<?php
$link = "publications";
$title = "Publications";
include("inc/header.php");

function print_pub_link($file)
{
   echo "<a href=\"$file\" onClick=\"recordOutboundLink(this, 'Publications', '$file');return false;\">";
}
?>

<script language="JavaScript" type="text/JavaScript">
function bibtexit(entry_name, entry)
{
   var span_entry_summary_link = document.getElementById(entry_name + "_summary_link");
   if (span_entry_summary_link)
      span_entry_summary_link.innerHTML = "+Summary";

   var span_entry_link = document.getElementById(entry_name + "_bibtex_link");
   var span_entry = document.getElementById(entry_name);

   if (span_entry_link.innerHTML == "+BibTeX")
   {
      span_entry_link.innerHTML = "-BibTeX";
      span_entry.style.color = '#F29900';
      span_entry.style.marginLeft = '10px';
      span_entry.style.marginRight = '25px';
      span_entry.style.padding = '5px';
      span_entry.style.border = '1px solid #7D8085';
      span_entry.style.background = '#000';
      span_entry.style.fontFamily = 'Courier New';
      span_entry.innerHTML = entry;
   }
   else
   {
      span_entry_link.innerHTML = "+BibTeX";
      span_entry.innerHTML = "";
      span_entry.style.margin = '0px';
      span_entry.style.padding = '0px';
      span_entry.style.border = '0px';
   }
}
function summaryit(entry_name, entry)
{
   var span_entry_bibtex_link = document.getElementById(entry_name + "_bibtex_link");
   if (span_entry_bibtex_link)
   span_entry_bibtex_link.innerHTML = "+BibTeX";

   var span_entry_link = document.getElementById(entry_name + "_summary_link");
   var span_entry = document.getElementById(entry_name);
   if (span_entry_link.innerHTML == "+Summary")
   {
      span_entry_link.innerHTML = "-Summary";
      span_entry.style.color = '#a0a0a0';
      span_entry.style.marginLeft = '10px';
      span_entry.style.marginRight = '25px';
      span_entry.style.padding = '5px';
      span_entry.style.border = '1px solid #7D8085';
      span_entry.style.background = '#000';
      span_entry.style.fontFamily = 'Sans-Serif';
      span_entry.innerHTML = entry;
   }
   else
   {
      span_entry_link.innerHTML = "+Summary";
      span_entry.innerHTML = "";
      span_entry.style.margin = '0px';
      span_entry.style.padding = '0px';
      span_entry.style.border = '0px';
   }
}

function printCode(link)
{
   return "<a href='" + link + "' onClick=\"recordOutboundLink(this, 'Publications', '" + link + "');return false;\">";
}

var codeShowing=-1;
function showCode(idnum)
{
	var codebox = document.getElementById("codeDesc");
	if (idnum==codeShowing)
	{
		codeShowing = -1;
		codebox.innerHTML = "";
		return;
	}

	codeShowing = idnum;
	if (idnum==0)
		codebox.innerHTML = "The following code is from our ICCV 2013 paper on layered tracking. It combines the level-set sampling algorithm in a temporal framework to track multiple objects in videos." +
			"<p align='center'>[ Coming Soon ]</p>";
	else if (idnum==1)
		codebox.innerHTML = "The following code is our latest level-set/MRF sampler as described in our ICCV 2013 paper. It fixes errors that were in the ICIP 2012 and CVPR 2011 paper that slightly biased results towards the mode (thank you <a href='http://www.mosaic.ethz.ch/people/janickc' target='_BLANK'>Janick</a> and <a href='http://www.mosaic.ethz.ch/people/ivos' target='_BLANK'>Ivo</a> identifying inconsistencies)." + 
			"<p align='center'>[ " + printCode('code/PGIMH/') + "zip/tar</a> ]</p>" + 
			"For completeness, our erroneous code is listed below. These should not be used and will not sample from the distribution of interest." +
			"<p align='center'>ICIP 2012: [ " + printCode('code/GIMH/') + "zip/tar</a> ]  &nbsp;&nbsp;&nbsp; CVPR 2011: [ " + printCode('code/BFPS/') + "zip/tar</a> ]</p>";
	else if (idnum==2)
		codebox.innerHTML = "The following code is the Dirichlet process mixture model sampler as described in our NIPS 2013 paper. It includes code for Gaussian observation models with Normal-Inverse Wishart and Multinomial observation models with a Dirichlet prior." +
			"<p align='center'>[ " + printCode('code/subclusters/dpmm_subclusters_2014-01-19.zip') + "zip</a> ] - [ " + printCode('code/subclusters/dpmm_subclusters_2014-01-19.tar.gz') + "tar.gz</a> ]</p>";
	else if (idnum==3)
		codebox.innerHTML = "The following code infers Temporal Superpixels for a particular video sequence as described in our CVPR 2013 paper." +
		"<p align='center'>[ " + printCode('code/TSP/TSP_2013-06-26.zip') + "zip</a> ] - [ " + printCode('code/TSP/TSP_2013-06-26.tar.gz') + "tar.gz</a> ]</p>";
}
</script>

<h2>Publications</h2>
<div class="boxpub">

<table class="research"><tr><td><img src="research/AISTATS2015.png"></td><td>
      Julian Straub, <u>Jason Chang</u>, Oren Freifeld, and John W. Fisher III<br>
      <mark>A Dirichlet Process Mixture Model for Spherical Data</mark><br>
      <i>International Conference on Artificial Intelligence and Statistics (AISTATS)</i>, San Diego, CA, USA May 2015. (Oral)<br>
[ <?php print_pub_link("pubs/publications/straub15_AISTATS.pdf");?>PDF</a> ] -
[ <?php print_pub_link("pubs/publications/straub15_AISTATS_supplemental.pdf");?>Supp. Material</a> ] -
&nbsp;
<font face="Courier New"><a href="javascript:bibtexit('straub15_AISTATS', '@INPROCEEDINGS{straub15dptgmm,<br>author = {Straub, J. and Chang, J. and Freifeld, O. and Fisher III, J. W.},<br>title = {A Dirichlet Process Mixture Model for Spherical Data},<br>booktitle = {Proceedings of the International Conference on Artificial Intelligence and Statistics (AISTATS)},<br>year = {2015},<br>month = {May}<br>}')"><span id="straub15_AISTATS_bibtex_link">+BibTeX</span></a></\font>
</td></tr></table>
<div id="straub15_AISTATS"></div>

<table class="research"><tr><td><img src="research/NIPS2014.jpg"></td><td>
      <u>Jason Chang</u> and John W. Fisher III<br>
      <mark>Parallel Sampling of HDPs using Sub-Cluster Splits</mark><br>
      <i>Neural Information and Processing Systems (NIPS)</i>, Montreal, Quebec, Canada Dec 2014.<br>
[ <?php print_pub_link("pubs/publications/chang14_NIPS.pdf");?>PDF</a> ] -
[ <?php print_pub_link("pubs/publications/chang14_NIPS_supplemental.pdf");?>Supp. Material</a> ] - 
[ <?php print_pub_link("pubs/posters/2014-NIPS.pdf");?>Poster</a> ] -
[ <a href="code.php#hdp_subclusters">Code</a> ]
&nbsp;
<font face="Courier New"><a href="javascript:bibtexit('chang14_NIPS', '@INPROCEEDINGS{chang14hdp,<br>author = {Chang, J and Fisher III, J. W.},<br>title = {Parallel Sampling of HDPs using Sub-Cluster Splits},<br>booktitle = {Proceedings of the Neural Information Processing Systems (NIPS)},<br>year = {2014},<br>month = {Dec}<br>}')"><span id="chang14_NIPS_bibtex_link">+BibTeX</span></a></font>
</td></tr></table>
<div id="chang14_NIPS"></div>
      

<table class="research"><tr><td><img src="research/ECCV2014.png"></td><td>
      <u>Jason Chang</u>, Randi Cabezas, and John W. Fisher III<br>
      <mark>Bayesian Nonparametric Intrinsic Image Decomposition</mark><br>
      <i>European Conference on Computer Vision (ECCV)</i>, Zurich, Switzerland, Sept 2014. (Oral)<br>
[ <?php print_pub_link("pubs/publications/chang14_ECCV.pdf");?>PDF</a> ] -
[ <?php print_pub_link("http://videolectures.net/eccv2014_chang_image_decomposition/");?>Presentation</a> ] - 
[ Code ] - 
&nbsp;
<font face="Courier New"><a href="javascript:bibtexit('chang14_ECCV', '@INPROCEEDINGS{chang14svdpgmm,<br>author = {Chang, J and Cabezas, R. and Fisher III, J. W.},<br>title = {Bayesian Nonparametric Intrinsic Image Decomposition},<br>booktitle = {Proceedings of the European Conference on Computer Vision (ECCV)},<br>year = {2014},<br>month = {Sept}<br>}')"><span id="chang14_ECCV_bibtex_link">+BibTeX</span></a></font>
</td></tr></table>
<div id="chang14_ECCV"></div>

<table class="research"><tr><td><img src="research/NIPS2013.png"></td><td>
      <u>Jason Chang</u> and John W. Fisher III<br>
      <mark>Parallel Sampling of DP Mixture Models using Sub-Clusters Splits</mark><br>
      <i>Neural Information and Processing Systems (NIPS)</i>, Lake Tahoe, NV, USA, Dec 2013.<br>
[ <?php print_pub_link("pubs/publications/chang13_NIPS.pdf");?>PDF</a> ] -
[ <?php print_pub_link("pubs/publications/chang13_NIPS_supplemental.pdf");?>Supp. Material</a> ] -
[ <?php print_pub_link("pubs/posters/2013-NIPS.pdf");?>Poster</a> ] -
[ <a href="code.php#dpmm_subclusters">Code</a> ]
&nbsp;
<font face="Courier New"><a href="javascript:bibtexit('chang13_NIPS', '@INPROCEEDINGS{chang13dpmm,<br>author = {Chang, J and Fisher III, J. W.},<br>title = {Parallel Sampling of DP Mixture Models using Sub-Clusters Splits},<br>booktitle = {Proceedings of the Neural Information Processing Systems (NIPS)},<br>year = {2013},<br>month = {Dec}<br>}')"><span id="chang13_NIPS_bibtex_link">+BibTeX</span></a></font>
</td></tr></table>
<div id="chang13_NIPS"></div>

<table class="research"><tr><td><img src="research/ICCV2013.png"></td><td>
      <u>Jason Chang</u> and John W. Fisher III<br>
      <mark>Topology-Constrained Layered Tracking with Latent Flow</mark><br>
      <i>IEEE International Conference on Computer Vision (ICCV)</i>, Syndey, Australia, Dec 2013.<br>
[ <?php print_pub_link("pubs/publications/chang13_ICCV.pdf");?>PDF</a> ] -
[ <?php print_pub_link("pubs/publications/chang13_ICCV_supplemental.zip");?>Supp. Material</a> ] -
[ <?php print_pub_link("pubs/posters/2013-ICCV.pdf");?>Poster</a> ] -
[ <a href="code.php#tracker">Code</a> ]
&nbsp;
<font face="Courier New"><a href="javascript:bibtexit('chang13_ICCV', '@INPROCEEDINGS{chang13tracking,<br>author = {Chang, J. and Fisher III, J. W.},<br>title = {Topology-Constrained Layered Tracking with Latent Flow},<br>booktitle = {Proceedings of the IEEE International Conference on Computer Vision (ICCV)},<br>year = {2013},<br>month = {Dec}<br>}')"><span id="chang13_ICCV_bibtex_link">+BibTeX</span></a></font>
&nbsp;
<!--font face="Courier New"><a href="javascript:summaryit('chang11_CVPR', '<img src=\'pubs/publications/summary/chang11_CVPR.png\'>We present a method for sampling from the posterior distribution of implicitly defined segmentations conditioned on the observed image. Segmentation is often formulated as an energy minimization or statistical inference problem in which either the optimal or most probable configuration is the goal. Exponentiating the negative energy functional provides a Bayesian interpretation in which the solutions are equivalent. Sampling methods enable evaluation of distribution properties that characterize the solution space via the computation of marginal event probabilities. We develop a Metropolis-Hastings sampling algorithm over level-sets which improves upon previous methods by allowing for topological changes while simultaneously decreasing computational times by orders of magnitude. An M-ary extension to the method is provided.<br><br>Results applying our sampling method to four different image features applied to the Berkeley Segmentation Dataset can be found <a href=\'pubs/image_browser/image_browser.html\' target=\'_BLANK\'>here</a>.')"><span id="chang11_CVPR_summary_link">+Summary</span></a></font-->
</td></tr></table>

<div id="chang13_ICCV"></div>


<table class="research"><tr><td><img src="research/CVPR2013.png"></td><td>
      <u>Jason Chang</u>, Donglai Wei, and John W. Fisher III<br>
      <mark>A Video Representation Using Temporal Superpixels</mark><br>
      <i>IEEE Computer Vision and Pattern Recognition (CVPR)</i>, Portland, OR, USA, June 2013.<br>
[ <?php print_pub_link("pubs/publications/chang13_CVPR.pdf");?>PDF</a> ] -
[ <?php print_pub_link("pubs/publications/chang13_CVPR_supplemental.zip");?>Supp. Material</a> ] -
[ <?php print_pub_link("pubs/posters/2013-CVPR.pdf");?>Poster</a> ] -
[ <a href="code.php#TSP">Code</a> ]
<!--[ <?php print_pub_link("code/TSP/TSP_2013-06-26.zip");?>Code</a> ] -->
&nbsp;
<font face="Courier New"><a href="javascript:bibtexit('chang13_CVPR', '@INPROCEEDINGS{chang13tsp,<br>author = {Chang, J. and Wei, D. and Fisher III, J. W.},<br>title = {A Video Representation Using Temporal Superpixels},<br>booktitle = {Proceedings of the IEEE Computer Vision and Pattern Recognition (CVPR)},<br>year = {2013},<br>month = {June}<br>}')"><span id="chang13_CVPR_bibtex_link">+BibTeX</span></a></font>
&nbsp;
<!--font face="Courier New"><a href="javascript:summaryit('chang11_CVPR', '<img src=\'pubs/publications/summary/chang11_CVPR.png\'>We present a method for sampling from the posterior distribution of implicitly defined segmentations conditioned on the observed image. Segmentation is often formulated as an energy minimization or statistical inference problem in which either the optimal or most probable configuration is the goal. Exponentiating the negative energy functional provides a Bayesian interpretation in which the solutions are equivalent. Sampling methods enable evaluation of distribution properties that characterize the solution space via the computation of marginal event probabilities. We develop a Metropolis-Hastings sampling algorithm over level-sets which improves upon previous methods by allowing for topological changes while simultaneously decreasing computational times by orders of magnitude. An M-ary extension to the method is provided.<br><br>Results applying our sampling method to four different image features applied to the Berkeley Segmentation Dataset can be found <a href=\'pubs/image_browser/image_browser.html\' target=\'_BLANK\'>here</a>.')"><span id="chang11_CVPR_summary_link">+Summary</span></a></font-->
</td></tr></table>
<div id="chang13_CVPR"></div>

<table class="research"><tr><td><img src="research/ICIP2012.png"></td><td>
	<u>Jason Chang</u> and John W. Fisher III<br>
	<mark>Efficient Topology-Controlled Sampling of Implicit Shapes</mark><br>
	<i>IEEE International Conference on Image Processing (ICIP)</i>, Orlando, FL, USA, Sept 2012.<br>
[ <?php print_pub_link("pubs/publications/chang12_ICIP.pdf");?>PDF</a> ] -
[ <a href="http://arxiv.org/abs/1205.3766">arXiv</a> ] -
[ <?php print_pub_link("pubs/posters/2012-ICIP.pdf");?>Poster</a> ] -
[ <a href="code.php#PGIMH">Code</a> ]
<!-- - [ <?php //print_pub_link("pubs/code/GIMH.php");?>Code</a> ] -->
&nbsp;
<font face="Courier New"><a href="javascript:bibtexit('chang12_ICIP', '@INPROCEEDINGS{chang12,<br>author = {Chang, J. and Fisher III, J. W.},<br>title = {Efficient Topology-Controlled Sampling of Implicit Shapes},<br>booktitle = {Proceedings of the IEEE International Conference on Image Processing (ICIP)},<br>year = {2012},<br>month = {Sept}<br>}')"><span id="chang12_ICIP_bibtex_link">+BibTeX</span></a></font>
&nbsp;
<!--font face="Courier New"><a href="javascript:summaryit('chang11_CVPR', '<img src=\'pubs/publications/summary/chang11_CVPR.png\'>We present a method for sampling from the posterior distribution of implicitly defined segmentations conditioned on the observed image. Segmentation is often formulated as an energy minimization or statistical inference problem in which either the optimal or most probable configuration is the goal. Exponentiating the negative energy functional provides a Bayesian interpretation in which the solutions are equivalent. Sampling methods enable evaluation of distribution properties that characterize the solution space via the computation of marginal event probabilities. We develop a Metropolis-Hastings sampling algorithm over level-sets which improves upon previous methods by allowing for topological changes while simultaneously decreasing computational times by orders of magnitude. An M-ary extension to the method is provided.<br><br>Results applying our sampling method to four different image features applied to the Berkeley Segmentation Dataset can be found <a href=\'pubs/image_browser/image_browser.html\' target=\'_BLANK\'>here</a>.')"><span id="chang11_CVPR_summary_link">+Summary</span></a></font-->
</td></tr></table>
<div id="chang12_ICIP"></div>

<table class="research"><tr><td><img src="research/CVPR2011.jpg"></td><td>
      <u>Jason Chang</u> and John W. Fisher III<br>
      <mark>Efficient MCMC Sampling with Implicit Shape Representations</mark><br>
      <i>IEEE Computer Vision and Pattern Recognition (CVPR)</i>, Colorado Springs, USA, June 2011.<br>
[ <?php print_pub_link("pubs/publications/chang11_CVPR.pdf");?>PDF</a> ] -
[ <?php print_pub_link("pubs/publications/chang11_CVPR_supplemental.zip");?>Supp. Material</a> ] -
[ <?php print_pub_link("pubs/posters/2011-CVPR.pdf");?>Poster</a> ] -
[ <?php print_pub_link("pubs/image_browser/image_browser.html");?>BSDS Results</a> ] -
[ <a href="code.php#PGIMH">Code</a> ]
<!-- - [ <?php //print_pub_link("pubs/code/BFPS.php");?>Code</a> ] -->
<!--[ <a href="pubs/publications/chang11_CVPR.pdf">PDF</a> ]
[ <a href="pubs/publications/chang11_CVPR_supplemental.zip">Supp. Material</a> ] -
[ <a href="pubs/image_browser/image_browser.html" target="_BLANK">BSDS Results</a> ] -
[ <a href="pubs/code/BFPS.php">Code</a> ]-->
&nbsp;
<font face="Courier New"><a href="javascript:bibtexit('chang11_CVPR', '@INPROCEEDINGS{chang11,<br>author = {Chang, J. and Fisher III, J. W.},<br>title = {Efficient MCMC Sampling with Implicit Shape Representations},<br>booktitle = {Proceedings of the IEEE Computer Vision and Pattern Recognition (CVPR)},<br>year = {2011},<br>month = {June}<br>}')"><span id="chang11_CVPR_bibtex_link">+BibTeX</span></a></font>
&nbsp;
<font face="Courier New"><a href="javascript:summaryit('chang11_CVPR', '<img src=\'pubs/publications/summary/chang11_CVPR.png\'>We present a method for sampling from the posterior distribution of implicitly defined segmentations conditioned on the observed image. Segmentation is often formulated as an energy minimization or statistical inference problem in which either the optimal or most probable configuration is the goal. Exponentiating the negative energy functional provides a Bayesian interpretation in which the solutions are equivalent. Sampling methods enable evaluation of distribution properties that characterize the solution space via the computation of marginal event probabilities. We develop a Metropolis-Hastings sampling algorithm over level-sets which improves upon previous methods by allowing for topological changes while simultaneously decreasing computational times by orders of magnitude. An M-ary extension to the method is provided.<br><br>Results applying our sampling method to four different image features applied to the Berkeley Segmentation Dataset can be found <a href=\'pubs/image_browser/image_browser.html\' target=\'_BLANK\'>here</a>.')"><span id="chang11_CVPR_summary_link">+Summary</span></a></font>
</td></tr></table>
<div id="chang11_CVPR"></div>


<table class="research"><tr><td><img src="research/ICCV2009.png"></td><td>
      <u>Jason Chang</u> and John W. Fisher III<br>
      <mark>Analysis of Orientation and Scale in Smoothly Varying Textures</mark><br>
      <i>IEEE International Conference on Computer Vision (ICCV)</i>, Kyoto, Japan, Sept 2009.<br>
[ <?php print_pub_link("pubs/publications/chang09_ICCV.pdf");?>PDF</a> ] -
[ <?php print_pub_link("pubs/publications/chang09_ICCV_supplemental.zip");?>Supp. Material</a> ] - 
[ <?php print_pub_link("pubs/posters/2009-ICCV.pdf");?>Poster</a> ]
<!--[ <a href="pubs/publications/chang09_ICCV.pdf">PDF</a> ] -
[ <a href="pubs/publications/chang09_ICCV_supplemental.zip">Supp. Material</a> ]-->
&nbsp;
<font face="Courier New"><a href="javascript:bibtexit('chang09_ICCV', '@INPROCEEDINGS{chang09,<br>author = {Chang, J. and Fisher III, J. W.},<br>title = {Analysis of Orientation and Scale in Smoothly Varying Textures},<br>booktitle = {Proceedings of the IEEE International Conference on Computer Vision (ICCV)},<br>year = {2009},<br>month = {Sept}<br>}')"><span id="chang09_ICCV_bibtex_link">+BibTeX</span></a></font>
&nbsp;
<font face="Courier New"><a href="javascript:summaryit('chang09_ICCV', '<img src=\'pubs/publications/summary/chang09_ICCV.png\'>We present a novel representation for modeling textured regions subject to smooth variations in orientation and scale. Utilizing the steerable pyramid of Simoncelli and Freeman as a basis, we decompose textured regions of natural images into explicit local attributes of contrast, bias, scale, and orientation. Additionally, we impose smoothness on these attributes via Markov random fields. The combination allows for demonstrable improvements in common scene analysis applications including unsupervised segmentation, reflectance and shading estimation, and estimation of the radiometric response function from a single image.')"><span id="chang09_ICCV_summary_link">+Summary</span></a></font>
</td></tr></table>
<div id="chang09_ICCV"></div>

</div>
<br>

<h2>Code</h2>
<div class="box">
   <!--table align="center"><tr>
      <td align><a href="javascript:showCode(0)"><div class="boxlink">Layered Tracking Sampler<br><img src="research/layered_particle_filter.png"></div></a></td>
      <td align><a href="javascript:showCode(1)"><div class="boxlink">Level-Set / MRF Sampler<br><img src="research/levelset_sampling.jpg"></div></a></td>
      <td align><a href="javascript:showCode(2)"><div class="boxlink">Dirichlet Process Mixtures<br><img src="research/dpmm.png"></div></a></td>
      <td align><a href="javascript:showCode(3)"><div class="boxlink">Temporal Superpixels<br><img src="research/tsp.png"></div></a></td>
   </tr></table>
   <p align="center">All code is interfaced through Matlab and has been tested in a Linux environment compiled with GCC.</p>
   <span id="codeDesc"></span-->
      Please visit <a href="code.php">here</a> to see code used in my publications.
</div>
<br>

<h2>Reports and Theses</h2>
<div class="box">
   <ul>
      <li>J. Chang, "Sampling in Computer Vision and Bayesian Nonparametric Mixtures", Ph.D. Thesis. MIT, Cambridge, MA June. 2014.
      [ <?php print_pub_link("pubs/publications/chang14_PHD_thesis.pdf");?>PDF</a> ]</li>
      <li>J. Chang and J. W. Fisher III, "Efficient topology-controlled sampling of implicit shapes," May 2012, arXiv:1205.3766v1 [cs.CV].
      [ <?php print_pub_link("http://arxiv.org/abs/1205.3766");?>PDF</a> ]</li-->
      <li>J. Chang, "Extracting Orientation and Scale from Smoothing Varying Textures with Application to Segmentation", S.M. Thesis. MIT, Cambridge, MA Sept. 2009.
      [ <?php print_pub_link("pubs/publications/chang09_SM_thesis.pdf");?>PDF</a> ]</li>
      <!--li>Master's Thesis Proposal (2009)
      [ <a href="pubs/reports/2009-ms_proposal.pdf">PDF</a> ]</li-->
      <li>Discrete-Time Signal Processing Project 1 - Selected for Solutions (2007)
      [ <?php print_pub_link("pubs/reports/2007-6341_project1.pdf");?>PDF</a> ]</li>
      <li>Discrete-Time Signal Processing Project 2 (2007)
      [ <?php print_pub_link("pubs/reports/2008-6341_project2.pdf");?>PDF</a> ]</li>
      <li>Undergraduate Senior Design Report (2007)
      [ <?php print_pub_link("pubs/reports/2007-senior_design.pdf");?>PDF</a> ]</li>
      <li>Fast Hierarchical Filtered Back Projection - Digital Imaging Final Project (2007)
      [ <?php print_pub_link("pubs/reports/2007-digital_imaging.pdf");?>PDF</a> ]</li>
   </ul>
</div>
<br>

<h2>Presentations</h2>
<div class="box">
   <ul>
      <li>A Video Representation using Temporal Superpixels - ECCV IWVS Invited Short Talk (2014)</li>
      <li>Bayesian Nonparametric Intrinsic Image Decomposition - ECCV Oral (2014)</li>
      <li>Modeling Textures and Computing Statistics of Object Shapes in Images - BU Image and Video Computer Group Seminar (2011)</li>
      <li>Inferring Relative Seismic Age - Stochastic Systems Group Seminar (2011)</li>
      <li>Efficient MCMC Sampling with Implicit Shape Representations - CSAIL Computer Vision Seminar (2011)
      [ <?php print_pub_link("pubs/presentations/2011-Vision_Talk.zip");?>PPT</a> ] [ <?php print_pub_link("pubs/presentations/2011-Vision_Talk.pdf");?>PDF</a> ]
      <li>Analysis of Smoothly Varying Textures - Research Qualifying Exam (2010)
      [ <?php print_pub_link("pubs/presentations/2010-RQE.pdf");?>PDF</a> ]
      <li>Analysis of Smoothly Varying Textures Applied to Segmentation - Automatic Target Recognition Center Workshop* (2009)
      [ <?php print_pub_link("pubs/presentations/2009-ATRC_presentation.zip");?>PPT</a> ]
      <li>Texture Based Image Segmentation - Stochastic Systems Group Seminar (2008)
      [ <?php print_pub_link("pubs/presentations/2008-ssg_presentation.zip");?>PPT</a> ] [ <?php print_pub_link("pubs/presentations/2008-ssg_presentation.pdf");?>PDF</a> ]
      <li>Texture Based Image Segmentation - Airforce Sponsors Visit (2008)
      [ <?php print_pub_link("pubs/presentations/2008-airforce.pdf");?>PDF</a> ]
      <li>Music Search Engine - Undergraduate Senior Design Presentation (2007)
      [ <?php print_pub_link("pubs/presentations/2007-senior_design.ppt");?>PPT</a> ] [ <?php print_pub_link("pubs/presentations/2007-senior_design.pdf");?>PDF</a> ]
      <li>Fast Hierarchical Filtered Back Projection - Digital Imaging Final Project (2007)
      [ <?php print_pub_link("pubs/presentations/2007-digital_imaging.ppt");?>PPT</a> ] [ <?php print_pub_link("pubs/presentations/2007-digital_imaging.pdf");?>PDF</a> ]
   </ul>
</div>
<br>

<h2>Posters</h2>
<div class="box">
   <ul>
      <li>MCMC Sampling in HDPs using Sub-Clusters - NIPS, Montreal, Canada (2014) [ PDF ]</li>
      <li>Bayesian Nonparametric Intrinsic Image Decomposition - ECCV, Zurich, Switzerland (2014) [ PDF ]</li>
      <li>Parallel Sampling of DP Mixture Models using Sub-Clusters Splits - NIPS, Lake Tahoe, NV (2013)
      [ <?php print_pub_link("pubs/posters/2013-NIPS.pdf");?>PDF</a> ]</li>
      <li>Topology-Constrained Layered Tracking with Latent Flow - ICCV, Sydney, AU (2013)
      [ <?php print_pub_link("pubs/posters/2013-ICCV.pdf");?>PDF</a> ]</li>
      <li>A Video Representation Using Temporal Superpixels - CVPR, Portland, OR (2013)
      [ <?php print_pub_link("pubs/posters/2013-CVPR.pdf");?>PDF</a> ]</li>
      <li>Efficient Topology-Controlled Sampling of Implicit Shapes - ICIP, Orlando, FL (2012)
      [ <?php print_pub_link("pubs/posters/2012-ICIP.pdf");?>PDF</a> ]</li>
      <li>Efficient MCMC Sampling with Implicit Shape Representations - CVPR, Colorado Springs, CO (2011)
      [ <?php print_pub_link("pubs/posters/2011-CVPR.pdf");?>PDF</a> ]</li>
      <li>Analysis of Orientation and Scale in Smoothly Varying Textures - ICCV, Kyoto, Japan (2009)
      [ <?php print_pub_link("pubs/posters/2009-ICCV.pdf");?>PDF</a> ]</li>
      <li>Analysis of Smoothly Varying Textures Applied to Segmentation - ATRC Workshop, Dayton, OH (2009)
      [ <?php print_pub_link("pubs/posters/2009-ATRC.pdf");?>PDF</a> ]</li>
      <li>Modeling Slowly Varying Textures - CSAIL Industry Affiliates Program Conference, MIT (2009)
      [ <?php print_pub_link("pubs/posters/2009-IAP.pdf");?>PDF</a> ]</li>
      <li>Decentralized Detection Schemes Applied to Cognitive Radio Networks - Vodafone Symposium (2007)
      [ <?php print_pub_link("pubs/posters/2007-vodafone.pdf");?>PDF</a> ]</li>
   </ul>
</div>
<br>

<h2>Patents</h2>
<div class="box">
   <ul>
      <li>Efficient MCMC Sampling with Implicit Shape Representations. U.S. Patent #8744185 B2, June 3, 2014.
   </ul>
</div>

<br><br>
*Note: Movies only play in powerpoint and are usually compressed using the <a target="BLANK"href="http://www.xvid.org/">Xvid</a> (*.avi) or the H264 (*.mp4) codec.
<?php include("inc/footer.php"); ?>
