Notice

Behavioural Analytics Competition Scoping Workshop

Published 2 August 2018

This Defence and Security Accelerator (DASA) competition seeks innovative solutions to how the Ministry of Defence (MOD) and partners across Defence and Security can unlock the potential to understand human behaviour from the estimated 2.5 quintillion bytes a day of user uploaded data using ‘behavioural analytics’; context specific insights into the ‘how’ and ‘why’ of individual, group and population behaviour, enabling predictions about how they are likely to act in the future.

The competition will see over £5 million of funding committed to innovation in this space, over a 26 month period, in a number of phases.

1. Aim

The aim of this workshop is to better define and focus the Behavioural Analytics competition scope based on the Defence and Security need, the latest research and development, and expertise within the community. This will support development of the full competition document planned for launch in October.

The event will be attended by military end users and practitioners, allowing Defence and Security professionals to share relevant current thinking and problem sets with potential suppliers from the wider external community including industry and academia.

2. How to attend

The workshop will take place at a venue in London on the 13 September 2018. If you would like the opportunity to help shape this upcoming competition, please register your interest..

3. Background

All Defence and Security activities are designed to have impact or influence on the behaviours of an individual, a group or community, a network or organisation, including on ourselves, and state or non-state actors (our allies and adversaries). Consequently, the audience, and the ability to influence that audience, is at the heart of everything that we do.

The vast amount of data generated by humans offers enormous potential to change how scientists can observe and understand human attitudes and behaviour. Whilst investment continues into novel data analytics and algorithms, machine learning and artificial intelligence, this challenge specifically focusses on the creators and users of this data.

We are interested in innovative contributions to how Defence and Security can develop and improve its capability in ‘behavioural analytics’

We are specifically interested in how we can:

1. improve understanding and measurement,
2. make confident and ethical predictions,
3. develop and deploy interventions, improving judgements and decisions, across Defence and Security.

In order to better articulate the particular challenges in Defence and Security within the three areas of Understanding, Prediction and Intervention, there will be 6 discussion topics, each initiated by the presentation of a Defence and Security scenario:

3.1 Understanding

• Measurement by proxy will look at how we can be confident that the traces of human behaviour in data link to basic psychological principles. It will explore how and why we need to invest in ways to compare proxy (indirect) and observed (direct) methods of measuring human behaviour from data.

• Opening the psychological window will look at how data provides a psychological window into the attitudes and behaviours of a broad spectrum of the world’s population. It will discuss how we need to identify what to let through this window and how we can address the complex technical storage, retrieval, analysis and verification of such large disparate data sets.

3.2 Prediction

• Finding predictors in a haystack will look at how such large quantities of data allows the identification of related factors but it does not assist in understanding what factors cause particular behaviours. It will discuss ways to validate predictors on multiple data sets to achieve sufficient levels of confidence and explore ways to employ different types of quasi-experimental designs.

• Avoiding dystopia will discuss the legal, ethical and moral challenges associated with continuous measurements of attitudes and behaviours, prediction and intervention. The need to invest in ways for humans to maintain control of data with models of regulation and supervision, detection of threat, hijack and misuse will be discussed.

3.3 Intervention

• Enhancements and overload will focus on human cognition, discussing how extant research suggests that humans learn incrementally from small amounts of data and have many different routes to understanding. It will discuss how we can incorporate cognition into models and also how we need to find ways to make voluminous data useful and understandable so that it can be acted upon in a timely manner by looking at design solutions and training techniques that focus on the human factors.

• The unblinking eye will bring together all the previous discussions to discuss how defence and security can develop an ‘unblinking eye’ approach to audiences, enabled by insight, evaluation and measurement. It will explore rapid filtering and analysis, rapid questioning with built-in analytical functions, visualisation and brain-computer interfaces