{"id":2758,"date":"2014-06-24T17:56:33","date_gmt":"2014-06-24T23:56:33","guid":{"rendered":"http:\/\/hunch.net\/?p=2758"},"modified":"2014-06-29T02:41:15","modified_gmt":"2014-06-29T08:41:15","slug":"interesting-papers-at-icml-2014","status":"publish","type":"post","link":"https:\/\/hunch.net\/?p=2758","title":{"rendered":"Interesting papers at ICML 2014"},"content":{"rendered":"<p>This year&#8217;s ICML had several papers which I want to read through more carefully and understand better.<\/p>\n<ol>\n<li>Chun-Liang Li, <a href=\"http:\/\/www.csie.ntu.edu.tw\/~htlin\/\">Hsuan-Tien Lin<\/a>, <a href=\"http:\/\/jmlr.org\/proceedings\/papers\/v32\/lia14.pdf\">Condensed Filter Tree for Cost-Sensitive Multi-Label Classification<\/a>.  Several tricks accumulate to give a new approach for addressing cost sensitive multilabel classification.<\/li>\n<li><a href=\"http:\/\/lowrank.net\/nikos\/index.html\">Nikos Karampatziakis<\/a> and <a href=\"http:\/\/www.machinedlearnings.com\/\">Paul Mineiro<\/a>, <a href=\"http:\/\/jmlr.org\/proceedings\/papers\/v32\/karampatziakis14.pdf\">Discriminative Features via Generalized Eigenvectors<\/a>.  An efficient, effective eigenvalue solution for supervised learning yields compelling nonlinear performance on several datasets.<\/li>\n<li><a href=\"http:\/\/www.cs.technion.ac.il\/~nailon\/homepage\/homepage.html\">Nir Ailon<\/a>, <a href=\"http:\/\/labs.yahoo.com\/author\/zkarnin\/\">Zohar Karnin<\/a>, <a href=\"http:\/\/www.cs.cornell.edu\/people\/tj\/\">Thorsten Joachims<\/a>, <a href=\"http:\/\/jmlr.org\/proceedings\/papers\/v32\/ailon14.pdf\">Reducing Dueling Bandits to Cardinal Bandits<\/a>.  An effective method for reducing dueling bandits to normal bandits that extends to contextual situations.<\/li>\n<li>Pedro Pinheiro, <a href=\"http:\/\/ronan.collobert.com\/\">Ronan Collobert<\/a>, <a href=\"http:\/\/jmlr.org\/proceedings\/papers\/v32\/pinheiro14.pdf\">Recurrent Convolutional Neural Networks for Scene Labeling<\/a>.  Image parsing remains a challenge, and this is plausibly a step forward.<\/li>\n<li><a href=\"http:\/\/researcher.watson.ibm.com\/researcher\/view.php?person=br-cicerons\">Cicero Dos Santos<\/a>, Bianca Zadrozny, <a href=\"http:\/\/jmlr.org\/proceedings\/papers\/v32\/santos14.pdf\">Learning Character-level Representations for Part-of-Speech Tagging<\/a>. Word morphology is clearly useful information, and yet almost all ML-for-NLP applications ignore it or hard-code it (by stemming).<\/li>\n<li><a href=\"http:\/\/research.microsoft.com\/en-us\/um\/people\/alekha\/\">Alekh Agarwal<\/a>, <a href=\"http:\/\/www.cs.columbia.edu\/~djhsu\/\">Daniel Hsu<\/a>, <a href=\"http:\/\/www.satyenkale.com\/\">Satyen Kale<\/a>, <a href=\"https:\/\/hunch.net\/~jl\">John Langford<\/a>, <a href=\"http:\/\/www.research.rutgers.edu\/~lihong\/\">Lihong Li<\/a>, <a href=\"http:\/\/www.cs.princeton.edu\/~schapire\/\">Robert Schapire<\/a>, <a href=\"http:\/\/jmlr.org\/proceedings\/papers\/v32\/agarwalb14.pdf\">Taming the Monster: A Fast and Simple Algorithm for Contextual Bandits<\/a>.  Statistically efficient interactive learning is now computationally feasible.  I wish this one had been done in time for the <a href=\"https:\/\/hunch.net\/~jl\/interact.pdf\">NIPS tutorial<\/a> \ud83d\ude42<\/li>\n<li>David Silver, Guy Lever, <a href=\"http:\/\/homepages.inf.ed.ac.uk\/s0677090\/\">Nicolas Heess<\/a>, <a href=\"http:\/\/people.bordeaux.inria.fr\/degris\/publications.html\">Thomas Degris<\/a>, <a href=\"http:\/\/www.idsia.ch\/~daan\/\">Daan Wierstra<\/a>, <a href=\"http:\/\/ml.informatik.uni-freiburg.de\/people\/riedmiller\/info\">Martin Riedmiller<\/a>, <a href=\"http:\/\/jmlr.org\/proceedings\/papers\/v32\/silver14.pdf\">Deterministic Policy Gradient Algorithms<\/a>.  A reduction in variance from working out the deterministic limit of policy gradient make policy gradient approaches look much more attractive.<\/li>\n<\/ol>\n<p>Edit: added one that I forgot.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>This year&#8217;s ICML had several papers which I want to read through more carefully and understand better. Chun-Liang Li, Hsuan-Tien Lin, Condensed Filter Tree for Cost-Sensitive Multi-Label Classification. Several tricks accumulate to give a new approach for addressing cost sensitive multilabel classification. Nikos Karampatziakis and Paul Mineiro, Discriminative Features via Generalized Eigenvectors. An efficient, effective &hellip; <\/p>\n<p class=\"link-more\"><a href=\"https:\/\/hunch.net\/?p=2758\" class=\"more-link\">Continue reading<span class=\"screen-reader-text\"> &#8220;Interesting papers at ICML 2014&#8221;<\/span><\/a><\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[33,29,18],"tags":[],"class_list":["post-2758","post","type-post","status-publish","format-standard","hentry","category-conferences","category-machine-learning","category-papers"],"_links":{"self":[{"href":"https:\/\/hunch.net\/index.php?rest_route=\/wp\/v2\/posts\/2758","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/hunch.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/hunch.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/hunch.net\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/hunch.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2758"}],"version-history":[{"count":0,"href":"https:\/\/hunch.net\/index.php?rest_route=\/wp\/v2\/posts\/2758\/revisions"}],"wp:attachment":[{"href":"https:\/\/hunch.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2758"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/hunch.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2758"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hunch.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2758"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}