<?xml version="1.0" encoding="UTF-8" ?>
<feed xmlns="http://www.w3.org/2005/Atom">
<id>http://research.google.com/pubs/atom.xml</id>
<title type="text">Recent Google Publications (Atom)</title>
<updated>2012-05-14T16:40:15Z</updated>
<link href="http://research.google.com/pubs/atom.xml" rel="self"/>
<link type="text/html" rel="alternate" href="http://research.google.com/pubs/papers.html"/>
<entry>
<title><![CDATA[Modelling the Distortion Produced by Cochlear Compression]]></title>
<updated>2012-05-14T16:39:04Z</updated>
<id>urn:googlelabs:37559</id>
<summary type="html"><![CDATA[Modelling the Distortion Produced by Cochlear Compression, Roy D. Patterson, Timothy Ives, Thomas C. Walters, Richard F. Lyon  Links:  [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=Modelling+the+Distortion+Produced+by+Cochlear+Compression+Patterson+Ives+Walters+Lyon">search</a>]]]></summary>
<link rel="alternate" href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=Modelling+the+Distortion+Produced+by+Cochlear+Compression+Patterson+Ives+Walters+Lyon" type="text/html"/>
<category term="Non-Speech Audio Processing" label="Non-Speech Audio Processing"/>
<author><name>Roy D. Patterson, Timothy Ives, Thomas C. Walters, Richard F. Lyon</name></author>
</entry>
<entry>
<title><![CDATA[Safe ICF: Pointer Safe and Unwinding Aware Identical Code Folding in Gold]]></title>
<updated>2012-05-14T10:44:22Z</updated>
<id>urn:googlelabs:36912</id>
<summary type="html"><![CDATA[We have found that large C++ applications and shared libraries tend to have many functions whose code is identical with another function. As much as 10% of the code could theoretically be eliminated by merging such identical functions into a single copy. This optimization, Identical Code Folding (ICF), has been implemented in the gold  linker. At link time, ICF detects functions with identical object code and merges them into a single copy. ICF can be unsafe, however, as it can change the run-time behaviour of code that relies on each function having a unique address. To address this, ICF can be used in a safe mode where it identifies and folds functions whose addresses are guaranteed not to have been used in comparison operations. 

Further, profiling and debugging binaries with merged functions can be confusing, as the PC values of merged functions cannot be always disambiguated to point to the correct function. To address this, we propose a new call table format for the DWARF debugging information to allow tools like the debugger and profiler to disambiguate PC values of merged functions correctly by examining the call chain. 

Detailed experiments on the x86 platform show that ICF can reduce the text size of a selection of Google binaries, whose average text size is 64 MB, by about 6%. Also, the code size savings of ICF with the safe option is almost as good as the code savings obtained without the safe option. Further, experiments also show that the run-time performance of the optimized binaries on the x86 platform does not change.  Links: [<a href="http://research.google.com/pubs/pub36912.html">abstract</a>] [<a href="http://research.google.com/pubs/archive/36912.pdf">pdf</a>]  [<a href="http://gcc.gnu.org/wiki/summit2010?action=AttachFile&do=view&target=tallam.pdf">gcc.gnu.org</a>] [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=Safe+ICF%3A+Pointer+Safe+and+Unwinding+Aware+Identical+Code+Folding+in+Gold+Tallam+Coutant+Taylor+Li+Demetriou">search</a>]]]></summary>
<link rel="alternate" href="http://research.google.com/pubs/pub36912.html" type="text/html"/>
<category term="Compilers" label="Compilers"/>
<author><name>Sriraman Tallam, Cary Coutant, Ian Lance Taylor, Xinliang David Li, Chris Demetriou</name></author>
</entry>
<entry>
<title><![CDATA[ShellOS: Enabling fast detection and forensic analysis of code injection attacks]]></title>
<updated>2012-05-14T09:24:55Z</updated>
<id>urn:googlelabs:38102</id>
<summary type="html"><![CDATA[The availability of off-the-shelf exploitation toolkits for
compromising hosts, coupled with the rapid rate of
exploit discovery and disclosure, has made exploit or
vulnerability-based detection far less effective than it
once was. For instance, the increasing use of metamorphic and polymorphic techniques to deploy code injection attacks continues to confound signature-based detection techniques. The key to detecting these attacks
lies in the ability to discover the presence of the injected
code (or, shellcode). One promising technique for doing so is to examine data (be that from network streams
or buffers of a process) and efﬁciently execute its content to ﬁnd what lurks within. Unfortunately, current approaches for achieving this goal are not robust to evasion or scalable, primarily because of their reliance on
software-based CPU emulators. In this paper, we argue that the use of software-based emulation techniques
are not necessary, and instead propose a new framework
that leverages hardware virtualization to better enable the
detection of code injection attacks. We also report on
our experience using this framework to analyze a corpus
of malicious Portable Document Format (PDF) ﬁles and
network-based attacks.  Links: [<a href="http://research.google.com/pubs/pub38102.html">abstract</a>] [<a href="http://static.usenix.org/events/sec11/tech/full_papers/Snow.pdf">static.usenix.org</a>] [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=ShellOS%3A+Enabling+fast+detection+and+forensic+analysis+of+code+injection+attacks+Snow+Krishnan+Monrose+Provos">search</a>]]]></summary>
<link rel="alternate" href="http://research.google.com/pubs/pub38102.html" type="text/html"/>
<category term="Security, Cryptography, and Privacy" label="Security, Cryptography, and Privacy"/>
<author><name>Kevin Snow, Srinivas Krishnan, Fabian Monrose, Niels Provos</name></author>
</entry>
<entry>
<title><![CDATA[RFC6583 - Operational Neighbor Discovery Problems]]></title>
<updated>2012-05-14T14:00:54Z</updated>
<id>urn:googlelabs:38101</id>
<summary type="html"><![CDATA[In IPv4, subnets are generally small, made just large enough to cover
   the actual number of machines on the subnet.  In contrast, the
   default IPv6 subnet size is a /64, a number so large it covers
   trillions of addresses, the overwhelming number of which will be
   unassigned.  Consequently, simplistic implementations of Neighbor
   Discovery (ND) can be vulnerable to deliberate or accidental denial
   of service (DoS), whereby they attempt to perform address resolution
   for large numbers of unassigned addresses.  Such denial-of-service
   attacks can be launched intentionally (by an attacker) or result from
   legitimate operational tools or accident conditions.  As a result of
   these vulnerabilities, new devices may not be able to &quot;join&quot; a
   network, it may be impossible to establish new IPv6 flows, and
   existing IPv6 transported flows may be interrupted.

   This document describes the potential for DoS in detail and suggests
   possible implementation improvements as well as operational
   mitigation techniques that can, in some cases, be used to protect
   against or at least alleviate the impact of such attacks.  Links: [<a href="http://research.google.com/pubs/pub38101.html">abstract</a>] [<a href="http://research.google.com/pubs/archive/38101.pdf">pdf</a>]  [<a href="http://tools.ietf.org/rfc/rfc6583.txt">tools.ietf.org</a>] [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=RFC6583+-+Operational+Neighbor+Discovery+Problems+Kumari+Yahoo%21+Zynga">search</a>]]]></summary>
<link rel="alternate" href="http://research.google.com/pubs/pub38101.html" type="text/html"/>
<category term="Networks" label="Networks"/>
<author><name>Warren Kumari, Igor Gashinsky, Yahoo!, Joel Jaeggli, Zynga</name></author>
</entry>
<entry>
<title><![CDATA[A Comparison of Chinese Parsers for Stanford Dependencies]]></title>
<updated>2012-05-12T16:47:34Z</updated>
<id>urn:googlelabs:38100</id>
<summary type="html"><![CDATA[A Comparison of Chinese Parsers for Stanford Dependencies, Wanxiang Che, Valentin I. Spitkovsky, Ting Liu  Links:  [<a href="http://research.google.com/pubs/archive/38100.pdf">pdf</a>]  [<a href="http://stanford.edu/~vals/pubs/stanford_dependencies_chinese.pdf">stanford.edu</a>] [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=A+Comparison+of+Chinese+Parsers+for+Stanford+Dependencies+Che+Spitkovsky+Liu">search</a>]]]></summary>
<link rel="alternate" href="http://research.google.com/pubs/archive/38100.pdf" type="application/pdf"/>
<category term="Natural Language Processing" label="Natural Language Processing"/>
<author><name>Wanxiang Che, Valentin I. Spitkovsky, Ting Liu</name></author>
</entry>
<entry>
<title><![CDATA[Capitalization Cues Improve Dependency Grammar Induction]]></title>
<updated>2012-05-12T16:45:01Z</updated>
<id>urn:googlelabs:38099</id>
<summary type="html"><![CDATA[Capitalization Cues Improve Dependency Grammar Induction, Valentin I. Spitkovsky, Daniel Jurafsky, Hiyan Alshawi  Links:  [<a href="http://research.google.com/pubs/archive/38099.pdf">pdf</a>]  [<a href="http://stanford.edu/~vals/pubs/capitalization.pdf">stanford.edu</a>] [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=Capitalization+Cues+Improve+Dependency+Grammar+Induction+Spitkovsky+Jurafsky+Alshawi">search</a>]]]></summary>
<link rel="alternate" href="http://research.google.com/pubs/archive/38099.pdf" type="application/pdf"/>
<category term="Natural Language Processing" label="Natural Language Processing"/>
<author><name>Valentin I. Spitkovsky, Daniel Jurafsky, Hiyan Alshawi</name></author>
</entry>
<entry>
<title><![CDATA[A Cross-Lingual Dictionary for English Wikipedia Concepts]]></title>
<updated>2012-05-12T16:39:57Z</updated>
<id>urn:googlelabs:38098</id>
<summary type="html"><![CDATA[A Cross-Lingual Dictionary for English Wikipedia Concepts, Valentin I. Spitkovsky, Angel X. Chang  Links:  [<a href="http://research.google.com/pubs/archive/38098.pdf">pdf</a>]  [<a href="http://stanford.edu/~vals/pubs/crosswikis.pdf">stanford.edu</a>] [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=A+Cross-Lingual+Dictionary+for+English+Wikipedia+Concepts+Spitkovsky+Chang">search</a>]]]></summary>
<link rel="alternate" href="http://research.google.com/pubs/archive/38098.pdf" type="application/pdf"/>
<category term="Hypertext and the Web" label="Hypertext and the Web"/>
<author><name>Valentin I. Spitkovsky, Angel X. Chang</name></author>
</entry>
<entry>
<title><![CDATA[Smart Pricing Grows the Pie]]></title>
<updated>2012-05-11T13:39:14Z</updated>
<id>urn:googlelabs:38097</id>
<summary type="html"><![CDATA[Some publisher advertising networks provide features intended to help advertisers bid more efficiently with a single bid in many publishers’ click auctions at once – Smart Pricing on the Google Display Network is one example. Typically such features involve discounting advertiser bids or prices for clicks on publisher websites according to how click values vary across sites (for some appropriate measure of advertiser value). Contrary to concerns that such features necessarily result in reduced publisher (and network) revenue we find that, in many simple cases, the modified auction dynamics produce rational incentives for advertisers to bid more – and spend more – than they would without the benefit of these features. So if advertisers act in their own interest then publishers and networks stand to make more revenue as well.  Links: [<a href="http://research.google.com/pubs/pub38097.html">abstract</a>] [<a href="http://research.google.com/pubs/archive/38097.pdf">pdf</a>]  [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=Smart+Pricing+Grows+the+Pie+Calvert">search</a>]]]></summary>
<link rel="alternate" href="http://research.google.com/pubs/pub38097.html" type="text/html"/>
<category term="Economics" label="Economics"/>
<author><name>Guy Calvert</name></author>
</entry>
<entry>
<title><![CDATA[Income Inequality in the Attention Economy]]></title>
<updated>2012-05-09T16:49:34Z</updated>
<id>urn:googlelabs:33367</id>
<summary type="html"><![CDATA[Income Inequality in the Attention Economy, Kevin S. McCurley  Links:  [<a href="http://research.google.com/pubs/archive/33367.pdf">pdf</a>]  [<a href="http://mccurley.org/papers/effective/">mccurley.org</a>] [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=Income+Inequality+in+the+Attention+Economy+McCurley">search</a>]]]></summary>
<link rel="alternate" href="http://research.google.com/pubs/archive/33367.pdf" type="application/pdf"/>
<category term="Hypertext and the Web" label="Hypertext and the Web"/>
<author><name>Kevin S. McCurley</name></author>
</entry>
<entry>
<title><![CDATA[Model Recommendation for Action Recognition]]></title>
<updated>2012-05-08T12:54:51Z</updated>
<id>urn:googlelabs:38093</id>
<summary type="html"><![CDATA[Simply choosing one model out of a large set of possibilities for a given vision task is a surprisingly difficult problem, especially if there is limited evaluation data with which to distinguish among models, such as when choosing the best ``walk&#39;&#39; action classifier from a large pool of classifiers tuned for different viewing angles, lighting conditions, and background clutter.  In this paper we suggest that this problem of selecting a good model can be recast as a recommendation problem, where the goal is to recommend a good model for a particular task based on how well a limited probe set of models appears to perform. Through this conceptual remapping, we can bring to bear all the collaborative filtering techniques developed for consumer recommender systems (e.g., Netflix, Amazon.com). We test this hypothesis on action recognition, and find that even when every model has been directly rated on a training set, recommendation finds better selections for the corresponding test set than the best performers on the training set.  Links: [<a href="http://research.google.com/pubs/pub38093.html">abstract</a>] [<a href="http://research.google.com/pubs/archive/38093.pdf">pdf</a>]  [<a href="http://www.google.com/search?lr=&amp;ie=UTF-8&amp;oe=UTF-8&amp;q=Model+Recommendation+for+Action+Recognition+Matikainen+Sukthankar+Hebert">search</a>]]]></summary>
<link rel="alternate" href="http://research.google.com/pubs/pub38093.html" type="text/html"/>
<category term="Computer Vision" label="Computer Vision"/>
<author><name>Pyry Matikainen, Rahul Sukthankar, Martial Hebert</name></author>
</entry>
</feed>

