{"id":2201,"date":"2019-02-23T14:06:39","date_gmt":"2019-02-23T14:06:39","guid":{"rendered":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/?p=2201"},"modified":"2019-02-23T14:06:39","modified_gmt":"2019-02-23T14:06:39","slug":"www-2019-paper-on-diversity-of-dialogue-response-generation-online","status":"publish","type":"post","link":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/www-2019-paper-on-diversity-of-dialogue-response-generation-online\/","title":{"rendered":"WWW 2019 paper on diversity of dialogue response generation online"},"content":{"rendered":"\n<p><em>Improving Neural Response Diversity with Frequency-Aware Cross-Entropy Loss<\/em> by Shaojie Jiang, Pengjie Ren, Christof Monz, and\u00a0Maarten de Rijke is online now at <a href=\"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-content\/papercite-data\/pdf\/jiang-2019-improving.pdf\">this location<\/a>.<\/p>\n\n\n\n<p>Sequence-to-Sequence (Seq2Seq) models have achieved encouraging performance on the dialogue response generation task. However, existing Seq2Seq-based response generation methods suffer from a\u00a0low-diversity problem: they frequently generate generic responses, which make the conversation less interesting. In this paper, we address the low-diversity problem by investigating its connection with\u00a0model over-confidence\u00a0reflected in predicted distributions. Specifically, we first analyze the influence of the commonly used Cross-Entropy (CE) loss function, and find that the CE loss function prefers high-frequency tokens, which results in low-diversity responses. We then propose a Frequency-Aware Cross-Entropy (FACE) loss function that improves over the CE loss function by incorporating a weighting mechanism conditioned on token frequency. Extensive experiments on benchmark datasets show that the FACE loss function is able to substantially improve the diversity of existing state-of-the-art Seq2Seq response generation methods, in terms of both automatic and human evaluations.<\/p>\n\n\n\n<p>The paper will be presented at The Web Conference 2019.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Improving Neural Response Diversity with Frequency-Aware Cross-Entropy Loss by Shaojie Jiang, Pengjie Ren, Christof Monz, and\u00a0Maarten de Rijke is online now at this location. Sequence-to-Sequence (Seq2Seq) models have achieved encouraging performance on the dialogue response generation task. However, existing Seq2Seq-based response generation methods suffer from a\u00a0low-diversity problem: they frequently generate generic responses, which make the&#8230;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"_links":{"self":[{"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/posts\/2201"}],"collection":[{"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/comments?post=2201"}],"version-history":[{"count":1,"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/posts\/2201\/revisions"}],"predecessor-version":[{"id":2202,"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/posts\/2201\/revisions\/2202"}],"wp:attachment":[{"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/media?parent=2201"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/categories?post=2201"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/staff.fnwi.uva.nl\/m.derijke\/wp-json\/wp\/v2\/tags?post=2201"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}