{"id":54780,"date":"2020-01-04T00:00:00","date_gmt":"2020-01-03T21:00:00","guid":{"rendered":"https:\/\/prohoster.info\/blog\/blog_prohoster\/nejroseti-kuda-eto-vse-dvizhetsya"},"modified":"2020-02-18T14:02:50","modified_gmt":"2020-02-18T11:02:50","slug":"nejroseti-kuda-eto-vse-dvizhetsya","status":"publish","type":"post","link":"https:\/\/prohoster.info\/az\/blog\/news\/nejroseti-kuda-eto-vse-dvizhetsya","title":{"rendered":"Nevro \u015f\u0259b\u0259k\u0259l\u0259ri. Bu h\u0259r \u015fey hara gedir?","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"<p>M\u0259qal\u0259 iki hiss\u0259d\u0259n ibar\u0259tdir:<\/p>\n<p><\/p>\n<ol>\n<li>\u015e\u0259kild\u0259 obyektl\u0259rin a\u015fkar edilm\u0259si v\u0259 \u015f\u0259kill\u0259rin seqmentasiyas\u0131 \u015f\u0259b\u0259k\u0259l\u0259rinin b\u0259zi arxitekturlar\u0131n\u0131 \u0259hat\u0259 ed\u0259n q\u0131sa bir izah, m\u0259n\u0259 \u0259n ayd\u0131n resurs ba\u011flant\u0131lar\u0131 il\u0259. \u0130zahat videolar\u0131n\u0131 se\u00e7m\u0259y\u0259 \u00e7al\u0131\u015fd\u0131m, m\u00fcmk\u00fcn q\u0259d\u0259r rus dilind\u0259.<\/li>\n<li>\u0130kinci hiss\u0259 neyron \u015f\u0259b\u0259k\u0259l\u0259rinin arxitekturalar\u0131n\u0131n inki\u015faf istiqam\u0259tini anlama\u011fa \u00e7al\u0131\u015fmaqla ba\u011fl\u0131d\u0131r. V\u0259 onlar\u0131n \u0259sas\u0131ndak\u0131 texnologiyalar.<\/li>\n<\/ol>\n<p><\/p>\n<p><img decoding=\"async\" alt=\"Nevro \u015f\u0259b\u0259k\u0259l\u0259ri. Bu h\u0259r \u015fey hara gedir?\" src=\"\/wp-content\/uploads\/2020\/01\/3e0238547dc956dbb069c11241e4534f.png\" style=\"display:block;margin: 0 auto;\" \/><\/p>\n<p><\/p>\n<p>\u015e\u0259kil 1 \u2013 Neyron \u015f\u0259b\u0259k\u0259l\u0259rinin arxitekturalar\u0131n\u0131 anlamaq asan deyil<\/p>\n<p><\/p>\n<p>H\u0259r \u015fey ondan ba\u015flad\u0131 ki, Android telefonunda t\u0259snifat v\u0259 obyekt a\u015fkar etm\u0259 \u00fczr\u0259 iki n\u00fcmayi\u015f t\u0259tbiqi etdim:<\/p>\n<p><\/p>\n<ul>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/foobar167\/junkyard\/tree\/master\/object_classifier\">Back-end demo<\/a><\/noindex>, m\u0259lumatlar\u0131n serverd\u0259 i\u015fl\u0259nm\u0259si v\u0259 telefona \u00f6t\u00fcr\u00fclm\u0259si il\u0259. \u00dc\u00e7 n\u00f6v ay\u0131 \u015f\u0259kill\u0259rinin t\u0259snifat\u0131: q\u0259hv\u0259yi, qara v\u0259 pl\u00fcmo.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/foobar167\/android\/tree\/master\/object_detection_demo\">Front-end demo<\/a><\/noindex>, m\u0259lumatlar\u0131n telefonun \u00f6z\u00fcnd\u0259 i\u015fl\u0259nm\u0259si il\u0259. \u00dc\u00e7 n\u00f6v obyektl\u0259rin a\u015fkar edilm\u0259si: f\u0131nd\u0131q, paxlava v\u0259 xurma.<\/li>\n<\/ul>\n<p><noindex><a rel=\"nofollow\" name=\"habracut\"><\/a><\/noindex><\/p>\n<p>\u015e\u0259kil t\u0259snifat\u0131, \u015f\u0259kil \u00fcz\u0259rind\u0259 obyektl\u0259rin a\u015fkar edilm\u0259si v\u0259 <noindex><a rel=\"nofollow\" href=\"https:\/\/medium.com\/analytics-vidhya\/image-classification-vs-object-detection-vs-image-segmentation-f36db85fe81\">\u015f\u0259kil seqmentasiyas\u0131<\/a><\/noindex>aras\u0131nda f\u0259rq var. Buna g\u00f6r\u0259 d\u0259, \u015f\u0259kill\u0259rd\u0259 obyektl\u0259ri a\u015fkar ed\u0259n v\u0259 seqmentasiya ed\u0259 bil\u0259n neyron \u015f\u0259b\u0259k\u0259l\u0259rinin arxitekturlar\u0131n\u0131 \u00f6yr\u0259nm\u0259y\u0259 ehtiyac var. M\u0259n\u0259 \u0259n ayd\u0131n resurs ba\u011flant\u0131lar\u0131 olan a\u015fa\u011f\u0131dak\u0131 arxitektura n\u00fcmun\u0259l\u0259rini tapd\u0131m:<\/p>\n<p><\/p>\n<ul>\n<li>R-CNN \u0259sasl\u0131 arxitekturalar seriyas\u0131 (<strong>R<\/strong>b\u00f6lg\u0259l\u0259r <strong>\u2014 TVM \u00fc\u00e7\u00fcn ikili icra kodu istehsal ed\u0259n Fift kitabxanas\u0131. Fift Y\u0131\u011fman\u0131n t\u0259rtibat\u00e7\u0131s\u0131 yoxdur. Bu,<\/strong>konsol x\u0259b\u0259rl\u0259ri \/ <strong>N<\/strong>\u015f\u0259b\u0259k\u0259 strukturunu): R-CNN, Fast R-CNN, <strong>N<\/strong>Faster R-CNN <noindex><a rel=\"nofollow\" href=\"https:\/\/medium.com\/@smallfishbigsea\/faster-r-cnn-explained-864d4fb7e3f8\">Mask R-CNN<\/a><\/noindex>, <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/0vt05rQqk_I\">obyektin \u015f\u0259kilind\u0259 a\u015fkar edilm\u0259si \u00fc\u00e7\u00fcn Region Proposal Network (RPN) mexanizmi il\u0259 m\u0259hdud b\u00f6lg\u0259l\u0259r (bounding boxes) se\u00e7ilir. \u018fvv\u0259lc\u0259 RPN yerin\u0259 daha yava\u015f Selective Search mexanizmi istifad\u0259 olunurdu. Sonra se\u00e7ilmi\u015f m\u0259hdud b\u00f6lg\u0259l\u0259r adi neyron \u015f\u0259b\u0259k\u0259sin\u0259 t\u0259snifat \u00fc\u00e7\u00fcn verilir. R-CNN arxitekturas\u0131nda m\u0259hdud b\u00f6lg\u0259l\u0259r \u00fcz\u0259rind\u0259 d\u00f6vr ard\u0131c\u0131ll\u0131\u011f\u0131 (for loops) var, \u00fcmumilikd\u0259 2000-\u0259 q\u0259d\u0259r AlexNet daxili \u015f\u0259b\u0259k\u0259si \u00fczr\u0259 ke\u00e7irilm\u0259 var. A\u00e7\u0131q d\u00f6vr ard\u0131c\u0131ll\u0131\u011f\u0131 s\u0259b\u0259bind\u0259n \u015f\u0259kill\u0259rin emal s\u00fcr\u0259ti azal\u0131r. H\u0259r yeni arxitektura versiyas\u0131 il\u0259 a\u00e7\u0131q d\u00f6vr ard\u0131c\u0131ll\u0131\u011f\u0131n\u0131n say\u0131 azal\u0131r v\u0259 onu art\u0131rmaq \u00fc\u00e7\u00fcn s\u00fcr\u0259ti art\u0131rmaq m\u0259qs\u0259dil\u0259 bir \u00e7ox dig\u0259r d\u0259yi\u015fiklikl\u0259r edilir, habel\u0259 obyekt a\u015fkarlanmas\u0131 tap\u015f\u0131r\u0131\u011f\u0131 Mask R-CNN-d\u0259 obyekt seqmentasiyas\u0131na d\u0259yi\u015fdirilir.<\/a><\/noindex>YOLO<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/L0tzmv--CGY\">sadece<\/a><\/noindex> (<strong>Y<\/strong>ilk <strong>N<\/strong>qiym\u0259t\u0259n <strong>h\u0259tta<\/strong>g\u00f6r\u00fcr <strong>N<\/strong>bir d\u0259f\u0259 \u2013 real zamanda mobil cihazlarda obyektl\u0259ri tan\u0131yan ilk neyron \u015f\u0259b\u0259k\u0259dir. F\u0259rqli x\u00fcsusiyy\u0259t: obyektl\u0259rin bir ke\u00e7id zaman\u0131 ayr\u0131-se\u00e7kilik edilm\u0259si (yaln\u0131z bir d\u0259f\u0259 baxmaq kifay\u0259tdir). Y\u0259ni YOLO arxitekturas\u0131nda a\u00e7\u0131q d\u00f6vr ard\u0131c\u0131ll\u0131\u011f\u0131 yoxdur, buna g\u00f6r\u0259 d\u0259 \u015f\u0259b\u0259k\u0259 s\u00fcr\u0259tli i\u015fl\u0259yir. M\u0259s\u0259l\u0259n, bel\u0259 bir analoqu: NumPy-d\u0259 matrisl\u0259rl\u0259 \u0259m\u0259liyyatlar zaman\u0131 da a\u00e7\u0131q d\u00f6vr ard\u0131c\u0131ll\u0131\u011f\u0131 yoxdur; bel\u0259 ki, NumPy-dak\u0131 h\u0259yata ke\u00e7irilm\u0259l\u0259r daha a\u015fa\u011f\u0131 s\u0259viyy\u0259d\u0259 C proqramla\u015fd\u0131rma dili il\u0259 h\u0259yata ke\u00e7irilir. YOLO \u0259vv\u0259lc\u0259d\u0259n m\u00fc\u0259yy\u0259n edilmi\u015f p\u0259nc\u0259r\u0259 \u015f\u0259b\u0259k\u0259sind\u0259n istifad\u0259 edir. Eyni obyektin d\u0259f\u0259l\u0259rl\u0259 m\u00fc\u0259yy\u0259n edilm\u0259m\u0259si \u00fc\u00e7\u00fcn p\u0259nc\u0259r\u0259 \u00fcst-\u00fcst\u0259 d\u00fc\u015fm\u0259 \u0259msal\u0131 (IoU, <strong>I<\/strong>k\u0259si\u015fm\u0259 <strong>o<\/strong>ver <strong>U<\/strong>sendikas\u0131). Bu arxitektura geni\u015f bir spektrda i\u015fl\u0259yir v\u0259 y\u00fcks\u0259k <noindex><a rel=\"nofollow\" href=\"https:\/\/ru.wikipedia.org\/wiki\/%D0%A0%D0%BE%D0%B1%D0%B0%D1%81%D1%82%D0%BD%D0%BE%D1%81%D1%82%D1%8C\">dayan\u0131ql\u0131l\u0131q<\/a><\/noindex>: model fotolar \u00fcz\u0259rind\u0259 t\u0259hsil ala bil\u0259r, amma eyni zamanda \u0259l il\u0259 \u00e7\u0259kilmi\u015f \u015f\u0259kill\u0259rd\u0259 d\u0259 yax\u015f\u0131 i\u015fl\u0259yir.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/P8e-G-Mhx4k\">SSD<\/a><\/noindex> (<strong>S<\/strong>t\u0259k <strong>S<\/strong>\u00e7oxsayl\u0131 MultiBox <strong>D<\/strong>detektor) \u2013 \u0259n yax\u015f\u0131 YOLO arxitekturas\u0131 \u201chaclar\u0131\u201d (m\u0259s\u0259l\u0259n, non-maximum suppression) istifad\u0259 olunur v\u0259 yeni \u0259lav\u0259 edilir ki, s\u00fcni intellekt daha s\u00fcr\u0259tli v\u0259 d\u0259qiq i\u015fl\u0259sin. X\u00fcsusi x\u00fcsusiyy\u0259t: obyektl\u0259ri t\u0259k bir ke\u00e7id il\u0259 t\u0259yin etm\u0259k \u00fc\u00e7\u00fcn verilmi\u015f \u015f\u0259b\u0259k\u0259 p\u0259nc\u0259r\u0259l\u0259ri (default box) vasit\u0259sil\u0259 \u015f\u0259kill\u0259r piramidas\u0131 \u00fcz\u0259rind\u0259 i\u015fl\u0259m\u0259k. \u015e\u0259kill\u0259r piramidas\u0131, ard\u0131c\u0131l konvolusiya v\u0259 max-pooling \u0259m\u0259liyyatlar\u0131nda konvolusiya tens\u00f6rl\u0259rind\u0259 kodlan\u0131r (max-pooling \u0259m\u0259liyyat\u0131 zaman\u0131 m\u0259kan \u00f6l\u00e7\u00fcs\u00fc azal\u0131r). Bu \u015f\u0259kild\u0259 h\u0259m b\u00f6y\u00fck, h\u0259m d\u0259 ki\u00e7ik obyektl\u0259r t\u0259k bir \u015f\u0259b\u0259k\u0259 ke\u00e7idi il\u0259 m\u00fc\u0259yy\u0259n edilir.<\/li>\n<li>MobileSSD (<strong>Mobil<\/strong>NetV2 + <strong>SSD<\/strong>) \u2013 iki neyron \u015f\u0259b\u0259k\u0259si arxitekturas\u0131n\u0131n birl\u0259\u015fm\u0259sidir. Birinci \u015f\u0259b\u0259k\u0259 <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/352804\/\">MobileNetV2<\/a><\/noindex> s\u00fcr\u0259tli i\u015fl\u0259yir v\u0259 tan\u0131ma d\u0259qiqliyini art\u0131r\u0131r. MobileNetV2 \u0259vv\u0259lc\u0259 istifad\u0259 olunan VGG-16-n\u0131n \u0259v\u0259zin\u0259 t\u0259tbiq olunur <noindex><a rel=\"nofollow\" href=\"https:\/\/arxiv.org\/abs\/1512.02325\">\u0259sas m\u0259qal\u0259d\u0259<\/a><\/noindex>. \u0130kinci \u015f\u0259b\u0259k\u0259 SSD \u015f\u0259kil \u00fcz\u0259rind\u0259 obyektl\u0259rin yerini m\u00fc\u0259yy\u0259n edir.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/ge_RT5wvHvY\">SqueezeNet<\/a><\/noindex> \u2013 \u00e7ox ki\u00e7ik, amma d\u0259qiq bir neyron \u015f\u0259b\u0259k\u0259sidir. \u00d6z ba\u015f\u0131na obyektl\u0259ri a\u015fkar etm\u0259k m\u0259s\u0259l\u0259sini h\u0259ll etmir. Lakin m\u00fcxt\u0259lif arxitekturalar\u0131n birl\u0259\u015fm\u0259sind\u0259 istifad\u0259 edil\u0259 bil\u0259r. Mobil cihazlarda da t\u0259tbiq oluna bil\u0259r. X\u00fcsusi x\u00fcsusiyy\u0259tl\u0259ri, \u0259vv\u0259lc\u0259 m\u0259lumatlar\u0131n d\u00f6rd 1\u00d71 konvolusiya filtri il\u0259 s\u0131x\u0131lmas\u0131ndan sonra d\u00f6rd 1\u00d71 v\u0259 d\u00f6rd 3\u00d73 konvolusiya filtrl\u0259ri il\u0259 geni\u015fl\u0259nm\u0259sidir. Bir bu c\u00fcr s\u0131x\u0131lma-geni\u015fl\u0259m\u0259 iterasiyas\u0131 \u201cFire Module\u201d adlan\u0131r.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/b6jhopSMit8\">DeepLab<\/a><\/noindex> (D\u0259rin \u0130dar\u0259edici \u015e\u0259kil Seqmentasiyas\u0131 D\u0259rin Konvolusiya \u015e\u0259b\u0259k\u0259l\u0259ri il\u0259) \u2013 \u015f\u0259kil \u00fcz\u0259rind\u0259 obyektl\u0259ri seqmentasiya edir. Arxitekturan\u0131n x\u00fcsusi x\u00fcsusiyy\u0259ti, m\u0259kan h\u0259llini qoruyan seyreltilmi\u015f (dilated convolution) konvolusiyad\u0131r. Daha sonra n\u0259tic\u0259l\u0259rin post-prosessinq m\u0259rh\u0259l\u0259si, ki\u00e7ik s\u0259sl\u0259ri seqmentasiyadan aradan qald\u0131rma\u011fa v\u0259 seqmentasiya olunmu\u015f \u015f\u0259kilin keyfiyy\u0259tini art\u0131rma\u011fa imkan ver\u0259n qrafik ehtimal modeli (\u015f\u0259rti t\u0259sad\u00fcfi sah\u0259) istifad\u0259 olunur. \u201cQrafik ehtimal modeli\u201d ad\u0131ndan gizl\u0259n\u0259n adi Gauss filtridir, be\u015f n\u00f6qt\u0259 il\u0259 yax\u0131nla\u015fd\u0131r\u0131lm\u0131\u015fd\u0131r.<\/li>\n<li>N\u0259tic\u0259 \u0259ld\u0259 etm\u0259y\u0259 \u00e7al\u0131\u015fd\u0131m <noindex><a rel=\"nofollow\" href=\"https:\/\/arxiv.org\/abs\/1711.06897\">RefineDet<\/a><\/noindex> (T\u0259k Pass <strong>D\u00fcz\u0259li\u015f<\/strong>m\u00fcxt\u0259liflik Neyron \u015e\u0259b\u0259k\u0259si Obyekti <strong>A\u015fkarlama\u0131<\/strong>lakin az \u015fey ba\u015fa d\u00fc\u015fd\u00fcm.<\/li>\n<li>H\u0259m\u00e7inin \u201cdiqq\u0259t\u201d (attention) texnologiyas\u0131n\u0131n nec\u0259 i\u015fl\u0259diyin\u0259 baxd\u0131m: <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/W2rWgXJBZhU\">video1<\/a><\/noindex>, <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/iDulhoQ2pro\">video2<\/a><\/noindex>, <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/H6Qiegq_36c\">video3<\/a><\/noindex>. \u201cDiqq\u0259t\u201d arxitekturas\u0131n\u0131n x\u00fcsusi x\u00fcsusiyy\u0259ti, \u015f\u0259kil \u00fcz\u0259rind\u0259 y\u00fcks\u0259k diqq\u0259t b\u00f6lg\u0259l\u0259rini avtomatik ay\u0131rmaqd\u0131r (RoI, <strong>R<\/strong>regionlar <strong>o<\/strong>f <strong>I<\/strong>Attention Unit adlanan neyron \u015f\u0259b\u0259k\u0259si il\u0259 maraql\u0131 b\u00f6lg\u0259l\u0259ri a\u015fkar etm\u0259k. Artan diqq\u0259t b\u00f6lg\u0259l\u0259ri m\u0259hdud b\u00f6lg\u0259l\u0259r\u0259 (bounding boxes) b\u0259nz\u0259yir, lakin onlardan f\u0259rqli olaraq t\u0259svir \u00fcz\u0259rind\u0259 sabitl\u0259\u015fm\u0259yib v\u0259 bulan\u0131k s\u0259rh\u0259dl\u0259r\u0259 malikdir. Sonra diqq\u0259t b\u00f6lg\u0259l\u0259rind\u0259n x\u00fcsusiyy\u0259tl\u0259r (features) \u00e7\u0131xar\u0131laraq, bunlar rekurent neyron \u015f\u0259b\u0259k\u0259l\u0259rin\u0259 verilir. <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/5lUUrREboSk\">LSDM, GRU v\u0259 ya Vanilla RNN<\/a><\/noindex>. Rekurent neyron \u015f\u0259b\u0259k\u0259l\u0259ri ard\u0131c\u0131ll\u0131qda x\u00fcsusiyy\u0259tl\u0259r aras\u0131ndak\u0131 m\u00fcst\u0259qil \u0259laq\u0259l\u0259ri analiz ed\u0259 bilir. Rekurent neyron \u015f\u0259b\u0259k\u0259l\u0259ri \u0259vv\u0259lc\u0259 m\u0259tnl\u0259ri dig\u0259r dill\u0259r\u0259 t\u0259rc\u00fcm\u0259 etm\u0259k \u00fc\u00e7\u00fcn istifad\u0259 olunurdu, indi is\u0259 <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/e-WB4lfg30M\">\u015f\u0259kill\u0259ri m\u0259tnd\u0259 \u00e7evirm\u0259k<\/a><\/noindex> v\u0259 <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/rAbhypxs1qQ\">m\u0259tni \u015f\u0259kil kimi \u00e7evirm\u0259k<\/a><\/noindex>.<\/li>\n<\/ul>\n<p><\/p>\n<p>Bu memarl\u0131\u011f\u0131 \u00f6yr\u0259n\u0259rk\u0259n <strong>he\u00e7 n\u0259 ba\u015fa d\u00fc\u015fm\u0259diyimi anlad\u0131m.<\/strong>. M\u0259s\u0259l\u0259 o deyil ki, m\u0259nim neyron \u015f\u0259b\u0259k\u0259min diqq\u0259t mexanizmi il\u0259 problemi var. B\u00fct\u00fcn bu memarlar\u0131n yarad\u0131lmas\u0131 b\u00f6y\u00fck bir hackathonu xat\u0131rlad\u0131r, burada m\u00fc\u0259llifl\u0259r hacklarda bir-birin\u0259 r\u0259qab\u0259t apar\u0131r. Hack - \u00e7\u0259tin bir proqram problemi \u00fc\u00e7\u00fcn s\u00fcr\u0259tli bir h\u0259ll. Y\u0259ni, bu memarlar\u0131n he\u00e7 biri aras\u0131nda g\u00f6r\u00fcn\u0259n v\u0259 anla\u015f\u0131lan m\u0259ntiqi ba\u011f yoxdur. Onlar\u0131 bir araya g\u0259tir\u0259n b\u00fct\u00fcn bunlar, b\u0259zil\u0259rinin bir-birind\u0259n ald\u0131klar\u0131 \u0259n u\u011furlu hackl\u0259r toplusudur, \u00fcst\u0259lik, ortaq bir <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/Ilg3gGewQ5U\">geri ba\u011flant\u0131 konvolyasiya \u0259m\u0259liyyat\u0131<\/a><\/noindex> (s\u0259hv yay\u0131lmas\u0131, backpropagation). Yoxdur <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/272473\/\">sistematik d\u00fc\u015f\u00fcnc\u0259<\/a><\/noindex>! N\u0259yi d\u0259yi\u015fm\u0259k v\u0259 olan nailiyy\u0259tl\u0259ri nec\u0259 optimalla\u015fd\u0131rmaq laz\u0131m oldu\u011fu ayd\u0131n deyil.<\/p>\n<p><\/p>\n<p>Hacks'lar aras\u0131nda m\u0259ntiqi ba\u011flant\u0131n\u0131n olmamas\u0131 n\u0259tic\u0259sind\u0259, onlar\u0131 xat\u0131rlamaq v\u0259 praktikada t\u0259tbiq etm\u0259k \u00e7ox \u00e7\u0259tindir. Bu, par\u00e7alanm\u0131\u015f bilikl\u0259rdir. \u018fn yax\u015f\u0131 halda, bir ne\u00e7\u0259 maraql\u0131 v\u0259 g\u00f6zl\u0259nilm\u0259z moment xat\u0131rlan\u0131r, lakin ba\u015fa d\u00fc\u015f\u00fcl\u0259n v\u0259 ba\u015fa d\u00fc\u015f\u00fclm\u0259y\u0259nl\u0259rin \u00e7oxu art\u0131q bir ne\u00e7\u0259 g\u00fcn \u0259rzind\u0259 yadda\u015fdan silinir. Bir h\u0259ft\u0259 sonra \u0259n az\u0131 memarl\u0131\u011f\u0131n ad\u0131 xat\u0131rlan\u0131rsa, bu yax\u015f\u0131d\u0131r. V\u0259 m\u0259qal\u0259l\u0259ri oxumaq v\u0259 icmal videolar\u0131n\u0131 izl\u0259m\u0259k \u00fc\u00e7\u00fcn bir ne\u00e7\u0259 saat v\u0259 h\u0259tta g\u00fcn i\u015f vaxt\u0131 s\u0259rf olunub!<\/p>\n<p><\/p>\n<p><img decoding=\"async\" alt=\"Nevro \u015f\u0259b\u0259k\u0259l\u0259ri. Bu h\u0259r \u015fey hara gedir?\" src=\"\/wp-content\/uploads\/2020\/01\/86f0f24e3be1d0f1a210f3132897d981.png\" style=\"display:block;margin: 0 auto;\" \/><\/p>\n<p><\/p>\n<p>\u015e\u0259kil 2 \u2013 <noindex><a rel=\"nofollow\" href=\"https:\/\/www.asimovinstitute.org\/neural-network-zoo\/\">Neyron \u015f\u0259b\u0259k\u0259l\u0259rinin zoo park\u0131<\/a><\/noindex><\/p>\n<p><\/p>\n<p>M\u0259nim \u015f\u0259xsi fikrim\u0259 g\u00f6r\u0259, \u0259ks\u0259r elmi m\u0259qal\u0259 m\u00fc\u0259llifl\u0259ri oxucunun bu par\u00e7alanm\u0131\u015f bilikl\u0259ri ba\u015fa d\u00fc\u015fm\u0259m\u0259si \u00fc\u00e7\u00fcn \u0259ll\u0259rind\u0259n g\u0259l\u0259ni edirl\u0259r. Lakin on c\u00fcml\u0259d\u0259n ibar\u0259t i\u015ftirak d\u0259y\u0259ri il\u0259 \u0259h\u0259miyy\u0259tli izahlar verm\u0259k - bu ayr\u0131ca bir m\u0259qal\u0259 m\u00f6vzusudur (problem <noindex><a rel=\"nofollow\" href=\"https:\/\/en.wikipedia.org\/wiki\/Publish_or_perish\">publish or perish<\/a><\/noindex>).<\/p>\n<p><\/p>\n<p>Bu s\u0259b\u0259bd\u0259n neyron \u015f\u0259b\u0259k\u0259l\u0259ri haqq\u0131nda m\u0259lumat\u0131 sisteml\u0259\u015fdirm\u0259k v\u0259 bel\u0259likl\u0259, anlama v\u0259 xat\u0131rlama keyfiyy\u0259tini art\u0131rmaq z\u0259rur\u0259ti yaran\u0131b. Bunun \u00fc\u00e7\u00fcn s\u00fcni neyron \u015f\u0259b\u0259k\u0259l\u0259rinin ayr\u0131-ayr\u0131 texnologiyalar\u0131 v\u0259 memarlar\u0131n\u0131 ara\u015fd\u0131rma\u011f\u0131n \u0259sas m\u0259s\u0259l\u0259si a\u015fa\u011f\u0131dak\u0131 oldu: <strong>bunun hara t\u0259r\u0259f ged\u0259c\u0259yini \u00f6yr\u0259nm\u0259k<\/strong>, hans\u0131sa spesifik neyron \u015f\u0259b\u0259k\u0259sinin qurulu\u015funa deyil.<\/p>\n<p><\/p>\n<p>B\u00fct\u00fcn bunlar nereye gedir. \u018fsas n\u0259tic\u0259l\u0259r:<\/p>\n<p><\/p>\n<ul>\n<li>Son iki ild\u0259 ma\u015f\u0131n \u00f6yr\u0259nm\u0259si sah\u0259sind\u0259 startaplar\u0131n say\u0131 <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/company\/recognitor\/blog\/455676\/\">k\u0259skin \u015f\u0259kild\u0259 azald\u0131<\/a><\/noindex>. M\u00fcmk\u00fcn s\u0259b\u0259bi: \"neyron \u015f\u0259b\u0259k\u0259l\u0259r art\u0131q yeni bir \u015fey deyil\".<\/li>\n<li>H\u0259r k\u0259s sad\u0259 bir problemi h\u0259ll etm\u0259k \u00fc\u00e7\u00fcn i\u015fl\u0259y\u0259n neyron \u015f\u0259b\u0259k\u0259si yarada bil\u0259r. Bunun \u00fc\u00e7\u00fcn \"model ziyar\u0259tgah\u0131\" (model zoo) i\u00e7ind\u0259n haz\u0131r bir modeli g\u00f6t\u00fcr\u0259c\u0259k v\u0259 neyron \u015f\u0259b\u0259k\u0259sinin son qat\u0131n\u0131 (<noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/yofjFQddwHE\">transfer learning<\/a><\/noindex>) haz\u0131r m\u0259lumatlarla <noindex><a rel=\"nofollow\" href=\"https:\/\/toolbox.google.com\/datasetsearch\">Google Dataset Search<\/a><\/noindex> v\u0259 ya <noindex><a rel=\"nofollow\" href=\"https:\/\/www.kaggle.com\/datasets\">Kaggle-dan 25 min datasetd\u0259n<\/a><\/noindex> pulsuz <noindex><a rel=\"nofollow\" href=\"https:\/\/www.dataschool.io\/cloud-services-for-jupyter-notebook\/\">Jupyter Notebook buludunda<\/a><\/noindex>.<\/li>\n<li>B\u00f6y\u00fck neyron \u015f\u0259b\u0259k\u0259si istehsal\u00e7\u0131lar\u0131 <strong>\"model ziyar\u0259tgahlar\u0131\"<\/strong> (model zoo) yaratma\u011fa ba\u015flad\u0131lar. Onlar\u0131n k\u00f6m\u0259yi il\u0259 tez bir kommersiya t\u0259tbiqi yaratmaq m\u00fcmk\u00fcnd\u00fcr: <noindex><a rel=\"nofollow\" href=\"https:\/\/tfhub.dev\/\">TF Hub<\/a><\/noindex> TensorFlow \u00fc\u00e7\u00fcn, <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/open-mmlab\/mmdetection\">MMDetection<\/a><\/noindex> PyTorch \u00fc\u00e7\u00fcn, <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/facebookresearch\/Detectron\">Detectron<\/a><\/noindex> Caffe2 \u00fc\u00e7\u00fcn, <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/wkentaro\/chainer-modelzoo\">chainer-modelzoo<\/a><\/noindex> Chainer \u00fc\u00e7\u00fcn v\u0259 <noindex><a rel=\"nofollow\" href=\"https:\/\/modelzoo.co\/\">dig\u0259r<\/a><\/noindex>.<\/li>\n<li>Real zamanda \u00e7al\u0131\u015fan neyron \u015f\u0259b\u0259k\u0259l\u0259ri <strong>(real-time) mobil cihazlarda. Saniy\u0259d\u0259 10-dan 50-y\u0259 q\u0259d\u0259r \u00e7\u0259r\u00e7iv\u0259.<\/strong> Neyron \u015f\u0259b\u0259k\u0259l\u0259rinin telefonlarda (TF Lite), brauzerl\u0259rd\u0259 (TF.js) v\u0259<\/li>\n<li>ev \u0259\u015fyalar\u0131nda <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/19ZNz2N79u4\">(IoT,<\/a><\/noindex> internet <strong>I<\/strong>Things). X\u00fcsusil\u0259 d\u0259 telefonlarda, hans\u0131 ki, art\u0131q \"\u00e7\u0259ki\" s\u0259viyy\u0259sind\u0259 neyron \u015f\u0259b\u0259k\u0259l\u0259rini d\u0259st\u0259kl\u0259yirl\u0259r (neyroakseleratorlar). <strong>o<\/strong>f <strong>H<\/strong>\u0259r bir cihaz, geyim \u0259\u015fyalar\u0131 v\u0259 b\u0259lk\u0259 d\u0259 h\u0259tta qida aras\u0131nda IP-v6 \u00fcnvan\u0131 olacaq<\/li>\n<li>v\u0259 bir-biri il\u0259 \u00fcnsiyy\u0259t quracaq\u201d \u2013 <strong>Sebastian Thrun<\/strong> M ma\u015f\u0131n \u00f6yr\u0259nm\u0259 \u00fczr\u0259 n\u0259\u015frl\u0259rin say\u0131n\u0131n artmas\u0131 <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/GG7H8Xa4m8I?t=85\">2015-ci ild\u0259n etibar\u0259n Mur qanununu<\/a><\/noindex>.<\/li>\n<li>(h\u0259r iki ild\u0259n bir iki d\u0259f\u0259 artma) a\u015fma\u011fa ba\u015flad\u0131. Ayd\u0131n m\u0259s\u0259l\u0259dir ki, m\u0259qal\u0259l\u0259ri analiz ed\u0259n neyron \u015f\u0259b\u0259k\u0259l\u0259ri laz\u0131md\u0131r. <noindex><a rel=\"nofollow\" href=\"http:\/\/data-mining.philippe-fournier-viger.com\/too-many-machine-learning-papers\">A\u015fa\u011f\u0131dak\u0131 texnologiyalar populyarl\u0131q qazan\u0131r:<\/a><\/noindex> \u2013 populyarl\u0131q s\u00fcr\u0259tl\u0259 art\u0131r v\u0259 g\u00f6r\u00fcn\u00fcr TensorFlow-u \u00fcst\u0259l\u0259yir.<\/li>\n<li>Avtomatik hiperparametrl\u0259rin se\u00e7imi\n<ul>\n<li><strong>PyTorch<\/strong> AutoML<\/li>\n<li>\u2013 populyarl\u0131\u011f\u0131 yava\u015f-yava\u015f art\u0131r. <strong>D\u00fczg\u00fcnl\u00fckd\u0259 t\u0259dric\u0259n azalma v\u0259 hesablama s\u00fcr\u0259tinin artmas\u0131:<\/strong> bulan\u0131k m\u0259ntiq<\/li>\n<li>, alqoritml\u0259r <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/rln_kZbYaWc\">boosting<\/a><\/noindex>, qeyri-d\u0259qiq (t\u0259xmini) hesablamalar, kvantla\u015fd\u0131rma (neyron \u015f\u0259b\u0259k\u0259sinin a\u011f\u0131rl\u0131qlar\u0131 tam \u0259d\u0259dl\u0259r\u0259 \u00e7evrilib kvantla\u015fd\u0131r\u0131ld\u0131\u011f\u0131nda), neyroakseleratorlar. <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/MIPkK5ZAsms\">Videodan \u00fc\u00e7\u00f6l\u00e7\u00fcl\u00fc obyektl\u0259rin yarad\u0131lmas\u0131<\/a><\/noindex>, art\u0131q real zamanda.<\/li>\n<li>m\u00fc\u0259llifin bu sual\u0131n <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/e-WB4lfg30M\">\u015f\u0259kill\u0259ri m\u0259tnd\u0259 \u00e7evirm\u0259k<\/a><\/noindex> v\u0259 <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/rAbhypxs1qQ\">m\u0259tni \u015f\u0259kil kimi \u00e7evirm\u0259k<\/a><\/noindex>.<\/li>\n<li>DL-d\u0259 \u0259sas m\u0259s\u0259l\u0259 \u2013 \u00e7ox m\u0259lumatd\u0131r, lakin onu toplamaq v\u0259 ba\u015fa d\u00fc\u015fm\u0259k \u00e7\u0259tindir. Bu s\u0259b\u0259bd\u0259n neyron \u015f\u0259b\u0259k\u0259l\u0259ri \u00fc\u00e7\u00fcn avtonom qeyd etm\u0259 ( <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/OrHLacCDZVQ\">automated annotation<\/a><\/noindex>) inki\u015faf edir.<\/li>\n<li>Neyron \u015f\u0259b\u0259k\u0259l\u0259ri M\u0259lumat Elml\u0259ri q\u0259fl\u0259t\u0259n<noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/NcKTn4C91Yc\">eksperimental elm hal\u0131na g\u0259ldi<\/a><\/noindex>v\u0259<\/li>\n<\/ul>\n<\/li>\n<li>t\u0259krarlanabilirlik b\u00f6hran\u0131 <strong>\u0130T pullar\u0131 v\u0259 neyron \u015f\u0259b\u0259k\u0259l\u0259rinin populyarl\u0131\u011f\u0131 eyni zamanda meydana g\u0259ldi, hesablama bazar d\u0259y\u0259ri hal\u0131na g\u0259ldikd\u0259. \u0130qtisadiyyat q\u0131z\u0131ldan-valyuta hal\u0131na g\u0259lir<\/strong> q\u0131z\u0131l-valyuta-hesablama. <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/480348\">M\u0259nim<\/a><\/noindex>.<\/li>\n<li>ekonofizika <strong>v\u0259 \u0130T pullar\u0131n\u0131n yaranma s\u0259b\u0259bi haqq\u0131nda m\u0259qal\u0259m\u0259 bax\u0131n.<\/strong>T\u0259dric\u0259n yeni <noindex><a rel=\"nofollow\" href=\"https:\/\/ru.wikipedia.org\/wiki\/%D0%AD%D0%BA%D0%BE%D0%BD%D0%BE%D1%84%D0%B8%D0%B7%D0%B8%D0%BA%D0%B0\">ML\/DL proqramla\u015fd\u0131rma metodologiyas\u0131<\/a><\/noindex> (Ma\u015f\u0131n \u00d6yr\u0259nm\u0259 &amp; D\u0259rin \u00d6yr\u0259nm\u0259) meydana \u00e7\u0131x\u0131r ki, bu da proqram\u0131 \u00f6yr\u0259nilmi\u015f neyron \u015f\u0259b\u0259k\u0259 modell\u0259rinin toplam\u0131 kimi t\u0259qdim etm\u0259k \u00fcz\u0259rind\u0259 qurulub.<\/li>\n<\/ul>\n<p><\/p>\n<p>\u015e\u0259kil 3 \u2013 ML\/DL yeni proqramla\u015fd\u0131rma metodologiyas\u0131 <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/481844\">Ancaq<\/a><\/noindex> \"neyron \u015f\u0259b\u0259k\u0259l\u0259rinin n\u0259z\u0259riyy\u0259si\"<\/p>\n<p><\/p>\n<p><img decoding=\"async\" alt=\"Nevro \u015f\u0259b\u0259k\u0259l\u0259ri. Bu h\u0259r \u015fey hara gedir?\" src=\"\/wp-content\/uploads\/2020\/01\/17df76a613191522145abeba177c2685.png\" style=\"display:block;margin: 0 auto;\" \/><\/p>\n<p><\/p>\n<p>h\u0259l\u0259 d\u0259 ortaya \u00e7\u0131xmay\u0131b ki, sistemli d\u00fc\u015f\u00fcnm\u0259k v\u0259 i\u015fl\u0259m\u0259k \u00fc\u00e7\u00fcn istifad\u0259 olunsun. Hal-haz\u0131rda \"n\u0259z\u0259riyy\u0259\" olaraq adland\u0131r\u0131lanlar \u0259slind\u0259 eksperimental, heuristik alqoritml\u0259rdir.<\/p>\n<p><\/p>\n<p>M\u0259nim v\u0259 dig\u0259r resurslara ba\u011flant\u0131lar: <strong>\u00abnevro\u015f\u0259b\u0259k\u0259 n\u0259z\u0259riyy\u0259l\u0259ri\u00bb<\/strong>, \u00e7\u0259r\u00e7iv\u0259sind\u0259 sistemli d\u00fc\u015f\u00fcnm\u0259k v\u0259 i\u015fl\u0259m\u0259k m\u00fcmk\u00fcnd\u00fcr. \u0130ndi \u00abn\u0259z\u0259riyy\u0259\u00bb adland\u0131r\u0131lan \u015fey \u0259slind\u0259 eksperimental, heuristik alqoritml\u0259rdir.<\/p>\n<p><\/p>\n<p>M\u0259nim v\u0259 yaln\u0131z m\u0259nim resurslar\u0131ma olan linkl\u0259r:<\/p>\n<p><\/p>\n<ul>\n<li>Data Science \u00fczr\u0259 x\u0259b\u0259r b\u00fclleteni. \u018fsas\u0259n, \u015f\u0259kill\u0259rin i\u015fl\u0259nm\u0259si haqq\u0131nda. Almaq ist\u0259y\u0259nl\u0259r e-mail g\u00f6nd\u0259rsin (foobar167gmailcom). M\u0259qal\u0259 v\u0259 videolar\u0131n ba\u011flant\u0131lar\u0131n\u0131 material y\u0131\u011fd\u0131qca g\u00f6nd\u0259rir\u0259m.<\/li>\n<li>\u00dcmumi <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/foobar167\/articles\/blob\/master\/Machine_Learning\/courses_on_machine_learning.md\">kurs v\u0259 m\u0259qal\u0259l\u0259rin siyah\u0131s\u0131<\/a><\/noindex>, ke\u00e7diyim v\u0259 ke\u00e7m\u0259k ist\u0259diyim.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/foobar167\/articles\/blob\/master\/Ubuntu\/13_Keras_and_TensorFlow_how-tos.md#exercises\">Ba\u015flayanlar \u00fc\u00e7\u00fcn kurslar v\u0259 videolar<\/a><\/noindex>, neyron \u015f\u0259b\u0259k\u0259l\u0259ri \u00f6yr\u0259nm\u0259y\u0259 ba\u015flamaq \u00fc\u00e7\u00fcn ba\u015flayaca\u011f\u0131n\u0131z yerl\u0259rdir. \u00dcst\u0259lik bro\u015fura <noindex><a rel=\"nofollow\" href=\"https:\/\/foobar167.github.io\/page\/vvedeniye-v-mashinnoye-obucheniye-i-iskusstvennyye-neyronnyye-seti.html\">\u00abMa\u015f\u0131n \u00d6yr\u0259nm\u0259si v\u0259 S\u00fcni Neyron \u015e\u0259b\u0259k\u0259l\u0259r\u0259 Giri\u015f\u00bb<\/a><\/noindex>.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/foobar167\/articles\/blob\/master\/Ubuntu\/13_Keras_and_TensorFlow_how-tos.md#tools\">Faydal\u0131 al\u0259tl\u0259r<\/a><\/noindex>, burada h\u0259r k\u0259s \u00f6z\u00fcn\u0259 maraql\u0131 bir \u015fey tapacaq.<\/li>\n<li>Son d\u0259r\u0259c\u0259 faydal\u0131 oldu <strong>elmi m\u0259qal\u0259l\u0259rin t\u0259hlili \u00fczr\u0259 video kanallar<\/strong> Data Science. Tap\u0131n, onlara abun\u0259 olun v\u0259 ba\u011flant\u0131lar\u0131 h\u0259mkarlar\u0131n\u0131za v\u0259 m\u0259n\u0259 d\u0259 g\u00f6nd\u0259rin. N\u00fcmun\u0259l\u0259r:\n<ul>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/www.youtube.com\/user\/keeroyz\">Two Minute Papers<\/a><\/noindex><\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/www.youtube.com\/channel\/UCHB9VepY6kYvZjj0Bgxnpbw\">Henry AI Labs<\/a><\/noindex><\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/www.youtube.com\/channel\/UCZHmQk67mSJgfCCTn7xBfew\">Yannic Kilcher<\/a><\/noindex><\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/www.youtube.com\/channel\/UC5_6ZD6s8klmMu9TXEB_1IA\">CodeEmporium<\/a><\/noindex><\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/www.dlology.com\">Chengwei Zhang\u0131n blogu<\/a><\/noindex> aka <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/Tony607\">Tony607<\/a><\/noindex> add\u0131m-add\u0131m t\u0259limatlarla v\u0259 a\u00e7\u0131q m\u0259nb\u0259 kodu il\u0259.<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p><\/p>\n<p>Diqq\u0259tiniz \u00fc\u00e7\u00fcn t\u0259\u015f\u0259kk\u00fcr edir\u0259m!<\/p>\n<p>M\u0259nb\u0259: <a content=\"nofollow\" rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/482794\/\">habr.com<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"<p>\u0421\u0442\u0430\u0442\u044c\u044f \u0441\u043e\u0441\u0442\u043e\u0438\u0442 \u0438\u0437 \u0434\u0432\u0443\u0445 \u0447\u0430\u0441\u0442\u0435\u0439: \u041a\u0440\u0430\u0442\u043a\u043e\u0435 \u043e\u043f\u0438\u0441\u0430\u043d\u0438\u0435 \u043d\u0435\u043a\u043e\u0442\u043e\u0440\u044b\u0445 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440 \u0441\u0435\u0442\u0435\u0439 \u043f\u043e \u043e\u0431\u043d\u0430\u0440\u0443\u0436\u0435\u043d\u0438\u044e \u043e\u0431\u044a\u0435\u043a\u0442\u043e\u0432 \u043d\u0430 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0438 \u0438 \u0441\u0435\u0433\u043c\u0435\u043d\u0442\u0430\u0446\u0438\u0438 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 \u0441 \u0441\u0430\u043c\u044b\u043c\u0438 \u043f\u043e\u043d\u044f\u0442\u043d\u044b\u043c\u0438 \u0434\u043b\u044f \u043c\u0435\u043d\u044f \u0441\u0441\u044b\u043b\u043a\u0430\u043c\u0438 \u043d\u0430 \u0440\u0435\u0441\u0443\u0440\u0441\u044b. \u0421\u0442\u0430\u0440\u0430\u043b\u0441\u044f \u0432\u044b\u0431\u0438\u0440\u0430\u0442\u044c \u0432\u0438\u0434\u0435\u043e \u043f\u043e\u044f\u0441\u043d\u0435\u043d\u0438\u044f \u0438 \u0436\u0435\u043b\u0430\u0442\u0435\u043b\u044c\u043d\u043e \u043d\u0430 \u0440\u0443\u0441\u0441\u043a\u043e\u043c \u044f\u0437\u044b\u043a\u0435. \u0412\u0442\u043e\u0440\u0430\u044f \u0447\u0430\u0441\u0442\u044c \u0441\u043e\u0441\u0442\u043e\u0438\u0442 \u0432 \u043f\u043e\u043f\u044b\u0442\u043a\u0435 \u043e\u0441\u043e\u0437\u043d\u0430\u0442\u044c \u043d\u0430\u043f\u0440\u0430\u0432\u043b\u0435\u043d\u0438\u0435 \u0440\u0430\u0437\u0432\u0438\u0442\u0438\u044f \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440 \u043d\u0435\u0439\u0440\u043e\u043d\u043d\u044b\u0445 \u0441\u0435\u0442\u0435\u0439. \u0418 \u0442\u0435\u0445\u043d\u043e\u043b\u043e\u0433\u0438\u0439 \u043d\u0430 \u0438\u0445 \u043e\u0441\u043d\u043e\u0432\u0435. \u0420\u0438\u0441\u0443\u043d\u043e\u043a 1 \u2013 \u041f\u043e\u043d\u0438\u043c\u0430\u0442\u044c [&hellip;]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-54780","post","type-post","status-publish","format-standard","hentry","category-news"],"aioseo_notices":[],"aioseo_head":"\n\t\t<!-- All in One SEO 5.0.0.1 - aioseo.com -->\n\t<meta name=\"description\" content=\"\u0421\u0442\u0430\u0442\u044c\u044f \u0441\u043e\u0441\u0442\u043e\u0438\u0442 \u0438\u0437 \u0434\u0432\u0443\u0445 \u0447\u0430\u0441\u0442\u0435\u0439: \u041a\u0440\u0430\u0442\u043a\u043e\u0435 \u043e\u043f\u0438\u0441\u0430\u043d\u0438\u0435 \u043d\u0435\u043a\u043e\u0442\u043e\u0440\u044b\u0445 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440 \u0441\u0435\u0442\u0435\u0439 \u043f\u043e \u043e\u0431\u043d\u0430\u0440\u0443\u0436\u0435\u043d\u0438\u044e \u043e\u0431\u044a\u0435\u043a\u0442\u043e\u0432 \u043d\u0430 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0438 \u0438 \u0441\u0435\u0433\u043c\u0435\u043d\u0442\u0430\u0446\u0438\u0438 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 \u0441 \u0441\u0430\u043c\u044b\u043c\u0438 \u043f\u043e\u043d\u044f\u0442\u043d\u044b\u043c\u0438 \u0434\u043b\u044f \u043c\u0435\u043d\u044f \u0441\u0441\u044b\u043b\u043a\u0430\u043c\u0438 \u043d\u0430 \u0440\u0435\u0441\u0443\u0440\u0441\u044b.\" \/>\n\t<meta name=\"robots\" content=\"max-image-preview:large\" \/>\n\t<meta name=\"author\" content=\"Yuri Gagarin\"\/>\n\t<link rel=\"canonical\" href=\"https:\/\/prohoster.info\/az\/blog\/news\/nejroseti-kuda-eto-vse-dvizhetsya\" \/>\n\t<meta name=\"generator\" content=\"All in One SEO (AIOSEO) 5.0.0.1\" \/>\n\t\t<meta property=\"og:locale\" content=\"az_AZ\" \/>\n\t\t<meta property=\"og:site_name\" content=\"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\" \/>\n\t\t<meta property=\"og:type\" content=\"article\" \/>\n\t\t<meta property=\"og:title\" content=\"\ud83e\udd47\u041d\u0435\u0439\u0440\u043e\u0441\u0435\u0442\u0438. \u041a\u0443\u0434\u0430 \u044d\u0442\u043e \u0432\u0441\u0435 \u0434\u0432\u0438\u0436\u0435\u0442\u0441\u044f | ProHoster\" \/>\n\t\t<meta property=\"og:description\" content=\"\u0421\u0442\u0430\u0442\u044c\u044f \u0441\u043e\u0441\u0442\u043e\u0438\u0442 \u0438\u0437 \u0434\u0432\u0443\u0445 \u0447\u0430\u0441\u0442\u0435\u0439: \u041a\u0440\u0430\u0442\u043a\u043e\u0435 \u043e\u043f\u0438\u0441\u0430\u043d\u0438\u0435 \u043d\u0435\u043a\u043e\u0442\u043e\u0440\u044b\u0445 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440 \u0441\u0435\u0442\u0435\u0439 \u043f\u043e \u043e\u0431\u043d\u0430\u0440\u0443\u0436\u0435\u043d\u0438\u044e \u043e\u0431\u044a\u0435\u043a\u0442\u043e\u0432 \u043d\u0430 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0438 \u0438 \u0441\u0435\u0433\u043c\u0435\u043d\u0442\u0430\u0446\u0438\u0438 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 \u0441 \u0441\u0430\u043c\u044b\u043c\u0438 \u043f\u043e\u043d\u044f\u0442\u043d\u044b\u043c\u0438 \u0434\u043b\u044f \u043c\u0435\u043d\u044f \u0441\u0441\u044b\u043b\u043a\u0430\u043c\u0438 \u043d\u0430 \u0440\u0435\u0441\u0443\u0440\u0441\u044b.\" \/>\n\t\t<meta property=\"og:url\" content=\"https:\/\/prohoster.info\/az\/blog\/news\/nejroseti-kuda-eto-vse-dvizhetsya\" \/>\n\t\t<meta property=\"og:image\" content=\"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg\" \/>\n\t\t<meta property=\"og:image:secure_url\" content=\"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg\" \/>\n\t\t<meta property=\"og:image:width\" content=\"350\" \/>\n\t\t<meta property=\"og:image:height\" content=\"350\" \/>\n\t\t<meta property=\"article:published_time\" content=\"2020-01-03T21:00:00+00:00\" \/>\n\t\t<meta property=\"article:modified_time\" content=\"2020-02-18T11:02:50+00:00\" \/>\n\t\t<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/prohoster\" \/>\n\t\t<meta property=\"article:author\" content=\"https:\/\/www.facebook.com\/prohoster\" \/>\n\t\t<!-- All in One SEO -->\n\n","aioseo_head_json":{"title":"\ud83e\udd47Neyron \u015f\u0259b\u0259k\u0259l\u0259ri. Bu h\u0259r \u015fey hara gedir | ProHoster","description":"M\u0259qal\u0259 iki hiss\u0259d\u0259n ibar\u0259tdir: \u015e\u0259kill\u0259rd\u0259 obyektin a\u015fkar olunmas\u0131 v\u0259 \u015f\u0259kill\u0259rin seqmentasiyas\u0131 \u00fc\u00e7\u00fcn b\u0259zi \u015f\u0259b\u0259k\u0259 arxitekturalar\u0131n\u0131n q\u0131sa t\u0259sviri, m\u0259nim \u00fc\u00e7\u00fcn \u0259n ayd\u0131n resurs ba\u011flant\u0131lar\u0131 il\u0259.","canonical_url":"https:\/\/prohoster.info\/az\/blog\/news\/nejroseti-kuda-eto-vse-dvizhetsya","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"az_AZ","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\u041d\u0435\u0439\u0440\u043e\u0441\u0435\u0442\u0438. \u041a\u0443\u0434\u0430 \u044d\u0442\u043e \u0432\u0441\u0435 \u0434\u0432\u0438\u0436\u0435\u0442\u0441\u044f | ProHoster","og:description":"\u0421\u0442\u0430\u0442\u044c\u044f \u0441\u043e\u0441\u0442\u043e\u0438\u0442 \u0438\u0437 \u0434\u0432\u0443\u0445 \u0447\u0430\u0441\u0442\u0435\u0439: \u041a\u0440\u0430\u0442\u043a\u043e\u0435 \u043e\u043f\u0438\u0441\u0430\u043d\u0438\u0435 \u043d\u0435\u043a\u043e\u0442\u043e\u0440\u044b\u0445 \u0430\u0440\u0445\u0438\u0442\u0435\u043a\u0442\u0443\u0440 \u0441\u0435\u0442\u0435\u0439 \u043f\u043e \u043e\u0431\u043d\u0430\u0440\u0443\u0436\u0435\u043d\u0438\u044e \u043e\u0431\u044a\u0435\u043a\u0442\u043e\u0432 \u043d\u0430 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0438 \u0438 \u0441\u0435\u0433\u043c\u0435\u043d\u0442\u0430\u0446\u0438\u0438 \u0438\u0437\u043e\u0431\u0440\u0430\u0436\u0435\u043d\u0438\u0439 \u0441 \u0441\u0430\u043c\u044b\u043c\u0438 \u043f\u043e\u043d\u044f\u0442\u043d\u044b\u043c\u0438 \u0434\u043b\u044f \u043c\u0435\u043d\u044f \u0441\u0441\u044b\u043b\u043a\u0430\u043c\u0438 \u043d\u0430 \u0440\u0435\u0441\u0443\u0440\u0441\u044b.","og:url":"https:\/\/prohoster.info\/az\/blog\/news\/nejroseti-kuda-eto-vse-dvizhetsya","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":"2020-01-03T21:00:00+00:00","article:modified_time":"2020-02-18T11:02:50+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"54780","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":"2026-01-24 12:43:20","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-02-28 19:57:38","updated":"2026-01-24 12:43:20","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/az\/wp-json\/wp\/v2\/posts\/54780","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/az\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/az\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/az\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/az\/wp-json\/wp\/v2\/comments?post=54780"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/az\/wp-json\/wp\/v2\/posts\/54780\/revisions"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/az\/wp-json\/wp\/v2\/media?parent=54780"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/az\/wp-json\/wp\/v2\/categories?post=54780"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/az\/wp-json\/wp\/v2\/tags?post=54780"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}