{"id":398369,"date":"2017-07-10T11:45:52","date_gmt":"2017-07-10T18:45:52","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-project&#038;p=398369"},"modified":"2023-04-03T10:54:30","modified_gmt":"2023-04-03T17:54:30","slug":"deep-learning-machine-reading-comprehension","status":"publish","type":"msr-project","link":"https:\/\/www.microsoft.com\/en-us\/research\/project\/deep-learning-machine-reading-comprehension\/","title":{"rendered":"Deep Learning for Machine Reading Comprehension"},"content":{"rendered":"<div class=\"asset-content\">The goal of this project is to teach a computer to read and answer general questions pertaining to a document. We recently released a large scale MRC dataset, <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/ms-marco-human-generated-machine-reading-comprehension-dataset\/\">MS MARCO<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>.\u00a0 We developed a <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/reasonet-learning-stop-reading-machine-comprehension\/\">ReasoNet\u00a0<span class=\"sr-only\"> (opens in new tab)<\/span><\/a> model to mimic the inference process of human readers. With a question in mind, ReasoNets read a document repeatedly, each time focusing on different parts of the document until a satisfying answer is found or formed. The extension of ReasoNet (<a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/implicit-reasonet-modeling-large-scale-structured-relationships-shared-memory\/\">ReasoNet-Memory<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>) incorporates the shared memory component in the model has been applied on Knowledge Graph Completition Task. We also develop a a <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/publication\/two-stage-synthesis-networks-transfer-learning-machine-comprehension\/\">two-stage synthesis network (SynNet)<span class=\"sr-only\"> (opens in new tab)<\/span><\/a>\u00a0 for transfer learning in machine reading comprehension. Our latest MRC model, called\u00a0<a class=\"msr-external-link glyph-append glyph-append-open-in-new-tab glyph-append-xsmall\" href=\"https:\/\/arxiv.org\/abs\/1712.03556\">SAN<span class=\"sr-only\"> (opens in new tab)<\/span><\/a> (Stochastic Answer Net), simulates multi-step reasoning using stochastic prediction dropout, achieving state-of-the-art on SQuAD.<\/div>\n<div id=\"en-usprojectsdssmdefault\" class=\"page-content\"><\/div>\n<pre><\/pre>\n<p>&nbsp;<\/p>\n","protected":false},"excerpt":{"rendered":"<p>The goal of this project is to teach a computer to read and answer general questions pertaining to a document. We recently released a large scale MRC dataset, MS MARCO.\u00a0 We developed a ReasoNet\u00a0 model to mimic the inference process of human readers. With a question in mind, ReasoNets read a document repeatedly, each time [&hellip;]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","footnotes":""},"research-area":[13556,13545],"msr-locale":[268875],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-398369","msr-project","type-msr-project","status-publish","hentry","msr-research-area-artificial-intelligence","msr-research-area-human-language-technologies","msr-locale-en_us","msr-archive-status-active"],"msr_project_start":"2016-09-01","related-publications":[294695,328118,328361,418055,450966,450972,168090,168302],"related-downloads":[571575],"related-videos":[],"related-groups":[],"related-events":[],"related-opportunities":[],"related-posts":[],"related-articles":[],"tab-content":[],"slides":[],"related-researchers":[{"type":"user_nicename","display_name":"Weizhu Chen","user_id":34863,"people_section":"Group name 1","alias":"wzchen"},{"type":"user_nicename","display_name":"Jianfeng Gao","user_id":32246,"people_section":"Group name 1","alias":"jfgao"},{"type":"user_nicename","display_name":"Xiaodong Liu","user_id":34877,"people_section":"Group name 1","alias":"xiaodl"},{"type":"guest","display_name":"Kevin Duh","user_id":400301,"people_section":"Group name 1","alias":""}],"msr_research_lab":[199565],"msr_impact_theme":[],"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/398369"}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-project"}],"version-history":[{"count":11,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/398369\/revisions"}],"predecessor-version":[{"id":451839,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/398369\/revisions\/451839"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=398369"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=398369"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=398369"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=398369"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=398369"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}