This week’s seminar is organized as follows:
1. The seminar time is 1.pm, Wed.<\/strong>, at Zhongguancun Technology Park ,Building 5, 1306.
2. The lecturer is Baohua Zhang<\/strong>, the paper’s title is Curriculum Learning for Natural Answer Generation<\/strong>.
3. The seminar will be hosted by Zhaoyang Wang.
4. Attachment is the paper of this seminar, please download in advance.

<\/p>\n\n\n\n

Anyone interested in this topic is welcomed to join us. the following is the abstract for this week\u2019s paper.<\/p>\n\n\n\n

\n\t\t
\n\t\t\tCurriculum Learning for Natural Answer Generation\n\t\t<\/p>\n\t\t
\n\t\t\tCao Liu, Shizhu He, Kang\nLiu, Jun Zhao\n\t\t<\/p>\n\t\t
\n\t\t\tAbstract\n\t\t<\/p>\n\t\t
\n\t\t\t By reason of being able to obtain natural\nlanguage responses, natural answers are more favored in real-world Question\nAnswering (QA) systems. Generative models learn to automatically generate\nnatural answers from large-scale question answer pairs (QA-pairs). However,\nthey are suffering from the uncontrollable and uneven quality of QA-pairs\ncrawled from the Internet. To address this problem, we propose a curriculum\nlearning based framework for natural answer generation (CL-NAG), which is able\nto take full advantage of the valuable learning data from a noisy and\nuneven-quality corpora. Specifically, we employ two practical measures to\nautomatically measure the quality (complexity) of QA-pairs. Based on the\nmeasurements, CLNAG firstly utilizes simple and low-quality QApairs to learn a\nbasic model, and then gradually learns to produce better answers with richer\ncontents and more complete syntaxes based on more complex and higher-quality\nQA-pairs. In this way, all valuable information in the noisy and unevenquality\ncorpora could be fully exploited. Experiments demonstrate that CL-NAG\noutperforms the state-of-the-art, which increases 6.8% and 8.7% in the accuracy\nfor simple and complex questions, respectively.<\/span>\n\t\t<\/p>\n\t\t
\n\t\t\t<\/span>\n\t\t<\/p>\n\t<\/div>\n<\/p>\n\n\n\n
$\"\"$ <\/figure>\n\n\n\n
Curriculum Learning for Natural Answer Generation<\/a>\u4e0b\u8f7d<\/a><\/div>\n\n\n\n\n\n\n\n
NLPIR SEMINAR 14th ISSUE COMPLETED<\/strong><\/h2>\n\n\n\n
Last week, BaohuaZhang gave a presentation about the paper, Curriculum Learning for Natural Answer Generation<\/strong>, and shared some opinion on it.<\/p>\n\n\n\n
$\"\"$ hdr<\/figcaption><\/figure>\n\n\n\n
After the presentation, several questions were asked. The Q&As are listed as following: <\/p>\n\n\n\n
Q1: Does the Chinese and English have the same model? A1: No, the model is only for Chinese.<\/p>\n\n\n\n
Q2: Is the accuracy of this model not higher the basic model?
\nA2: Most of the model’s anwsers are right, and the answers are more natural and closer to our human’s expressions.<\/p>\n\n\n\n
Q3: This paper is about curriculum leaning, what’s the difference between it and general deep learning?
\nA3: Curriculum learning starts from easy instances and then gradually handles harder ones. In this paper the author uses instance scheduler, the instance contains target instances and common instance, and trains the model from common instances to target instances.<\/p>\n\n\n\n
Q4: Which model does NAG use?
\nA4: NAG uses gold topic entities to simplify the model.<\/p>\n","protected":false},"excerpt":{"rendered":"
NLPIR SEMINAR Y2019#1 INTRO\ufeff &nbsp … \u7ee7\u7eed\u9605\u8bfb →<\/span><\/a><\/p>\n","protected":false},"author":862,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[37,38],"tags":[],"_links":{"self":[{"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/posts\/6282"}],"collection":[{"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/users\/862"}],"replies":[{"embeddable":true,"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/comments?post=6282"}],"version-history":[{"count":6,"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/posts\/6282\/revisions"}],"predecessor-version":[{"id":6603,"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/posts\/6282\/revisions\/6603"}],"wp:attachment":[{"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/media?parent=6282"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/categories?post=6282"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.nlpir.org\/wordpress\/wp-json\/wp\/v2\/tags?post=6282"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}

NLPIR SEMINAR Y2019#1<\/h2>\n\n\n\n