Hi everyone,
Below you can find an interesting talk that is on the application of network techniques on complex, dynamic and stochastic processes such as the spread of diseases within population. We had first seen network analysis in the OEP’s engagement with pipeline optimization problems. Then if I am not wrong this sort of thinking came up again with Lyle’s engagement with syndromic surveillance, and we had discussed both in New York and Berkeley how these techniques might have lead to a reproblematization and reproblematization of ‘population’ as an ontological being under a new and novel form different from its 19th century understanding (as we know from Foucault’s and his followers’ work on this period).
Perhaps one of the epistemic shifts that become possible with the application of techniques such as network analysis and monte carlo simulations is the ability to intervene in very calculated and precise ways as the abstract below suggests. Fuzzy and macro processes that have been the object of intervention in the 19th century and most of the 20th century are now unconcealed in novel ways with these techniques which allow the experts to examine the object in exact and precise ways by visually re-constructing the object of intervention in its exact detail. To my knowledge there is almost no work that pays attention to this aspect, and furthermore to me this seems to be one of the crucial points of the vital systems argument.
Â
Â
Date: Tue, 14 Oct 2008 09:23:56 -0400 (EDT)
From: Rocco A. Servedio <ras2105@columbia.edu>
To: theoryread@lists.cs.columbia.edu
Subject: Theory reading group: David Kempe talk Mon Oct 20, 2:30pm
Â
Please join us for a theory seminar on Monday, October 20 at 2:30pm in the
CS Conference Room. David Kempe of USC will speak about “Optimization
Problems in Social Networks”; details below.
Â
Please send me email if you would like to meet with David during his
visit.
Â
Hope to see you at the talk,
Â
Rocco
Â
=========================================================================
Â
Title:Â Optimization Problems in Social Networks
Â
Abstract: A social network – the graph of relationships and interactions
within a group of individuals – plays a fundamental role as a medium for
the spread of information, ideas, influence, or diseases among its
members. An idea or innovation will appear, and it can either die out
quickly or make significant inroads into the population. Similarly, an
infectious disease may either affect a large share of the population, or
be confined to a small fraction.
Â
The collective behavior of individuals and the spread of diseases in a
social network have a long history of study in sociology and epidemiology.
In this talk, we will investigate graph-theoretic optimization problems
relating to the spread of information or diseases. Specifically, we will
focus on two types of questions: influence maximization, wherein we seek
to identify influential individuals to start a cascade of an innovation to
maximize the expected number of eventual adopters; and infection
minimization, wherein we seek to remove nodes so as to keep a given
infected component small.
Â
We will present constant factor and bicriteria algorithms for versions of
these problems, and also touch on many open problems and issues regarding
competition among multiple innovators.
Â
(This talk represents joint work with Jon Kleinberg, Eva Tardos, Elliot
Anshelevich, Shishir Bharathi, Ara Hayrapetyan, Martin Pal, Mahyar Salek,
and Zoya Svitkina.)
Thanks for this Onur, very interesting. It does seem to map usefully on to the approach to ‘vital nodes’ we’ve seen in other domains, particularly electricity. Treating population as a network that can be optimized for norms, not of ‘health’ per se, but infection minimization. Also interesting, and perhaps contestable, is the mirroring of information and disease spread, by focusing on the object of ‘relationships’ and ‘interactions’. Influenza and influence of course have the same root. I’m not certain how much this conceptualization has been taken up in public health practice, although something like ring-vaccination strategies (in which you try to vaccinate people in rings around a node of infection rather than mass vaccination of a population) use clunkier version of the basic idea.
Hi Lyle,
I was posting an extensive comment to your comment, but somehow it did not go through and I lost it. As comment was pretty long, I am not sure if i can reproduce it. Yet, here is a brief version:
But just as a small note: I think you are absolutely right. The distinction between optimization and minimization/maximization is very interesting, and I think it serves a very interesting purpose for reconciling deeply contradictory and often mutually exclusive norms and normative rationalities. So, for instance the problem you are posing in your post maximization of the health of the population vs intervention and then vs the question of the autonomy of the population as a political ontological form within liberal thinking is in a way answered through the technique of optimization.
I think it would not bee too wild to imagine a case where an infectious disease epidemic breaks out in an isolated town in the wilderness, and you have to quarantine the town (an example F. referred to in STP). The 19th century response would be isolate the town. But with this technique you can isolate critical nodes without necessarily isolating the town. Probably in reality these two practices of population-security would be used in tandem with each other and respond to different pragmatic problems that the experts are having on the ground.
As to the application of these techniques. I am not sure how wide spread they are, but I know different funding agencies including NIH is extremely interested in studies deploying these techniques. One example is this paper: