{"612961":{"#nid":"612961","#data":{"type":"news","title":"DCL Seminar Series: Anthony Yezzi","body":[{"value":"\u003Cp\u003EAbstract:\u003C\/p\u003E\r\n\r\n\u003Cp\u003EFollowing the seminal work of Nesterov, accelerated optimization methods\u0026nbsp;\u003Cbr \/\u003E\r\n(sometimes referred to as momentum methods) have been used to powerfully\u0026nbsp;\u003Cbr \/\u003E\r\nboost the performance of first-order, gradient-based parameter\u0026nbsp;\u003Cbr \/\u003E\r\nestimation in scenarios were second-order optimization strategies are\u0026nbsp;\u003Cbr \/\u003E\r\neither inapplicable or impractical. Not only does accelerated gradient\u0026nbsp;\u003Cbr \/\u003E\r\ndescent converge considerably faster than traditional gradient descent,\u0026nbsp;\u003Cbr \/\u003E\r\nbut it performs a more robust local search of the parameter space by\u0026nbsp;\u003Cbr \/\u003E\r\ninitially overshooting and then oscillating back as it settles into a\u0026nbsp;\u003Cbr \/\u003E\r\nfinal configuration, thereby selecting only local minimizers with a\u0026nbsp;\u003Cbr \/\u003E\r\nattraction basin large enough to accommodate the initial overshoot. This\u0026nbsp;\u003Cbr \/\u003E\r\nbehavior has made accelerated search methods particularly popular within\u0026nbsp;\u003Cbr \/\u003E\r\nthe machine learning community where stochastic variants have been\u0026nbsp;\u003Cbr \/\u003E\r\nproposed as well.\u0026nbsp; So far, however, accelerated optimization methods\u0026nbsp;\u003Cbr \/\u003E\r\nhave been applied to searches over finite parameter spaces. We show how\u0026nbsp;\u003Cbr \/\u003E\r\na variational framework for these finite dimensional methods (recently\u0026nbsp;\u003Cbr \/\u003E\r\nformulated by Wibisono, Wilson, and Jordan) can be extended to the\u0026nbsp;\u003Cbr \/\u003E\r\ninfinite dimensional setting and, in particular, to the manifold of\u0026nbsp;\u003Cbr \/\u003E\r\nplanar curves in order to yield a new class of accelerated geometric,\u0026nbsp;\u003Cbr \/\u003E\r\nPDE-based active contours.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EBio:\u003C\/p\u003E\r\n\r\n\u003Cp\u003EProfessor Yezzi\u0026nbsp;obtained both his Bachelor\u0026#39;s degree and his Ph.D. in the Department of Electrical Engineering at the University of Minnesota with minors in mathematics and music.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EAfter completing his Ph.D., he continued his research as a post-Doctoral Research Associate at the Laboratory for Information and Decision Systems at Massachusetts Institute of Technology in Boston, MA.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EHis research interests fall broadly within the fields of image processing and computer vision. In particular he is interested in curve and surface evolution theory and partial differential equation techniques as they apply to topics within these fields.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EMuch of Dr. Yezzi\u0026#39;s work is particularly tailored to problems in medical imaging, including cardiac ultrasound, MRI, and CT. He joined the Georgia Tech faculty in the fall of 1999 where he has taught courses in DSP and is working to develop advanced courses in computer vision and medical image processing.\u0026nbsp;\u003C\/p\u003E\r\n","summary":null,"format":"limited_html"}],"field_subtitle":[{"value":"Accelerated Optimization in the PDE Framework"}],"field_summary":[{"value":"\u003Cp\u003EAccelerated Optimization in the PDE Framework\u003C\/p\u003E\r\n","format":"limited_html"}],"field_summary_sentence":[{"value":"Accelerated Optimization in the PDE Framework"}],"uid":"34840","created_gmt":"2018-10-18 16:57:55","changed_gmt":"2018-10-18 16:57:55","author":"mamstutz3","boilerplate_text":"","field_publication":"","field_article_url":"","dateline":{"date":"2018-03-16T00:00:00-04:00","iso_date":"2018-03-16T00:00:00-04:00","tz":"America\/New_York"},"extras":[],"hg_media":{"612960":{"id":"612960","type":"image","title":"Yezzi","body":null,"created":"1539881434","gmt_created":"2018-10-18 16:50:34","changed":"1539881434","gmt_changed":"2018-10-18 16:50:34","alt":"","file":{"fid":"233357","name":"tony_yezzi_2.jpg","image_path":"\/sites\/default\/files\/images\/tony_yezzi_2_1.jpg","image_full_path":"http:\/\/tlwarc.hg.gatech.edu\/\/sites\/default\/files\/images\/tony_yezzi_2_1.jpg","mime":"image\/jpeg","size":71135,"path_740":"http:\/\/tlwarc.hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/images\/tony_yezzi_2_1.jpg?itok=a0SbBndh"}}},"media_ids":["612960"],"groups":[{"id":"66191","name":"Decision and Control Lab (DCL)"}],"categories":[],"keywords":[{"id":"179403","name":"DCL Seminar Series"}],"core_research_areas":[],"news_room_topics":[],"event_categories":[],"invited_audience":[],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}