Human Dynamics Observatory
Yesterday, I was at the MIT — FBK Human Dynamics Observatory at FBK in Trento, Italy. It was organized by Bruno Lepri. The video of the full event is below:
- Based on a study of credit card transactions, researchers noticed similarities between people’s shopping behavior and animal’s foraging behavior, and found that the shopping “exploration” changed during periods of financial problems [plus one 2015 singh, bosky, peantland]
- From a communication network derived from online data, you can predict the wealth of people. nyu folks predicted the wealth of a person based on how many people that person was able to reach in 2 steps (friends-of-friend), with an R2 as high as 0.9
- Enigma project is quite cool and has been implemented in real-world settings. Enigma as a new “public commons for data”. The main idea is to bring the algoritms to the data, don’t share the data itself. It “is a decentralized cloud platform with guaranteed privacy. Private data is stored, shared and analyzed without ever being fully revealed to any party. Secure multi-party computation, empowered by the blockchain, is the magical technology behind it.” Initial paper here.
Nicola Segata @nsegata presented a brilliant projct — “The human microbiome”. It was all about the fascinating field of metagenomics:
Apparently, 600 species ride the subway with you! And we know that because of a project called “Metasub”. Researchers led by Prof Chris Mason (video) at Cornell University are trying to map the microbes in our cities at global scale!
Susan Athey of Stanford has done some interesting work on Machine Learning Methods for Causal Effects (pdf). At 4.07 in the video, you can see that she spoke about a natural experiment related to when Google News shut down in Spain. The question was: is Google News impacting publishers’ revenues and news consumption? For sure, yes. The researchers found that big news outlets compete with Google News, while small outlets are helped by the presence of Google News. And they quantified the extent to which it is so.
Finally, the last panel was on data-driven cities and government. Speakers included me (Bell Labs, Cambridge), Emmanuel Letouzé (Data-pop Alliance), Fosca Giannotti (CNR), Francesco Calabrese (Vodafone Italia).
I gave an overview of a lot of projects, including Smelly Maps, Chatty Maps, the inference of restaurant/cafe’ ambiance from the profile pictures of the customers (video), algorithmic regulation for the sharing economy (in the news), and the testing of Jane Jacobs’s ideas at the scale of entire cities. I also mentioned two upcoming projects: Cultural Capital 2.0, and Innovation Capital (which goes beyond the idea that innovation scales superlinearly with city size). More on those coming up soon on periscope (event invitation).
Rahul Nair of IBM presented works on data for humanitarian sector.
Good design humanises, bad design brutalises.
p.s. I missed a few things because of other meetings. A panel on data-driven health. The speakers included: Andrea Pezzi (Innovation Director Unipol-SAI), Andrea Cavalli (IIT), Nicola Segata (CIBIO), Cesare Furlanello (FBK), Ciro Cattuto (ISI Foundation). And the panel on data-driven finance had Cristiano Motto (Head of of Information Management Europe, Barclays Europe), Dino Pedreschi (University of Pisa), Espen Beer Prydz (World Bank), Raffaele Mauro(Endeavor Italia).