From awd at cs.cmu.edu  Wed Apr  3 15:17:07 2013
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Wed, 03 Apr 2013 15:17:07 -0400
Subject: [Research] "...guitar-like instruments made out of catfish skins..."
Message-ID: <515C8033.7060309@cs.cmu.edu>

What you learn at the Auton Lab can be applied in millions of ways:

http://www.businessinsider.com/globein-gives-a-new-way-to-shop-2013-4

Congrats Vlad!

Artur


From awd at cs.cmu.edu  Mon Apr  8 08:32:12 2013
From: awd at cs.cmu.edu (Artur Dubrawski)
Date: Mon, 08 Apr 2013 08:32:12 -0400
Subject: [Research] AutonLab meeting this Tuesday April 9th at noon in NSH
	1507
In-Reply-To: <020c01ce3161$ca61aa30$5f24fe90$@andrew.cmu.edu>
References: <020c01ce3161$ca61aa30$5f24fe90$@andrew.cmu.edu>
Message-ID: <5162B8CC.5060408@cs.cmu.edu>

Speaker: Dr. Kyle Miller

Topic: T-Cube Data Server makes large scale analyses of multivariate 
time series widely accessible and configurable. Come and learn how to 
best take advantage of that in your work.

Food: Will be provided.




From schneide at cs.cmu.edu  Mon Apr 22 15:08:49 2013
From: schneide at cs.cmu.edu (Jeff Schneider)
Date: Mon, 22 Apr 2013 15:08:49 -0400
Subject: [Research] Barnabas AISTATS practice talk tomorrow morning 10am
Message-ID: <51758AC1.2090101@cs.cmu.edu>

Hi Everyone,

Sorry for the late notice.  Tomorrow (Tue) morning at 10am Barnabas will give a 
practice talk for his AISTATS paper (abstract below).  We will do this in NSH 
1507.  Please come by and hear about the work on regressions from distributions 
to real-values.  This will also serve as a warm-up to a future brainstorming 
session from Junier on regressing distributions to distributions.

See you tomorrow morning!
Jeff.




Distribution-Free Distribution Regression

     `Distribution regression' refers to the situation where a response Y 
depends on a covariate P where P is a probability distribution. The model is 
Y=f(P) + mu where f is an unknown regression function and mu is a random error. 
Typically, we do not observe P directly, but rather, we observe a sample from P. 
In this paper we develop theory and methods for distribution-free versions of 
distribution regression. This means that we do not make distributional 
assumptions about the error term mu and covariate P. We prove that when the 
effective dimension is small enough (as measured by the doubling dimension), 
then the excess prediction risk converges to zero with a polynomial rate.