{"id":34491,"date":"2019-10-31T21:58:39","date_gmt":"2019-10-31T18:58:39","guid":{"rendered":"https:\/\/prohoster.info\/blog\/intel-rabotaet-nad-opticheskimi-chipami-dlya-bolee-effektivnogo-ii\/"},"modified":"2019-10-31T21:58:39","modified_gmt":"2019-10-31T18:58:39","slug":"intel-rabotaet-nad-opticheskimi-chipami-dlya-bolee-effektivnogo-ii","status":"publish","type":"post","link":"https:\/\/prohoster.info\/da\/blog\/news\/intel-rabotaet-nad-opticheskimi-chipami-dlya-bolee-effektivnogo-ii","title":{"rendered":"Intel arbejder p\u00e5 optiske chips til mere effektiv AI.","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Photoniske integrerede kredsl\u00f8b eller optiske chips har potensialet til at tilbyde mange fordele sammenlignet med deres elektroniske modparter, s\u00e5som reduceret energiforbrug og lavere latens i beregninger. Det er derfor mange forskere mener, at de kan v\u00e6re ekstremt effektive i opgaver inden for maskinl\u00e6ring og kunstig intelligens (AI). Intel ser ogs\u00e5 store muligheder for anvendelse af siliciumfotonik i denne retning. Gruppens forskere i en videnskabelig artikel<\/a><\/noindex> har detaljeret beskrevet nye metoder, der kan bringe optiske neurale netv\u00e6rk et skridt t\u00e6ttere p\u00e5 virkeligheden.<\/p>\n

\"Intel<\/p>\n

I en nylig blogindl\u00e6g fra Intel<\/a><\/noindex>, der handler om maskinl\u00e6ring, beskriver, hvordan forskningen inden for optiske neurale netv\u00e6rk startede. Forskning udf\u00f8rt af David A. B. Miller og Michael Reck viste, at en type fotonisk kredsl\u00f8b kendt som Mach-Zehnder interferometer (MZI) kan konfigureres til at udf\u00f8re matrixmultiplikation af st\u00f8rrelsen 2 \u00d7 2, og hvis man placerer MZI i et trekantsnet til multiplikation af st\u00f8rre matricer, kan man opn\u00e5 et kredsl\u00f8b, der implementerer algoritmen til matrix gange vektor \u2014 den grundl\u00e6ggende beregning, der anvendes i maskinl\u00e6ring.<\/p>\n

En ny unders\u00f8gelse fra Intel fokuserede p\u00e5 at unders\u00f8ge, hvad der sker, n\u00e5r forskellige defekter opst\u00e5r i optiske chips under produktionen (da beregningsfotonik er analog af natur), hvilket skaber forskelle i beregningens n\u00f8jagtighed mellem forskellige chips af samme type. Selvom lignende unders\u00f8gelser allerede er blevet udf\u00f8rt, har de tidligere v\u00e6ret mere fokuseret p\u00e5 post-fremstilling optimering for at rette eventuelle un\u00f8jagtigheder. Men denne tilgang har d\u00e5rlig skalerbarhed, da netv\u00e6rkene bliver st\u00f8rre i st\u00f8rrelse, hvilket f\u00f8rer til en stigning i den beregningskraft, der kr\u00e6ves for at justere optiske netv\u00e6rk. I stedet for at optimere efter fremstillingen overvejede Intel muligheden for engangs tr\u00e6ning af chips f\u00f8r fremstilling ved hj\u00e6lp af en st\u00f8jresistent arkitektur. En standard optisk neuralt netv\u00e6rk blev tr\u00e6net \u00e9n gang, hvorefter l\u00e6ringsparametrene blev fordelt over flere fremstillede eksemplarer af netv\u00e6rket med forskelle i deres komponenter.<\/p>\n

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

Intel-teamet har overvejet to arkitekturer til opbygning af kunstig intelligens-systemer baseret p\u00e5 MZI: GridNet og FFTNet. GridNet placerer forudsigeligt MZI i et netv\u00e6rk, mens FFTNet placerer dem i form af 'sommerfugle'. Efter tr\u00e6ning af begge i simuleringen af en benchmark-opgave for dyb l\u00e6ring til genkendelse af h\u00e5ndskrevne cifre (MNIST), fandt forskerne, at GridNet opn\u00e5ede h\u00f8jere n\u00f8jagtighed end FFTNet (98 % mod 95 %), men at FFTNet-arkitekturen viste sig at v\u00e6re 'betydeligt mere robust'. Faktisk faldt GridNets ydeevne under 50 % med tilf\u00f8jelsen af kunstigt st\u00f8j (forstyrrelser, der imiterer mulige defekter ved fremstillingen af optiske chips), mens FFTNet n\u00e6rmest forblev konstant.<\/p>\n

Forskere h\u00e6vder, at deres forskning l\u00e6gger grunden for metoder til tr\u00e6ning af kunstig intelligens, der vil hj\u00e6lpe med at undg\u00e5 behovet for finjustering af optiske chips efter deres produktion, hvilket sparer v\u00e6rdifuld tid og ressourcer.<\/p>\n

\u2019Som i enhver produktionsproces opst\u00e5r der visse defekter, hvilket betyder, at der vil v\u00e6re sm\u00e5 variationer mellem chipsene, og de vil p\u00e5virke n\u00f8jagtigheden af beregningerne,\u2019 skriver senior direkt\u00f8r for Intel AIs produktgruppe, Casimir Wierzynski. \u2019Hvis optiske neurale netv\u00e6rk skal blive en levedygtig del af den hardware\u00f8kosystem for kunstig intelligens, vil de have brug for at bev\u00e6ge sig mod st\u00f8rre chips og industrielle produktionsmetoder. Vores forskning viser, at valget af den rigtige arkitektur p\u00e5 forh\u00e5nd kan \u00f8ge chancerne for, at de opn\u00e5ede chips n\u00e5r den \u00f8nskede ydeevne, selv i tilf\u00e6lde af produktionsvariationer.\u2019<\/p>\n

Samtidig, mens Intel prim\u00e6rt udf\u00f8rer forskning, har ph.d. kandidat fra Massachusetts Institute of Technology, Yichen Shen, grundlagt startup'en Lightelligence, som er baseret i Boston og har tiltrukket venturekapital p\u00e5 10,7 millioner US dollars og for nylig demonstreret<\/a><\/noindex> en prototype af en optisk chip til maskinl\u00e6ring, som er 100 gange hurtigere end nuv\u00e6rende elektroniske chips samt signifikant reducerer energiforbruget, hvilket endnu engang tydeligt demonstrerer potentialet i fotoniske teknologier.<\/p>\n

<\/p>\n

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

\t\t\t\t


\n\t\t\t\t\t\t\t\t<\/center>
Kilde: 3dnews.ru<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"

\u0424\u043e\u0442\u043e\u043d\u043d\u044b\u0435 \u0438\u043d\u0442\u0435\u0433\u0440\u0430\u043b\u044c\u043d\u044b\u0435 \u0441\u0445\u0435\u043c\u044b \u0438\u043b\u0438 \u043e\u043f\u0442\u0438\u0447\u0435\u0441\u043a\u0438\u0435 \u0447\u0438\u043f\u044b \u043f\u043e\u0442\u0435\u043d\u0446\u0438\u0430\u043b\u044c\u043d\u043e \u043e\u0431\u043b\u0430\u0434\u0430\u044e\u0442 \u043c\u043d\u043e\u0436\u0435\u0441\u0442\u0432\u043e\u043c \u043f\u0440\u0435\u0438\u043c\u0443\u0449\u0435\u0441\u0442\u0432 \u043f\u043e \u0441\u0440\u0430\u0432\u043d\u0435\u043d\u0438\u044e \u0441 \u044d\u043b\u0435\u043a\u0442\u0440\u043e\u043d\u043d\u044b\u043c\u0438 \u0430\u043d\u0430\u043b\u043e\u0433\u0430\u043c\u0438, \u0442\u0430\u043a\u0438\u043c\u0438 \u043a\u0430\u043a \u0441\u043d\u0438\u0436\u0435\u043d\u0438\u0435 \u044d\u043d\u0435\u0440\u0433\u043e\u043f\u043e\u0442\u0440\u0435\u0431\u043b\u0435\u043d\u0438\u044f \u0438 \u0443\u043c\u0435\u043d\u044c\u0448\u0435\u043d\u0438\u0435 \u0437\u0430\u0434\u0435\u0440\u0436\u043a\u0438 \u0432 \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u044f\u0445. \u0412\u043e\u0442 \u043f\u043e\u0447\u0435\u043c\u0443 \u043c\u043d\u043e\u0436\u0435\u0441\u0442\u0432\u043e \u0438\u0441\u0441\u043b\u0435\u0434\u043e\u0432\u0430\u0442\u0435\u043b\u0435\u0439 \u0441\u0447\u0438\u0442\u0430\u044e\u0442, \u0447\u0442\u043e \u043e\u043d\u0438 \u043c\u043e\u0433\u0443\u0442 \u043e\u043a\u0430\u0437\u0430\u0442\u044c\u0441\u044f \u043a\u0440\u0430\u0439\u043d\u0435 \u044d\u0444\u0444\u0435\u043a\u0442\u0438\u0432\u043d\u044b \u0432 \u0437\u0430\u0434\u0430\u0447\u0430\u0445 \u043c\u0430\u0448\u0438\u043d\u043d\u043e\u0433\u043e \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u044f \u0438 \u0441\u043e\u0437\u0434\u0430\u043d\u0438\u044f \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u043e\u0433\u043e \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442\u0430 (\u0418\u0418). \u041a\u043e\u043c\u043f\u0430\u043d\u0438\u044f Intel \u0442\u0430\u043a\u0436\u0435 \u0432\u0438\u0434\u0438\u0442 \u0431\u043e\u043b\u044c\u0448\u0438\u0435 \u043f\u0435\u0440\u0441\u043f\u0435\u043a\u0442\u0438\u0432\u044b \u0434\u043b\u044f \u043f\u0440\u0438\u043c\u0435\u043d\u0435\u043d\u0438\u044f \u043a\u0440\u0435\u043c\u043d\u0438\u0435\u0432\u043e\u0439 \u0444\u043e\u0442\u043e\u043d\u0438\u043a\u0438 \u0432 […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":25984,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-34491","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\n","aioseo_head_json":{"title":"\ud83e\udd47Intel arbejder p\u00e5 optiske chips for en mere effektiv AI | ProHoster","description":"Fotoniske integrerede kredsl\u00f8b eller optiske chips har potentielt mange fordele sammenlignet med deres elektroniske modstykker, s\u00e5som reduceret energiforbrug og mindsket latenstid i beregninger.","canonical_url":"https:\/\/prohoster.info\/da\/blog\/news\/intel-rabotaet-nad-opticheskimi-chipami-dlya-bolee-effektivnogo-ii","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"da_DK","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd47Intel \u0440\u0430\u0431\u043e\u0442\u0430\u0435\u0442 \u043d\u0430\u0434 \u043e\u043f\u0442\u0438\u0447\u0435\u0441\u043a\u0438\u043c\u0438 \u0447\u0438\u043f\u0430\u043c\u0438 \u0434\u043b\u044f \u0431\u043e\u043b\u0435\u0435 \u044d\u0444\u0444\u0435\u043a\u0442\u0438\u0432\u043d\u043e\u0433\u043e \u0418\u0418 | ProHoster","og:description":"\u0424\u043e\u0442\u043e\u043d\u043d\u044b\u0435 \u0438\u043d\u0442\u0435\u0433\u0440\u0430\u043b\u044c\u043d\u044b\u0435 \u0441\u0445\u0435\u043c\u044b \u0438\u043b\u0438 \u043e\u043f\u0442\u0438\u0447\u0435\u0441\u043a\u0438\u0435 \u0447\u0438\u043f\u044b \u043f\u043e\u0442\u0435\u043d\u0446\u0438\u0430\u043b\u044c\u043d\u043e \u043e\u0431\u043b\u0430\u0434\u0430\u044e\u0442 \u043c\u043d\u043e\u0436\u0435\u0441\u0442\u0432\u043e\u043c \u043f\u0440\u0435\u0438\u043c\u0443\u0449\u0435\u0441\u0442\u0432 \u043f\u043e \u0441\u0440\u0430\u0432\u043d\u0435\u043d\u0438\u044e \u0441 \u044d\u043b\u0435\u043a\u0442\u0440\u043e\u043d\u043d\u044b\u043c\u0438 \u0430\u043d\u0430\u043b\u043e\u0433\u0430\u043c\u0438, \u0442\u0430\u043a\u0438\u043c\u0438 \u043a\u0430\u043a \u0441\u043d\u0438\u0436\u0435\u043d\u0438\u0435 \u044d\u043d\u0435\u0440\u0433\u043e\u043f\u043e\u0442\u0440\u0435\u0431\u043b\u0435\u043d\u0438\u044f \u0438 \u0443\u043c\u0435\u043d\u044c\u0448\u0435\u043d\u0438\u0435 \u0437\u0430\u0434\u0435\u0440\u0436\u043a\u0438 \u0432 \u0432\u044b\u0447\u0438\u0441\u043b\u0435\u043d\u0438\u044f\u0445.","og:url":"https:\/\/prohoster.info\/da\/blog\/news\/intel-rabotaet-nad-opticheskimi-chipami-dlya-bolee-effektivnogo-ii","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2019-10-31T18:58:39+00:00","article:modified_time":"2019-10-31T18:58:39+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"34491","title":null,"description":null,"keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":null,"og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"","isEnabled":true},"graphs":[]},"schema_type":null,"schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":null,"robots_max_videopreview":null,"robots_max_imagepreview":"large","priority":null,"frequency":null,"local_seo":null,"seo_analyzer_scan_date":"2026-01-21 19:28:19","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-03-01 02:20:55","updated":"2026-01-21 19:28:19","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/posts\/34491","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/comments?post=34491"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/posts\/34491\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/media\/25984"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/media?parent=34491"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/categories?post=34491"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/da\/wp-json\/wp\/v2\/tags?post=34491"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}