apply for up to £50,000 to develop their solutions in Phase 2. Up to 10 awards of £10,000 will also be given tothe highest scoring projects from phase 1. The £10,000 can only be used to support with the growth of your business and the costs associated with this must be evidenced.

This application process represents the UK side of the challenge. US applicants must apply to the US competition. Each organisation can only apply once into either the UK or US competition. This includes if your organisation is linked in anyway.

Eligibility

Your organisation must be a UK registered:

• business of any size
• academic institution
• charity
• not for profit
• public sector organisation
• research and technology organisation (RTO)

Your organisation can lead on one proposal in each theme only. You can only receive funding for one successful proposal.

Scope

The aim of this competition is to develop innovative privacy-preserving solutions that address one or both of the specific challenge use cases in financial crime or public health.

You must develop privacy-preserving federated learning solutions that:

• use a combination of input and output privacy techniques
• demonstrate the ability to protect privacy against a set of defined attacks and threat models
• effectively accomplish a set of analytical or predictive tasks specified in the use case provided

You must determine the set of privacy technologies used in your solutions.

For example:

• any de-identification techniques
• differential privacy
• cryptographic techniques
• combinations that could be a part of the end-to-end solutions

We encourage projects that:

• display a high degree of novel innovation
• rigorously describe how their solution will provide guarantees of privacy appropriate to the use case
• consider how their solution, or a future version of it, could be applied in a production environment

Your project must address one or both of the pre-defined, high-impact use cases. You must specify the use case in your application, and can submit technical solutions for both use cases. The technical evaluation of these will be treated separately.

A full technical briefing is attached to each use case, providing details of the datasets, analytical tasks, and evaluation criteria that will be used during the challenge. The briefings will also outline the requirements for what should be included in the white paper.

Use case 1: Financial Crime Prevention

This use case is focused on enhancing cross-organisation and cross-border data access, supporting efforts to combat money laundering and other financial crime.

You will utilise synthetic datasets representing data held by the SWIFT payments network and datasets held by partner banks.

This is a high-impact use case for novel privacy enhancing technologies. Successful solutions will allow for effective detection of illegal financial activity while addressing the challenges arising between enabling sufficient access to data and successfully limiting the identifiability of innocent individuals and possibility of inference of their sensitive information from that data.

The scale of the problem is vast: the UN estimates that US$800-2000bn is laundered each year, representing 2-5% of global GDP.

A full technical briefing for this use case can be found here.

Use case 2: Pandemic response and forecasting

This use case is focused on enabling privacy-preserving access to health and mobility data in order to improve forecasting related to public health emergencies, and there by bolster response capabilities for future emergencies, including pandemics.

You will utilise synthetic datasets representing data held by the University of Virginia. This use case is an opportunity to prepare for future epidemics and public health emergencies.

A full technical briefing for this use case can be found here.

Briefing

An online briefing will be held on 26th July at 11.30am: click here to register to attend.

Challenge website

Click here to visit the dedicated website for the PETs Prize Challenge, where you can find potential UK and US collaborators and learn more about the technical requirements.