{"id":4572,"date":"2015-12-11T09:00:12","date_gmt":"2015-12-11T17:00:12","guid":{"rendered":"https:\/\/blogs.msdn.microsoft.com\/msr_er\/?p=4572"},"modified":"2016-08-17T16:38:02","modified_gmt":"2016-08-17T23:38:02","slug":"convex-optimization-highlights-iccv-program-session","status":"publish","type":"post","link":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/blog\/convex-optimization-highlights-iccv-program-session\/","title":{"rendered":"Convex optimization highlights ICCV program session"},"content":{"rendered":"

By John Kaiser, Research News<\/em><\/p>\n

Convex optimization, prized for its efficiency and utility in solving small and medium-size problems across multiple disciplines, could soon be extended to handle much larger problems and tasks, such as those found in image and vision processing.<\/p>\n

\"Prof-Stephen-Boyd\"<\/a>Explaining how and why that might happen, Prof. Stephen Boyd of Stanford University gave the\u00a0plenary address Sunday at\u00a0the\u00a0biennial International Conference on Computer Vision (opens in new tab)<\/span><\/a> in Santiago Chile.<\/p>\n

To date, convex optimization used in many applications like AI, automatic control, and automated trading, starts to falter when the problem size exceeds 10,000 variables. That\u2019s sufficient to control an engine or inform hedge fund portfolios, but not nearly enough to handle the complex image recognition calculations required for autonomous vehicle navigation through busy urban areas.<\/p>\n

Boyd is a leading authority on convex optimization, a century-old branch of mathematics that was largely theoretical until the early 1990s. That\u2019s when faster computing speeds, and advances in algorithms for solving them, made it practical to use for applications like predicting peak power usage. Fast forward to today and convex optimization is routinely used across a variety of disciplines not only in computer science but also genetics, bioengineering, economics, behavioral sciences and various other fields.<\/p>\n

In the plenary talk and accompanying paper, Convex Optimization with Abstract Linear Operators<\/em> (opens in new tab)<\/span><\/a>, Boyd describes \u201crecent progress toward the goal of extending (domain specific languages or DSLs) to handle large-scale problems that involve linear operators given as abstract operators with fast transforms, such as those arising in image processing and vision, medical imaging, and other application areas. This involves re-thinking the entire stack, from the high-level DSL design down to the low level solvers.\u201d<\/p>\n

The research was conducted in collaboration with Steven Diamond, PhD candidate at Stanford.<\/p>\n

\"Example\"<\/a>

Example<\/p><\/div>\n

Although faster and cheaper CPUs will always be significant in charting the rise of any newly adopted technology, the use of DSLs may prompt wider adoption by enabling non-experts to articulate exactly what they want using a more natural language.<\/p>\n

\u201cDomain specific languages allow a broad variety of people to use (convex optimization) tools without knowing the details of how they work,\u201d Boyd said.<\/p>\n

It\u2019s the prospect of solving very difficult computational problems with just several lines of code that has far reaching, \u201cgame changing\u201d implications.<\/p>\n

Eventual goal: \u201cPeople who do images and work with 1 million variables can just write a 5-line Python script,\u201d Boyd said.<\/p>\n

Domain specific languages for convex optimization are easy to use, since they are \u201csimple declarative languages,\u201d where programmers declare what they want, but not how to achieve it.\u00a0 This is unlike the usual procedural languages where the programmers state what to do when something occurs.<\/p>\n

\u201cIn most quantitative hedge funds you don\u2019t have hardcoded logic that says what to sell or buy depending on how the market moves,\u201d Boyd explained. \u201cWhat they do instead is solve an optimization problem that takes everything into account and then decides what to trade.\u201d<\/p>\n

Boyd\u2019s PhD-level course on convex optimization is among the largest graduate courses at Stanford with more than 330 students (compared to a median graduate class size of 12). For access to open source code, papers, and related materials, see the Stephen P. Boyd site (opens in new tab)<\/span><\/a> at Stanford University.<\/p>\n

Sponsored by the Institute of Electrical and Electronics Engineers (opens in new tab)<\/span><\/a> (IEEE), ICCV runs from Dec. 11-18, 2015.<\/p>\n","protected":false},"excerpt":{"rendered":"

By John Kaiser, Research News Convex optimization, prized for its efficiency and utility in solving small and medium-size problems across multiple disciplines, could soon be extended to handle much larger problems and tasks, such as those found in image and vision processing. Explaining how and why that might happen, Prof. Stephen Boyd of Stanford University […]<\/p>\n","protected":false},"author":32627,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":[],"msr_hide_image_in_river":0,"footnotes":""},"categories":[194471,194483,194488,194459],"tags":[195175,195871,195929,195953,197299],"research-area":[13562,13546,13560],"msr-region":[],"msr-event-type":[],"msr-locale":[268875],"msr-post-option":[],"msr-impact-theme":[],"msr-promo-type":[],"msr-podcast-series":[],"class_list":["post-4572","post","type-post","status-publish","format-standard","hentry","category-computer-vision","category-mathematics","category-program-languages-and-software-engineering","category-researchnews","tag-convex-optimization","tag-ieee","tag-institute-of-electrical-and-electronics-engineers","tag-international-conference-on-computer-vision","tag-stanford-university","msr-research-area-computer-vision","msr-research-area-computational-sciences-mathematics","msr-research-area-programming-languages-software-engineering","msr-locale-en_us"],"msr_event_details":{"start":"","end":"","location":""},"podcast_url":"","podcast_episode":"","msr_research_lab":[],"msr_impact_theme":[],"related-publications":[],"related-downloads":[],"related-videos":[],"related-academic-programs":[],"related-groups":[],"related-projects":[],"related-events":[],"related-researchers":[],"msr_type":"Post","byline":"","formattedDate":"December 11, 2015","formattedExcerpt":"By John Kaiser, Research News Convex optimization, prized for its efficiency and utility in solving small and medium-size problems across multiple disciplines, could soon be extended to handle much larger problems and tasks, such as those found in image and vision processing. Explaining how and…","locale":{"slug":"en_us","name":"English","native":"","english":"English"},"_links":{"self":[{"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/posts\/4572","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/users\/32627"}],"replies":[{"embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/comments?post=4572"}],"version-history":[{"count":4,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/posts\/4572\/revisions"}],"predecessor-version":[{"id":278574,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/posts\/4572\/revisions\/278574"}],"wp:attachment":[{"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/media?parent=4572"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/categories?post=4572"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/tags?post=4572"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=4572"},{"taxonomy":"msr-region","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-region?post=4572"},{"taxonomy":"msr-event-type","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-event-type?post=4572"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=4572"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=4572"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=4572"},{"taxonomy":"msr-promo-type","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-promo-type?post=4572"},{"taxonomy":"msr-podcast-series","embeddable":true,"href":"https:\/\/new-cm-edgedigital.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-podcast-series?post=4572"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}