{"id":30936,"date":"2019-10-31T21:38:20","date_gmt":"2019-10-31T18:38:20","guid":{"rendered":"https:\/\/prohoster.info\/blog\/mashinnoe-obuchenie-bez-python-anaconda-i-prochih-presmykayushhihsya\/"},"modified":"2019-10-31T21:38:20","modified_gmt":"2019-10-31T18:38:20","slug":"mashinnoe-obuchenie-bez-python-anaconda-i-prochih-presmykayushhihsya","status":"publish","type":"post","link":"https:\/\/prohoster.info\/hr\/blog\/news\/mashinnoe-obuchenie-bez-python-anaconda-i-prochih-presmykayushhihsya","title":{"rendered":"Strojno u\u010denje bez Pythona, Anaconde i drugih gmazova","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Ne, naravno da ne mislim ozbiljno. Mora postojati granica do koje je mogu\u0107e pojednostaviti temu. Ali za prve korake, razumevanje osnovnih koncepcija i brzo 'uklju\u010divanje' u temu, mo\u017eda je i prihvatljivo. A kako ispravno imenovati ovaj materijal (mogu\u0107nosti: 'Ma\u0161insko u\u010denje za po\u010detnike', 'Analiza podataka od nule', 'Algoritmi za najmla\u0111e'), razgovara\u0107emo na kraju. <\/p>\n

Pre\u0111imo na stvar. Napisao sam nekoliko prakti\u010dnih programa u MS Excelu za vizualizaciju i slikovito predstavljanje procesa koji se odvijaju u razli\u010ditim metodama ma\u0161inskog u\u010denja prilikom analize podataka. Videti je verovati, na kraju krajeva, kako ka\u017eu pripadnici kulture koja je razvila ve\u0107inu ovih metoda (usput, nije sve. Mo\u0107na 'metoda potpornih vektora', ili SVM, support vector machine \u2013 izum je na\u0161eg sugra\u0111anina Vladimira Vapnika, Moskovski institut za upravljanje. Godina 1963, uzgred! Sada, me\u0111utim, predaje i radi u SAD-u).<\/p>\n

Tri datoteke na uvid
\n<\/a><\/noindex><\/p>\n

1. Klasifikacija metodom k-srednjih<\/h2>\n

\nOve vrste zadataka pripadaju 'u\u010denju bez nadzora', kada treba da podelimo ulazne podatke na unapred poznat broj kategorija, ali pritom nemamo nijednu koli\u010dinu 'ta\u010dnih odgovora', njih moramo izvu\u0107i iz samih podataka. Fundamentalni klasi\u010dni zadatak pronala\u017eenja podvrsta cvetova irisa (Ronald Fisher, 1936!) koji se smatra prvim korakom u ovoj oblasti znanja \u2013 upravo je ovakve prirode.<\/p>\n

Metoda je dovoljno jednostavna. Imamo skup objekata predstavljenih u obliku vektora (setova N brojeva). Za iris, to su setovi od 4 broja koji karakterizuju cvet: du\u017eina i \u0161irina spolja\u0161njeg i unutra\u0161njeg lista oblatnika, redom (Iris Fisher \u2014 Wikipedija<\/a><\/noindex>). Kao mera udaljenosti ili bliskosti izme\u0111u objekata koristi se obi\u010dna dekartova metrika.<\/p>\n

Zatim se na proizvoljan na\u010din (ili neproizvoljan, pogledajte dalje) biraju centri klastera i izra\u010dunavaju se udaljenosti od svakog objekta do centara klastera. Svaki objekat na ovom koraku iteracije ozna\u010dava se kao pripadaju\u0107i najbli\u017eem centru. Zatim se centar svakog klastera premesti na aritmeti\u010dku sredinu koordinata svojih \u010dlanova (po analogiji s fizikom, ovo se jo\u0161 naziva 'centar mase'), a procedura se ponavlja.<\/p>\n

Proces brzo konvergira. Na slikama u dva dimenzije to izgleda ovako:<\/p>\n

1. Po\u010detna slu\u010dajna raspodjela to\u010daka na ravnini i broj klastera<\/p>\n

\"Strojno
\n
\n2. Postavljanje centara klastera i pripisivanje to\u010daka svojim klasterima<\/p>\n

\"Strojno
\n
\n3. Pomicanje koordinata centara klastera, ponovni izra\u010dun pripadnosti to\u010daka, sve dok centri ne postanu stabilni. Vidljiva je putanja kretanja centra klastera prema kona\u010dnom polo\u017eaju.<\/p>\n

\"Strojno
\n
\nU bilo kojem trenutku mo\u017eete postaviti nove centre klastera (ne generiraju\u0107i novu raspodjelu to\u010daka!) i vidjeti da proces podjele ne mora uvijek biti jednozna\u010dan. Matemat\u010dki to zna\u010di da u optimiziranoj funkciji (sumi kvadrata udaljenosti to\u010daka do centara svojih klastera) nalazimo ne globalni, ve\u0107 lokalni minimum. Ovaj problem mo\u017eemo rije\u0161iti ili ne slu\u010dajnim odabirom po\u010detnih centara klastera, ili isprobavanjem mogu\u0107ih centara (ponekad ih je korisno postaviti to\u010dno na neku od to\u010daka, jer bar postoji jamstvo da ne\u0107emo dobiti prazne klastere). U svakom slu\u010daju, kona\u010dna skupina uvek ima to\u010dnu donju granicu. <\/p>\n

Mo\u017eete se igrati s ovom datotekom putem ove poveznice<\/a><\/noindex> (ne zaboravite uklju\u010diti podr\u0161ku makronaredbama. Datoteke su provjerene na viruse)<\/p>\n

Opis metode na Wikipediji \u2014 Metoda k-srednjih<\/a><\/noindex><\/p>\n

2. Aproksimacija polinomima i podjela podataka. Preu\u010denje<\/h2>\n

\nIzvanredni znanstvenik i popularizator znanosti o podacima K.V. Voroncov kratko govori o metodama strojnog u\u010denja kao o \"znanosti o provo\u0111enju krivulja kroz to\u010dke\". U ovom primjeru \u0107emo prona\u0107i zakonitost u podacima metodom najmanjih kvadrata. <\/p>\n

Prikazana je tehnika podjele izvornih podataka na \"obu\u010dene\" i \"kontrolne\", kao i fenomen preu\u010denje, ili \"prilagodba\" na podatke. Uz pravilnu aproksimaciju imat \u0107emo odre\u0111enu pogre\u0161ku na obu\u010dnim podacima i ne\u0161to ve\u0107u \u2013 na kontrolnim. Uz nepravilnu \u2013 to\u010dnu prilagodbu na obu\u010dene podatke i ogroman pogre\u0161ku na kontrolnim.<\/p>\n

(Poznata \u010dinjenica je da kroz N to\u010daka mo\u017ee se povu\u0107i jedinstvena krivulja N-1. stupnja, i ovaj na\u010din u op\u0107em slu\u010daju ne daje \u017eeljeni rezultat. Interpolacijski polinom Lagrangea na Wikipediji<\/a><\/noindex>)<\/p>\n

1. Postavljamo inicijalnu raspodjelu<\/p>\n

\"Strojno
\n
\n2. Dijelimo to\u010dke na \"obu\u010dene\" i \"kontrolne\" u omjeru 70 prema 30.<\/p>\n

\"Strojno
\n
\n3. Provodimo aproksimacijsku krivulju prema to\u010dkama za obuku, vidimo pogre\u0161ku koju ona daje na kontrolnim podacima.<\/p>\n

\"Strojno
\n
\n4. Provodimo to\u010dnu krivulju kroz to\u010dke za obuku i vidimo ogromnu pogre\u0161ku na kontrolnim podacima (i nultu na obu\u010davaju\u0107im, ali kakva je to korist?).<\/p>\n

\"Strojno
\n
\nPoka\u017ee se, naravno, najjednostavnija varijanta s jednim razdvajanjem na \"obu\u010davaju\u0107e\" i \"kontrolne\" podskupine; u op\u0107em slu\u010daju to se radi vi\u0161ekratno radi bolje prilagodbe koeficijenata.<\/p>\n

Datoteka je dostupna ovdje, provjerena antivirusom.<\/a><\/noindex> Omogu\u0107ite makronaredbe za ispravan rad.<\/p>\n

3. Gradientni silazak i dinamika promjene pogre\u0161ke.<\/h2>\n

\nOvdje \u0107e biti 4-dimenzionalni slu\u010daj i linearna regresija. Koeficijenti linearne regresije odre\u0111ivat \u0107e se koracima metodom gradientnog silaska, isprva su svi koeficijenti \u2013 nule. Na zasebnom grafikonu vidi se dinamika smanjenja pogre\u0161ke s obzirom na sve precizniju prilagodbu koeficijenata. Postoji mogu\u0107nost da se pogleda sve \u010detiri 2-dimenzionalne projekcije.<\/p>\n

Ako se postavi prevelik korak gradientnog silaska, vidi se da \u0107emo svaki put presko\u010diti minimum i do rezultata \u0107emo do\u0107i za ve\u0107i broj koraka, iako \u0107emo, na kraju, svejedno do\u0107i (samo ako ne pove\u0107amo korak silaska previ\u0161e \u2014 tada \u0107e algoritam oti\u0107i \"izvan kontrole\"). I graph ovisnosti pogre\u0161ke o koraku iteracije ne\u0107e biti glatki, nego \"grub\".<\/p>\n

1. Generiramo podatke, postavljamo korak gradientnog silaska.<\/p>\n

\"Strojno
\n
\n2. Uz pravilno odabran korak gradientnog silaska glatko i dovoljno brzo dolazimo do minimuma.<\/p>\n

\"Strojno
\n
\n3. Uz pogre\u0161no odabran korak gradientnog silaska proma\u0161ujemo maksimum, graf pogre\u0161ke je \"grub\", a konvergencija traje vi\u0161e koraka.<\/p>\n

\"Strojno
\ni<\/p>\n

\"Strojno
\n
\n4. Uz potpuno pogre\u0161no odabran korak gradientnog silaska udaljavamo se od minimuma.<\/p>\n

\"Strojno
\n
\n(Da biste reproducirali proces pri prikazanim vrijednostima koraka gradientnog silaska, ozna\u010dite \"referentne podatke\").<\/p>\n

Datoteka \u2013 putem ove poveznice, treba uklju\u010diti makronaredbe, virusi nisu prisutni.<\/a><\/noindex><\/p>\n

Kako smatra uva\u017eena zajednica, je li dopu\u0161teno takvo pojednostavljenje i metoda izlaganja? Treba li \u010dlanak prevesti na engleski? <\/b>
\n
Izvor: habr.com<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"

\u041d\u0435\u0442, \u043d\u0443 \u044f, \u043a\u043e\u043d\u0435\u0447\u043d\u043e, \u043d\u0435 \u0432\u0441\u0435\u0440\u044c\u0435\u0437. \u0414\u043e\u043b\u0436\u0435\u043d \u0436\u0435 \u0431\u044b\u0442\u044c \u043f\u0440\u0435\u0434\u0435\u043b, \u0434\u043e \u043a\u0430\u043a\u043e\u0439 \u0441\u0442\u0435\u043f\u0435\u043d\u0438 \u0432\u043e\u0437\u043c\u043e\u0436\u043d\u043e \u0443\u043f\u0440\u043e\u0449\u0430\u0442\u044c \u043f\u0440\u0435\u0434\u043c\u0435\u0442. \u041d\u043e \u0434\u043b\u044f \u043f\u0435\u0440\u0432\u044b\u0445 \u044d\u0442\u0430\u043f\u043e\u0432, \u043f\u043e\u043d\u0438\u043c\u0430\u043d\u0438\u044f \u0431\u0430\u0437\u043e\u0432\u044b\u0445 \u043a\u043e\u043d\u0446\u0435\u043f\u0446\u0438\u0439 \u0438 \u0431\u044b\u0441\u0442\u0440\u043e\u0433\u043e \u00ab\u0432\u044a\u0435\u0437\u0436\u0430\u043d\u0438\u044f\u00bb \u0432 \u0442\u0435\u043c\u0443, \u043c\u043e\u0436\u0435\u0442 \u0431\u044b\u0442\u044c, \u0438 \u0434\u043e\u043f\u0443\u0441\u0442\u0438\u043c\u043e. \u0410 \u043a\u0430\u043a \u043f\u0440\u0430\u0432\u0438\u043b\u044c\u043d\u043e \u043f\u043e\u0438\u043c\u0435\u043d\u043e\u0432\u0430\u0442\u044c \u0434\u0430\u043d\u043d\u044b\u0439 \u043c\u0430\u0442\u0435\u0440\u0438\u0430\u043b (\u0432\u0430\u0440\u0438\u0430\u043d\u0442\u044b: \u00ab\u041c\u0430\u0448\u0438\u043d\u043d\u043e\u0435 \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u0435 \u0434\u043b\u044f \u0447\u0430\u0439\u043d\u0438\u043a\u043e\u0432\u00bb, \u00ab\u0410\u043d\u0430\u043b\u0438\u0437 \u0434\u0430\u043d\u043d\u044b\u0445 \u0441 \u043f\u0435\u043b\u0435\u043d\u043e\u043a\u00bb, \u00ab\u0410\u043b\u0433\u043e\u0440\u0438\u0442\u043c\u044b \u0434\u043b\u044f \u0441\u0430\u043c\u044b\u0445 \u043c\u0430\u043b\u0435\u043d\u044c\u043a\u0438\u0445\u00bb), \u043e\u0431\u0441\u0443\u0434\u0438\u043c \u0432 \u043a\u043e\u043d\u0446\u0435. \u041a […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":22912,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-30936","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\udd47Strojno u\u010denje bez Pythona, Anaconde i ostalih gmazova | ProHoster","description":"Ne, pa naravno da to ne mislim ozbiljno. Mora postojati granica do koje mjere je mogu\u0107e pojednostaviti predmet. No, za prve faze, razumijevanje osnovnih koncepcija i brzo \"uranjanje\" u temu, mo\u017eda je to i dopu\u0161teno.","canonical_url":"https:\/\/prohoster.info\/hr\/blog\/news\/mashinnoe-obuchenie-bez-python-anaconda-i-prochih-presmykayushhihsya","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"hr_HR","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\udd47\u041c\u0430\u0448\u0438\u043d\u043d\u043e\u0435 \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u0435 \u0431\u0435\u0437 Python, Anaconda \u0438 \u043f\u0440\u043e\u0447\u0438\u0445 \u043f\u0440\u0435\u0441\u043c\u044b\u043a\u0430\u044e\u0449\u0438\u0445\u0441\u044f | ProHoster","og:description":"\u041d\u0435\u0442, \u043d\u0443 \u044f, \u043a\u043e\u043d\u0435\u0447\u043d\u043e, \u043d\u0435 \u0432\u0441\u0435\u0440\u044c\u0435\u0437. \u0414\u043e\u043b\u0436\u0435\u043d \u0436\u0435 \u0431\u044b\u0442\u044c \u043f\u0440\u0435\u0434\u0435\u043b, \u0434\u043e \u043a\u0430\u043a\u043e\u0439 \u0441\u0442\u0435\u043f\u0435\u043d\u0438 \u0432\u043e\u0437\u043c\u043e\u0436\u043d\u043e \u0443\u043f\u0440\u043e\u0449\u0430\u0442\u044c \u043f\u0440\u0435\u0434\u043c\u0435\u0442. \u041d\u043e \u0434\u043b\u044f \u043f\u0435\u0440\u0432\u044b\u0445 \u044d\u0442\u0430\u043f\u043e\u0432, \u043f\u043e\u043d\u0438\u043c\u0430\u043d\u0438\u044f \u0431\u0430\u0437\u043e\u0432\u044b\u0445 \u043a\u043e\u043d\u0446\u0435\u043f\u0446\u0438\u0439 \u0438 \u0431\u044b\u0441\u0442\u0440\u043e\u0433\u043e \u00ab\u0432\u044a\u0435\u0437\u0436\u0430\u043d\u0438\u044f\u00bb \u0432 \u0442\u0435\u043c\u0443, \u043c\u043e\u0436\u0435\u0442 \u0431\u044b\u0442\u044c, \u0438 \u0434\u043e\u043f\u0443\u0441\u0442\u0438\u043c\u043e.","og:url":"https:\/\/prohoster.info\/hr\/blog\/news\/mashinnoe-obuchenie-bez-python-anaconda-i-prochih-presmykayushhihsya","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":"2019-10-31T18:38:20+00:00","article:modified_time":"2019-10-31T18:38:20+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"30936","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 03:45:19","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-03-01 03:27:05","updated":"2026-01-21 03:45:19","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/posts\/30936","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/comments?post=30936"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/posts\/30936\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/media\/22912"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/media?parent=30936"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/categories?post=30936"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/hr\/wp-json\/wp\/v2\/tags?post=30936"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}