Differences

This shows you the differences between two versions of the page.

--- papers:tr-bottou-2012 [2012/09/12 08:13]
leonb
+++ papers:tr-bottou-2012 [2013/05/21 18:10]
leonb
@@ Line 2: / Line 2: @@
 //Abstract//:
-Using the search engine ad placement problem as an example, we explain the
+This work shows how to leverage causal inference to understand the behavior of complex learning systems interacting with their environment and predict the consequences of changes to the system. Such predictions allow both humans and algorithms to select changes that improve both the short-term and long-term performance of such systems. This work is illustrated by experiments carried out on the ad placement system associated with the Bing search engine.
-central role of causal inference for the design of learning systems interacting
-with their environments. Thanks to importance sampling techniques, data
-collected during randomized experiments gives precious cues to assist the
-designer of such learning systems and useful signals to drive learning
-algorithms. Thanks to a sharp distinction between the learning algorithms and
-the extraction of the signals that drive them, these methods can be tailored
-to causal models with different structures.  Thanks to mathematical
-foundations shared with physics, these signals can describe the response of
-the system when equilibrium conditions are reached.
 <box 99% green>
@@ Line 18: / Line 9: @@
 \\
 [[http://arxiv.org/abs/1209.2355|arXiv link]]
+<html>&nbsp;&nbsp;</html>
+Latest:
 [[http://leon.bottou.org/publications/djvu/tr-2012-09-12.djvu|tr-2012-09-12.djvu]]
 [[http://leon.bottou.org/publications/pdf/tr-2012-09-12.pdf|tr-2012-09-12.pdf]]

User Tools

Site Tools

Differences

Page Tools