Newer
Older
Digital_Repository / Misc / Mass downloads / UTas / 825.html
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html>
  <head>
    <title>UTas ePrints - Gradient Descent Style Leveraging of Decision Trees and Stumps for Misclassification Cost Performance</title>
    <script type="text/javascript" src="http://eprints.utas.edu.au/javascript/auto.js"><!-- padder --></script>
    <style type="text/css" media="screen">@import url(http://eprints.utas.edu.au/style/auto.css);</style>
    <style type="text/css" media="print">@import url(http://eprints.utas.edu.au/style/print.css);</style>
    <link rel="icon" href="/images/eprints/favicon.ico" type="image/x-icon" />
    <link rel="shortcut icon" href="/images/eprints/favicon.ico" type="image/x-icon" />
    <link rel="Top" href="http://eprints.utas.edu.au/" />
    <link rel="Search" href="http://eprints.utas.edu.au/cgi/search" />
    <meta content="Cameron-Jones, Mike" name="eprints.creators_name" />
<meta content="conference_item" name="eprints.type" />
<meta content="2007-03-19" name="eprints.datestamp" />
<meta content="2008-01-08 15:30:00" name="eprints.lastmod" />
<meta content="show" name="eprints.metadata_visibility" />
<meta content="Gradient Descent Style Leveraging of Decision Trees and Stumps for Misclassification Cost Performance" name="eprints.title" />
<meta content="pub" name="eprints.ispublished" />
<meta content="280200" name="eprints.subjects" />
<meta content="280000" name="eprints.subjects" />
<meta content="public" name="eprints.full_text_status" />
<meta content="paper" name="eprints.pres_type" />
<meta content="The original publication is available at www.springerlink.com" name="eprints.note" />
<meta content="This paper investigates the use, for the task of classifier learning in the presence of misclassification costs, of some gradient descent style leveraging approaches to classifier learning: Schapire and Singer's AdaBoost.MH and AdaBoost.MR [16], and Collins et al's multi-class logistic regression method [4], and some modifications that retain the gradient descent style approach. Decision trees and stumps are used as the underlying base classifiers, learned from modified versions of Quinlan's C4.5 [15]. Experiments are reported comparing the performance, in terms of average cost, of the modified methods to that of the originals, and to the previously suggested &quot;Cost Boosting&quot; methods of Ting and Zheng [21] and Ting [18], which also use decision trees based upon modified C4.5 code, but do not have an interpretation in the gradient descent framework. While some of the modifications improve upon the originals in terms of cost performance for both trees and stumps, the comparison with tree-based Cost Boosting suggests that out of the methods first experimented with here, it is one based on stumps that has the most promise.
" name="eprints.abstract" />
<meta content="2001-12" name="eprints.date" />
<meta content="published" name="eprints.date_type" />
<meta content="12" name="eprints.pages" />
<meta content="AI 2001: Advances in Artificial Intelligence, 14th Australian Joint Conference on Artificial Intelligence" name="eprints.event_title" />
<meta content="Adelaide, Australia" name="eprints.event_location" />
<meta content="10-14 Dec 2001" name="eprints.event_dates" />
<meta content="conference" name="eprints.event_type" />
<meta content="UNSPECIFIED" name="eprints.thesis_type" />
<meta content="TRUE" name="eprints.refereed" />
<meta content="http://www.springerlink.com/content/8wa6n94gm9ufe6dv/" name="eprints.official_url" />
<meta content="Cameron-Jones, Mike (2001) Gradient Descent Style Leveraging of Decision Trees and Stumps for Misclassification Cost Performance. In: AI 2001: Advances in Artificial Intelligence, 14th Australian Joint Conference on Artificial Intelligence, 10-14 Dec 2001, Adelaide, Australia." name="eprints.citation" />
<meta content="http://eprints.utas.edu.au/825/1/OzAI2001.pdf" name="eprints.document_url" />
<link rel="schema.DC" href="http://purl.org/DC/elements/1.0/" />
<meta content="Gradient Descent Style Leveraging of Decision Trees and Stumps for Misclassification Cost Performance" name="DC.title" />
<meta content="Cameron-Jones, Mike" name="DC.creator" />
<meta content="280200 Artificial Intelligence and Signal and Image Processing" name="DC.subject" />
<meta content="280000 Information, Computing and Communication Sciences" name="DC.subject" />
<meta content="This paper investigates the use, for the task of classifier learning in the presence of misclassification costs, of some gradient descent style leveraging approaches to classifier learning: Schapire and Singer's AdaBoost.MH and AdaBoost.MR [16], and Collins et al's multi-class logistic regression method [4], and some modifications that retain the gradient descent style approach. Decision trees and stumps are used as the underlying base classifiers, learned from modified versions of Quinlan's C4.5 [15]. Experiments are reported comparing the performance, in terms of average cost, of the modified methods to that of the originals, and to the previously suggested &quot;Cost Boosting&quot; methods of Ting and Zheng [21] and Ting [18], which also use decision trees based upon modified C4.5 code, but do not have an interpretation in the gradient descent framework. While some of the modifications improve upon the originals in terms of cost performance for both trees and stumps, the comparison with tree-based Cost Boosting suggests that out of the methods first experimented with here, it is one based on stumps that has the most promise.
" name="DC.description" />
<meta content="2001-12" name="DC.date" />
<meta content="Conference or Workshop Item" name="DC.type" />
<meta content="PeerReviewed" name="DC.type" />
<meta content="application/pdf" name="DC.format" />
<meta content="http://eprints.utas.edu.au/825/1/OzAI2001.pdf" name="DC.identifier" />
<meta content="http://www.springerlink.com/content/8wa6n94gm9ufe6dv/" name="DC.relation" />
<meta content="Cameron-Jones, Mike (2001) Gradient Descent Style Leveraging of Decision Trees and Stumps for Misclassification Cost Performance. In: AI 2001: Advances in Artificial Intelligence, 14th Australian Joint Conference on Artificial Intelligence, 10-14 Dec 2001, Adelaide, Australia." name="DC.identifier" />
<meta content="http://eprints.utas.edu.au/825/" name="DC.relation" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/BibTeX/epprod-eprint-825.bib" title="BibTeX" type="text/plain" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/ContextObject/epprod-eprint-825.xml" title="OpenURL ContextObject" type="text/xml" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/ContextObject::Dissertation/epprod-eprint-825.xml" title="OpenURL Dissertation" type="text/xml" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/ContextObject::Journal/epprod-eprint-825.xml" title="OpenURL Journal" type="text/xml" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/DC/epprod-eprint-825.txt" title="Dublin Core" type="text/plain" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/DIDL/epprod-eprint-825.xml" title="DIDL" type="text/xml" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/EndNote/epprod-eprint-825.enw" title="EndNote" type="text/plain" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/HTML/epprod-eprint-825.html" title="HTML Citation" type="text/html; charset=utf-8" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/METS/epprod-eprint-825.xml" title="METS" type="text/xml" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/MODS/epprod-eprint-825.xml" title="MODS" type="text/xml" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/RIS/epprod-eprint-825.ris" title="Reference Manager" type="text/plain" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/Refer/epprod-eprint-825.refer" title="Refer" type="text/plain" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/Simple/epprod-eprint-825text" title="Simple Metadata" type="text/plain" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/Text/epprod-eprint-825.txt" title="ASCII Citation" type="text/plain; charset=utf-8" />
<link rel="alternate" href="http://eprints.utas.edu.au/cgi/export/825/XML/epprod-eprint-825.xml" title="EP3 XML" type="text/xml" />

  </head>
  <body bgcolor="#ffffff" text="#000000" onLoad="loadRoutine(); MM_preloadImages('images/eprints/ePrints_banner_r5_c5_f2.gif','images/eprints/ePrints_banner_r5_c7_f2.gif','images/eprints/ePrints_banner_r5_c8_f2.gif','images/eprints/ePrints_banner_r5_c9_f2.gif','images/eprints/ePrints_banner_r5_c10_f2.gif','images/eprints/ePrints_banner_r5_c11_f2.gif','images/eprints/ePrints_banner_r6_c4_f2.gif')">
    
    <div class="ep_noprint"><noscript><style type="text/css">@import url(http://eprints.utas.edu.au/style/nojs.css);</style></noscript></div>




<table width="795" border="0" cellspacing="0" cellpadding="0">
  <tr>
    <td><script language="JavaScript1.2">mmLoadMenus();</script>
      <table border="0" cellpadding="0" cellspacing="0" width="795">
        <!-- fwtable fwsrc="eprints_banner_final2.png" fwbase="ePrints_banner.gif" fwstyle="Dreamweaver" fwdocid = "1249563342" fwnested="0" -->
        <tr>
          <td><img src="/images/eprints/spacer.gif" width="32" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="104" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="44" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="105" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="41" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="16" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="68" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="68" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="68" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="82" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="69" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="98" height="1" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="1" height="1" border="0" alt="" /></td>
        </tr>
        <tr>
          <td colspan="12"><img name="ePrints_banner_r1_c1" src="/images/eprints/ePrints_banner_r1_c1.gif" width="795" height="10" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="1" height="10" border="0" alt="" /></td>
        </tr>
        <tr>
          <td rowspan="6"><img name="ePrints_banner_r2_c1" src="/images/eprints/ePrints_banner_r2_c1.gif" width="32" height="118" border="0" alt="" /></td>
          <td rowspan="5"><a href="http://www.utas.edu.au/"><img name="ePrints_banner_r2_c2" src="/images/eprints/ePrints_banner_r2_c2.gif" width="104" height="103" border="0" alt="" /></a></td>
          <td colspan="10"><img name="ePrints_banner_r2_c3" src="/images/eprints/ePrints_banner_r2_c3.gif" width="659" height="41" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="1" height="41" border="0" alt="" /></td>
        </tr>
        <tr>
          <td colspan="3"><a href="http://eprints.utas.edu.au/"><img name="ePrints_banner_r3_c3" src="/images/eprints/ePrints_banner_r3_c3.gif" width="190" height="31" border="0" alt="" /></a></td>
          <td rowspan="2" colspan="7"><img name="ePrints_banner_r3_c6" src="/images/eprints/ePrints_banner_r3_c6.gif" width="469" height="37" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="1" height="31" border="0" alt="" /></td>
        </tr>
        <tr>
          <td colspan="3"><img name="ePrints_banner_r4_c3" src="/images/eprints/ePrints_banner_r4_c3.gif" width="190" height="6" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="1" height="6" border="0" alt="" /></td>
        </tr>
        <tr>
          <td colspan="2"><img name="ePrints_banner_r5_c3" src="/images/eprints/ePrints_banner_r5_c3.gif" width="149" height="1" border="0" alt="" /></td>
          <td rowspan="2" colspan="2"><a href="/information.html" onMouseOut="MM_swapImgRestore();MM_startTimeout()" onMouseOver="MM_showMenu(window.mm_menu_0821132634_0,0,25,null,'ePrints_banner_r5_c5');MM_swapImage('ePrints_banner_r5_c5','','/images/eprints/ePrints_banner_r5_c5_f2.gif',1);"><img name="ePrints_banner_r5_c5" src="/images/eprints/ePrints_banner_r5_c5.gif" width="57" height="25" border="0" alt="About" /></a></td>
          <td rowspan="2"><a href="/view/" onMouseOut="MM_swapImgRestore();MM_startTimeout()" onMouseOver="MM_showMenu(window.mm_menu_0821133021_1,0,25,null,'ePrints_banner_r5_c7');MM_swapImage('ePrints_banner_r5_c7','','/images/eprints/ePrints_banner_r5_c7_f2.gif',1);"><img name="ePrints_banner_r5_c7" src="/images/eprints/ePrints_banner_r5_c7.gif" width="68" height="25" border="0" alt="Browse" /></a></td>
          <td rowspan="2"><a href="/perl/search/simple" onMouseOut="MM_swapImgRestore();MM_startTimeout()" onMouseOver="MM_showMenu(window.mm_menu_0821133201_2,0,25,null,'ePrints_banner_r5_c8');MM_swapImage('ePrints_banner_r5_c8','','/images/eprints/ePrints_banner_r5_c8_f2.gif',1);"><img name="ePrints_banner_r5_c8" src="/images/eprints/ePrints_banner_r5_c8.gif" width="68" height="25" border="0" alt="Search" /></a></td>
          <td rowspan="2"><a href="/perl/register" onMouseOut="MM_swapImgRestore();MM_startTimeout();" onMouseOver="MM_showMenu(window.mm_menu_1018171924_3,0,25,null,'ePrints_banner_r5_c9');MM_swapImage('ePrints_banner_r5_c9','','/images/eprints/ePrints_banner_r5_c9_f2.gif',1);"><img name="ePrints_banner_r5_c9" src="/images/eprints/ePrints_banner_r5_c9.gif" width="68" height="25" border="0" alt="register" /></a></td>
          <td rowspan="2"><a href="/perl/users/home" onMouseOut="MM_swapImgRestore();MM_startTimeout()" onMouseOver="MM_showMenu(window.mm_menu_0821133422_4,0,25,null,'ePrints_banner_r5_c10');MM_swapImage('ePrints_banner_r5_c10','','/images/eprints/ePrints_banner_r5_c10_f2.gif',1);"><img name="ePrints_banner_r5_c10" src="/images/eprints/ePrints_banner_r5_c10.gif" width="82" height="25" border="0" alt="user area" /></a></td>
          <td rowspan="2"><a href="/help/" onMouseOut="MM_swapImgRestore();MM_startTimeout()" onMouseOver="MM_showMenu(window.mm_menu_0821133514_5,0,25,null,'ePrints_banner_r5_c11');MM_swapImage('ePrints_banner_r5_c11','','/images/eprints/ePrints_banner_r5_c11_f2.gif',1);"><img name="ePrints_banner_r5_c11" src="/images/eprints/ePrints_banner_r5_c11.gif" width="69" height="25" border="0" alt="Help" /></a></td>
          <td rowspan="3" colspan="4"><img name="ePrints_banner_r5_c12" src="/images/eprints/ePrints_banner_r5_c12.gif" width="98" height="40" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="1" height="1" border="0" alt="" /></td>
        </tr>
        <tr>
          <td rowspan="2"><img name="ePrints_banner_r6_c3" src="/images/eprints/ePrints_banner_r6_c3.gif" width="44" height="39" border="0" alt="ePrints home" /></td>
          <td><a href="/" onMouseOut="MM_swapImgRestore()" onMouseOver="MM_swapImage('ePrints_banner_r6_c4','','/images/eprints/ePrints_banner_r6_c4_f2.gif',1);"><img name="ePrints_banner_r6_c4" src="/images/eprints/ePrints_banner_r6_c4.gif" width="105" height="24" border="0" alt="ePrints home" /></a></td>
          <td><img src="/images/eprints/spacer.gif" width="1" height="24" border="0" alt="" /></td>
        </tr>
        <tr>
          <td><img name="ePrints_banner_r7_c2" src="/images/eprints/ePrints_banner_r7_c2.gif" width="104" height="15" border="0" alt="" /></td>
          <td colspan="8"><img name="ePrints_banner_r7_c4" src="/images/eprints/ePrints_banner_r7_c4.gif" width="517" height="15" border="0" alt="" /></td>
          <td><img src="/images/eprints/spacer.gif" width="1" height="15" border="0" alt="" /></td>
        </tr>
      </table></td>
  </tr>
    <tr><td><table width="100%" style="font-size: 90%; border: solid 1px #ccc; padding: 3px"><tr>
      <td align="left"><a href="http://eprints.utas.edu.au/cgi/users/home">Login</a> | <a href="http://eprints.utas.edu.au/cgi/register">Create Account</a></td>
      <td align="right" style="white-space: nowrap">
        <form method="get" accept-charset="utf-8" action="http://eprints.utas.edu.au/cgi/search" style="display:inline">
          <input class="ep_tm_searchbarbox" size="20" type="text" name="q" />
          <input class="ep_tm_searchbarbutton" value="Search" type="submit" name="_action_search" />
          <input type="hidden" name="_order" value="bytitle" />
          <input type="hidden" name="basic_srchtype" value="ALL" />
          <input type="hidden" name="_satisfyall" value="ALL" />
        </form>
      </td>
    </tr></table></td></tr>
  <tr>
    <td class="toplinks"><!-- InstanceBeginEditable name="content" -->


<div align="center">
  
  <table width="720" class="ep_tm_main"><tr><td align="left">
    <h1 class="ep_tm_pagetitle">Gradient Descent Style Leveraging of Decision Trees and Stumps for Misclassification Cost Performance</h1>
    <p style="margin-bottom: 1em" class="not_ep_block"><span class="person_name">Cameron-Jones, Mike</span> (2001) <xhtml:em>Gradient Descent Style Leveraging of Decision Trees and Stumps for Misclassification Cost Performance.</xhtml:em> In: AI 2001: Advances in Artificial Intelligence, 14th Australian Joint Conference on Artificial Intelligence, 10-14 Dec 2001, Adelaide, Australia.</p><p style="margin-bottom: 1em" class="not_ep_block"></p><table style="margin-bottom: 1em" class="not_ep_block"><tr><td valign="top" style="text-align:center"><a onmouseover="EPJS_ShowPreview( event, 'doc_preview_825' );" href="http://eprints.utas.edu.au/825/1/OzAI2001.pdf" onmouseout="EPJS_HidePreview( event, 'doc_preview_825' );"><img alt="[img]" src="http://eprints.utas.edu.au/style/images/fileicons/application_pdf.png" class="ep_doc_icon" border="0" /></a><div class="ep_preview" id="doc_preview_825"><table><tr><td><img alt="" src="http://eprints.utas.edu.au/825/thumbnails/1/preview.png" class="ep_preview_image" border="0" /><div class="ep_preview_title">Preview</div></td></tr></table></div></td><td valign="top"><a href="http://eprints.utas.edu.au/825/1/OzAI2001.pdf"><span class="ep_document_citation">PDF (Author Version)</span></a> - Requires a PDF viewer<br />424Kb</td></tr></table><p style="margin-bottom: 1em" class="not_ep_block">Official URL: <a href="http://www.springerlink.com/content/8wa6n94gm9ufe6dv/">http://www.springerlink.com/content/8wa6n94gm9ufe6dv/</a></p><div class="not_ep_block"><h2>Abstract</h2><p style="padding-bottom: 16px; text-align: left; margin: 1em auto 0em auto">This paper investigates the use, for the task of classifier learning in the presence of misclassification costs, of some gradient descent style leveraging approaches to classifier learning: Schapire and Singer's AdaBoost.MH and AdaBoost.MR [16], and Collins et al's multi-class logistic regression method [4], and some modifications that retain the gradient descent style approach. Decision trees and stumps are used as the underlying base classifiers, learned from modified versions of Quinlan's C4.5 [15]. Experiments are reported comparing the performance, in terms of average cost, of the modified methods to that of the originals, and to the previously suggested "Cost Boosting" methods of Ting and Zheng [21] and Ting [18], which also use decision trees based upon modified C4.5 code, but do not have an interpretation in the gradient descent framework. While some of the modifications improve upon the originals in terms of cost performance for both trees and stumps, the comparison with tree-based Cost Boosting suggests that out of the methods first experimented with here, it is one based on stumps that has the most promise.
</p></div><table style="margin-bottom: 1em" cellpadding="3" class="not_ep_block" border="0"><tr><th valign="top" class="ep_row">Item Type:</th><td valign="top" class="ep_row">Conference or Workshop Item (Paper)</td></tr><tr><th valign="top" class="ep_row">Additional Information:</th><td valign="top" class="ep_row">The original publication is available at www.springerlink.com</td></tr><tr><th valign="top" class="ep_row">Subjects:</th><td valign="top" class="ep_row"><a href="http://eprints.utas.edu.au/view/subjects/280200.html">280000 Information, Computing and Communication Sciences &gt; 280200 Artificial Intelligence and Signal and Image Processing</a><br /><a href="http://eprints.utas.edu.au/view/subjects/280000.html">280000 Information, Computing and Communication Sciences</a></td></tr><tr><th valign="top" class="ep_row">ID Code:</th><td valign="top" class="ep_row">825</td></tr><tr><th valign="top" class="ep_row">Deposited By:</th><td valign="top" class="ep_row"><span class="ep_name_citation"><span class="person_name">Dr R. Mike Cameron-Jones</span></span></td></tr><tr><th valign="top" class="ep_row">Deposited On:</th><td valign="top" class="ep_row">19 Mar 2007</td></tr><tr><th valign="top" class="ep_row">Last Modified:</th><td valign="top" class="ep_row">09 Jan 2008 02:30</td></tr><tr><th valign="top" class="ep_row">ePrint Statistics:</th><td valign="top" class="ep_row"><a target="ePrintStats" href="/es/index.php?action=show_detail_eprint;id=825;">View statistics for this ePrint</a></td></tr></table><p align="right">Repository Staff Only: <a href="http://eprints.utas.edu.au/cgi/users/home?screen=EPrint::View&amp;eprintid=825">item control page</a></p>
  </td></tr></table>
</div>



    <!-- InstanceEndEditable --></td>
  </tr>
  <tr>
    <td><!-- #BeginLibraryItem "/Library/footer_eprints.lbi" -->
    <table width="795" border="0" align="left" cellpadding="0" class="footer">
  <tr valign="top">
<td colspan="2"><div align="center"><a href="http://www.utas.edu.au">UTAS home</a> | <a href="http://www.utas.edu.au/library/">Library home</a> | <a href="/">ePrints home</a> | <a href="/contact.html">contact</a> | <a href="/information.html">about</a> | <a href="/view/">browse</a> | <a href="/perl/search/simple">search</a> | <a href="/perl/register">register</a> | <a href="/perl/users/home">user area</a> | <a href="/help/">help</a></div><br /></td>
</tr>
<tr><td colspan="2"><p><img src="/images/eprints/footerline.gif" width="100%" height="4" /></p></td></tr>
      <tr valign="top">
        <td width="68%" class="footer">Authorised by the University Librarian<br />
© University of Tasmania ABN 30 764 374 782<br />
      <a href="http://www.utas.edu.au/cricos/">CRICOS Provider Code 00586B</a> | <a href="http://www.utas.edu.au/copyright/copyright_disclaimers.html">Copyright &amp; Disclaimers</a> | <a href="http://www.utas.edu.au/accessibility/index.html">Accessibility</a> | <a href="http://eprints.utas.edu.au/feedback/">Site Feedback</a>  </td>
        <td width="32%"><div align="right">
            <p align="right" class="NoPrint"><a href="http://www.utas.edu.au/"><img src="http://www.utas.edu.au/shared/logos/unioftasstrip.gif" alt="University of Tasmania Home Page" width="260" height="16" border="0" align="right" /></a></p>
            <p align="right" class="NoPrint"><a href="http://www.utas.edu.au/"><br />
            </a></p>
        </div></td>
      </tr>
      <tr valign="top">
        <td><p>  </p></td>
        <td><div align="right"><span class="NoPrint"><a href="http://www.eprints.org/software/"><img src="/images/eprintslogo.gif" alt="ePrints logo" width="77" height="29" border="0" align="bottom" /></a></span></div></td>
      </tr>
    </table>
    <!-- #EndLibraryItem -->
    <div align="center"></div></td>
  </tr>
</table>

  </body>
</html>