People Watching with Machine Learning

Data Science
Location: Room 1-6 Level: Intermediate
Average rating: ****.
(4.33, 3 ratings)

Observing how other human beings interact is so interesting that we have a name for when we do it recreationally, we call it “people-watching.” Evolution has equipped us both with a desire to people-watch and with the tools we need to do it. In social situations, we can spot patterns such as groups forming and dispersing fairly easily, but it’s harder to describe that observation process logically. If we could do that, we could make machines people-watch for us.

Modern smart phone platforms come with a growing range of sensors. They also have a (near-) ubiquitous data connection, and the ability to report user positioning via multiple methods. They are almost
invariably carried everywhere on their owner’s person. Given all that, it’s now fairly easy to build large, distributed datasets from people’s smart phones — data that when collated together has a vast amount of information regarding the social graph.

We’ve developed algorithms incorporating machine learning techniques that, combined with these large spatio-temporal datasets, enable us to automatically characterize groups of people from the spatially coherent behaviour of the individuals that form them, as well as to look for other forms of interactions between people and distinguish between these different types of interactions.

These algorithms are going to give machines access to our social interactions in ways that weren’t possible before. While humans find it harder to spot social groups in large crowds because of the amount of data involved, additional data makes group identification algorithmically easier. Given enough data, machines might become better at finding patterns by people-watching than we are ourselves, and as a result give us novel insights into our own social interactions.

Photo of Alasdair Allan

Alasdair Allan

Babilim Light Industries

Alasdair Allan is a director at Babilim Light Industries and a scientist, author, hacker, maker, and journalist. An expert on the internet of things and sensor systems, he’s famous for hacking hotel radios, deploying mesh networked sensors through the Moscone Center during Google I/O, and for being behind one of the first big mobile privacy scandals when, back in 2011, he revealed that Apple’s iPhone was tracking user location constantly. He’s written eight books and writes regularly for Hackster.io, Hackaday, and other outlets. A former astronomer, he also built a peer-to-peer autonomous telescope network that detected what was, at the time, the most distant object ever discovered.

Zena Wood

University of Exeter

Zena Wood is employed as a lecturer in Computer Science at the University of Exeter, she works in the field of Applied Ontology and spatiotemporal reasoning. Entitled “Detecting and Identifying Collective Phenomena within Movement Data,” her PhD focused on identifying what is meant by the term collective, the different types of collective that exist and developing a method that allowed the identification of such phenomena within large spatiotemporal datasets.

Zena has continued to work within these fields and is currently developing similar techniques that can be used to study human and animal behaviour. In addition to her roles as lecturer, Zena also coordinates the eskills’s endorsed undergraduate and postgraduate IT Management for Business degrees and Computer Science outreach at Exeter.

Sponsors

Sponsorship Opportunities

For information on exhibition and sponsorship opportunities, contact Susan Stewart at sstewart@oreilly.com or +1 (707) 827-7148

Media Partner Opportunities

For information on trade opportunities contact Kathy Yu at mediapartners
@oreilly.com

Press and Media

For media-related inquiries, contact Maureen Jennings at maureen@oreilly.com

Contact Us

View a complete list of Strata contacts.