{"469831":{"#nid":"469831","#data":{"type":"event","title":"Ph.D. Dissertation Defense - I-Fan Chen","body":[{"value":"\u003Cp\u003E\u003Cstrong\u003ETitle\u003C\/strong\u003E\u003Cem\u003E:\u0026nbsp; \u003C\/em\u003E\u003Cem\u003EResource-dependent Acoustic and Language Modeling for Spoken Keyword Search\u003C\/em\u003E\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003ECommittee:\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003EDr. C.-H. Lee , Advisor\u003C\/p\u003E\u003Cp\u003EDr. Biing-Hwang Juang, ECE\u003C\/p\u003E\u003Cp\u003EDr. Mark Clements, ECE\u003C\/p\u003E\u003Cp\u003EDr. Gee-Kung Chang, ECE\u003C\/p\u003E\u003Cp\u003EDr. Yao Xie, ISyE\u003C\/p\u003E\u003Cp\u003E\u003Cstrong\u003EAbstract:\u0026nbsp;\u003C\/strong\u003E\u003C\/p\u003E\u003Cp\u003E\u0026nbsp;\u003C\/p\u003E\u003Cp\u003EIn this dissertation, three research directions were explored to alleviate two major issues, i.e., the use of incorrect models and training\/test condition mismatches, in the modeling frameworks of modern spoken keyword search (KWS) systems. Each of the three research directions, which include (i) data-efficient training processes, (ii) system optimization objectives, and (iii) data augmentation, utilizes different types and amounts of training resources in different ways to ameliorate the two issues of acoustic and language modeling in modern KWS systems. To be more specific, \u003Cem\u003Eresource-dependent keyword modeling\u003C\/em\u003E, \u003Cem\u003Ekeyword-boosted sMBR (state-level minimum Bayes risk) training\u003C\/em\u003E, and \u003Cem\u003Emultilingual acoustic modeling\u003C\/em\u003E are proposed and investigated for acoustic modeling in this research. For language modeling, \u003Cem\u003Ekeyword-aware language modeling\u003C\/em\u003E, \u003Cem\u003Ediscriminative keyword-aware language modeling,\u003C\/em\u003E and \u003Cem\u003Eweb text augmented language modeling\u003C\/em\u003E are presented and discussed.\u003C\/p\u003E\u003Cp\u003EThe dissertation provides a comprehensive collection of solutions and strategies to the acoustic and language modeling problems in KWS. It also offers insights into the realization of good-performance KWS systems. Experimental results show that the data-efficient training process and data augmentation are the two directions providing the most prominent performance improvement for KWS systems. While modifying system optimization objectives provides smaller yet consistent performance enhancement in KWS systems with different configurations. The effects of the proposed acoustic and language modeling approaches in the three directions are also shown to be additive and can be combined to further improve the overall KWS system performance.\u003C\/p\u003E","summary":null,"format":"limited_html"}],"field_subtitle":"","field_summary":"","field_summary_sentence":[{"value":"ECE PhD Dissertation Defense"}],"uid":"28475","created_gmt":"2015-11-15 13:03:29","changed_gmt":"2016-10-08 02:14:50","author":"Daniela Staiculescu","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2015-11-17T14:00:00-05:00","event_time_end":"2015-11-17T14:00:00-05:00","event_time_end_last":"2015-11-17T14:00:00-05:00","gmt_time_start":"2015-11-17 19:00:00","gmt_time_end":"2015-11-17 19:00:00","gmt_time_end_last":"2015-11-17 19:00:00","rrule":null,"timezone":"America\/New_York"},"extras":[],"groups":[{"id":"434381","name":"ECE Ph.D. Dissertation Defenses"}],"categories":[],"keywords":[{"id":"1808","name":"graduate students"},{"id":"100811","name":"Phd Defense"}],"core_research_areas":[],"news_room_topics":[],"event_categories":[{"id":"1788","name":"Other\/Miscellaneous"}],"invited_audience":[{"id":"78771","name":"Public"}],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}