{"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\/sq\/blog\/news\/nejroseti-kuda-eto-vse-dvizhetsya","title":{"rendered":"Rrjetet nervore. Ku po shkon gjith\u00eb kjo","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"<p>Artikulli p\u00ebrb\u00ebhet nga dy pjes\u00eb:<\/p>\n<p><\/p>\n<ol>\n<li>Nj\u00eb p\u00ebrshkrim i shpejt\u00eb i disa arkitekturave t\u00eb rrjeteve p\u00ebr zbulimin e objekteve n\u00eb imazhe dhe segmentimin e imazheve me lidhjet m\u00eb t\u00eb kuptueshme p\u00ebr mua. P\u00ebrpiqesha t\u00eb zgjidhja video shpjeguese dhe p\u00ebr sa m\u00eb shum\u00eb n\u00eb gjuh\u00ebn shqipe.<\/li>\n<li>Pjesa e dyt\u00eb \u00ebsht\u00eb p\u00ebrpjekja p\u00ebr t\u00eb kuptuar drejtimin e zhvillimit t\u00eb arkitekturave t\u00eb rrjeteve neuronale dhe teknologjive mbi baz\u00ebn e tyre.<\/li>\n<\/ol>\n<p><\/p>\n<p><img decoding=\"async\" alt=\"Rrjetet nervore. Ku po shkon gjith\u00eb kjo\" src=\"\/wp-content\/uploads\/2020\/01\/3e0238547dc956dbb069c11241e4534f.png\" style=\"display:block;margin: 0 auto;\" \/><\/p>\n<p><\/p>\n<p>Figur 1 \u2013 Nuk \u00ebsht\u00eb e leht\u00eb t\u00eb kuptosh arkitekturat e rrjeteve neuronale<\/p>\n<p><\/p>\n<p>T\u00eb gjitha filluan kur b\u00ebra dy aplikacione demonstruese p\u00ebr klasifikimin dhe zbulimin e objekteve n\u00eb telefonin Android:<\/p>\n<p><\/p>\n<ul>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/foobar167\/junkyard\/tree\/master\/object_classifier\">Demo fenomeni<\/a><\/noindex>, n\u00eb t\u00eb cilin t\u00eb dh\u00ebnat p\u00ebrpunohen n\u00eb server dhe d\u00ebrgohen n\u00eb telefon. Klasifikimi i imazheve (image classification) t\u00eb tre llojeve t\u00eb medveve: t\u00eb bardh\u00eb, t\u00eb zez\u00eb dhe lod\u00ebr.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/foobar167\/android\/tree\/master\/object_detection_demo\">Demo n\u00eb front<\/a><\/noindex>, n\u00eb t\u00eb cilin t\u00eb dh\u00ebnat p\u00ebrpunohen n\u00eb telefonin vet. Zbulimi i objekteve (object detection) t\u00eb tre llojeve: lajthi, hurm\u00eb dhe datelin\u00eb.<\/li>\n<\/ul>\n<p><noindex><a rel=\"nofollow\" name=\"habracut\"><\/a><\/noindex><\/p>\n<p>Ka nj\u00eb dallim midis detyrave t\u00eb klasifikimit t\u00eb imazheve, zbulimit t\u00eb objekteve n\u00eb nj\u00eb imazh dhe <noindex><a rel=\"nofollow\" href=\"https:\/\/medium.com\/analytics-vidhya\/image-classification-vs-object-detection-vs-image-segmentation-f36db85fe81\">segmentimit t\u00eb imazheve<\/a><\/noindex>. Prandaj, erdhi nevoja p\u00ebr t\u00eb m\u00ebsuar se cilat arkitektura t\u00eb rrjeteve neuronale zbulojn\u00eb objekte n\u00eb imazhe dhe cilat mund t\u00eb segmentejn\u00eb. Gjeta shembuj t\u00eb m\u00ebposht\u00ebm t\u00eb arkitekturave me lidhjet m\u00eb t\u00eb kuptueshme p\u00ebr mua:<\/p>\n<p><\/p>\n<ul>\n<li>Nj\u00eb seri arkitekturash t\u00eb bazuara n\u00eb R-CNN (<strong>R<\/strong>regjione me <strong>C<\/strong>rrjete neuronale karakteristikash): R-CNN, Fast R-CNN, <strong>N<\/strong>Faster R-CNN <strong>N<\/strong>Mask R-CNN <noindex><a rel=\"nofollow\" href=\"https:\/\/medium.com\/@smallfishbigsea\/faster-r-cnn-explained-864d4fb7e3f8\">. P\u00ebr zbulimin e objekteve n\u00eb nj\u00eb imazh, p\u00ebrmes mekanizmit Region Proposal Network (RPN), dallohet rajoni i kufizuar (bounding boxes). Fillimisht, n\u00eb vend t\u00eb RPN u p\u00ebrdor nj\u00eb mekaniz\u00ebm m\u00eb i ngadalsh\u00ebm, k\u00ebrkimi i selektiv. M\u00eb pas, rajonet e kufizuara t\u00eb dalluara u jepen si input n\u00eb nj\u00eb rrjet neuronal t\u00eb zakonsh\u00ebm p\u00ebr klasifikim. N\u00eb arkitektur\u00ebn R-CNN ka cikle t\u00eb dukshme \"for\" p\u00ebr t\u00eb kaluar n\u00ebp\u00ebr rajonet e kufizuara, total deri n\u00eb 2000 kalime p\u00ebrmes rrjetit t\u00eb brendsh\u00ebm AlexNet. P\u00ebr shkak t\u00eb cikleve t\u00eb dukshme \"for\", ngadal\u00ebsohet shpejt\u00ebsia e p\u00ebrpunimit t\u00eb imazheve. Numri i cikleve t\u00eb dukshme, kalimeve p\u00ebrmes rrjetit t\u00eb brendsh\u00ebm t\u00eb neuronave, zvog\u00eblohet me \u00e7do version t\u00eb ri t\u00eb arkitektur\u00ebs, si dhe b\u00ebhen dhjet\u00ebra ndryshime t\u00eb tjera p\u00ebr t\u00eb rritur shpejt\u00ebsin\u00eb dhe p\u00ebr t\u00eb z\u00ebvend\u00ebsuar detyr\u00ebn e zbulimit t\u00eb objekteve me segmentimin e objekteve n\u00eb Mask R-CNN.<\/a><\/noindex>, <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/0vt05rQqk_I\">YOLO<\/a><\/noindex>do<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/L0tzmv--CGY\">t\u00eb<\/a><\/noindex> (<strong>Y<\/strong>shikojm\u00eb <strong>O<\/strong>vet\u00ebm <strong>L<\/strong>at\u00eb <strong>O<\/strong>nce) \u2013 rrjeti i par\u00eb nervor q\u00eb njihej objekte n\u00eb koh\u00eb reale n\u00eb pajisjet mobile. Karakteristika e tij dalluese: diferencimi i objekteve me nj\u00eb kalim (mjafton t\u00eb shikosh nj\u00eb her\u00eb). K\u00ebshtu, n\u00eb arkitektur\u00ebn YOLO nuk ka cikle t\u00eb dukshme \u00abfor\u00bb, \u00e7ka b\u00ebn q\u00eb rrjeti t\u00eb punoj\u00eb shpejt. P\u00ebr shembull, nj\u00eb krahasim: n\u00eb NumPy gjat\u00eb operacioneve me matrica gjithashtu nuk ka cikle t\u00eb dukshme \u00abfor\u00bb, q\u00eb n\u00eb NumPy realizohen n\u00eb nivele m\u00eb t\u00eb ul\u00ebta t\u00eb arkitektur\u00ebs p\u00ebrmes gjuh\u00ebs s\u00eb programimit C. YOLO p\u00ebrdor nj\u00eb rrjet t\u00eb caktuar me dritare. P\u00ebr t\u00eb mos e njohur t\u00eb nj\u00ebjtin objekt shum\u00eb her\u00eb, p\u00ebrdoret koeficienti i mbulimit t\u00eb dritareve (IoU, <strong>I<\/strong>ntersection <strong>o<\/strong>ver <strong>U<\/strong>nion). Kjo arkitektur\u00eb punon n\u00eb nj\u00eb gam\u00eb t\u00eb gjer\u00eb dhe ka nj\u00eb <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\">robust\u00ebsi<\/a><\/noindex>: modeli mund t\u00eb st\u00ebrvitet mbi fotografi, por gjithashtu funksionon mir\u00eb mbi piktura t\u00eb vizatuara.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/P8e-G-Mhx4k\">SSD<\/a><\/noindex> (<strong>S<\/strong>ingle <strong>S<\/strong>hot MultiBox <strong>D<\/strong>etector) \u2013 p\u00ebrdoren \u00abhack\u00bb-et m\u00eb t\u00eb suksesshme t\u00eb arkitektur\u00ebs YOLO (p\u00ebr shembull, non-maximum suppression) dhe shtohen t\u00eb reja, q\u00eb rrjeti nervor t\u00eb punoj\u00eb m\u00eb shpejt dhe m\u00eb sakt\u00eb. Karakteristika e tij dalluese: diferencimi i objekteve me nj\u00eb kalim p\u00ebrmes nj\u00eb rrjeti dritare t\u00eb caktuar (default box) n\u00eb piramid\u00ebn e imazheve. Piramida e imazheve \u00ebsht\u00eb koduar n\u00eb tenzor\u00eb konvencional gjat\u00eb operacioneve t\u00eb radh\u00ebs t\u00eb konvencioneve dhe pooling (gjat\u00eb operacionit max-pooling, dimensioni hap\u00ebsinor zvog\u00eblohet). K\u00ebshtu p\u00ebrcaktohen si objekte t\u00eb m\u00ebdha ashtu edhe t\u00eb vogla me nj\u00eb kalim t\u00eb rrjetit.<\/li>\n<li>MobileSSD (<strong>Mobile<\/strong>NetV2 + <strong>SSD<\/strong>) \u2013 kombinim i dy arkitekturave t\u00eb rrjeteve nervore. Rrjeti i par\u00eb <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/352804\/\">MobileNetV2<\/a><\/noindex> punon shpejt dhe rrit sakt\u00ebsin\u00eb e njohjes. MobileNetV2 z\u00ebvend\u00ebson VGG-16, e cila p\u00ebrdorej fillimisht n\u00eb <noindex><a rel=\"nofollow\" href=\"https:\/\/arxiv.org\/abs\/1512.02325\">artikullit origjinal<\/a><\/noindex>. Rrjeti i dyt\u00eb SSD p\u00ebrcakton vendndodhjen e objekteve n\u00eb imazh.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/ge_RT5wvHvY\">SqueezeNet<\/a><\/noindex> \u2013 nj\u00eb rrjet nervor shum\u00eb t\u00eb vog\u00ebl, por t\u00eb sakt\u00eb. Vet\u00eb nuk zgjidh problemin e identifikimit t\u00eb objekteve. Megjithat\u00eb, mund t\u00eb p\u00ebrdoret n\u00eb kombinim me arkitektura t\u00eb ndryshme. Dhe p\u00ebrdoret n\u00eb pajisje mobile. Karakteristika e tij \u00ebsht\u00eb se s\u00eb pari t\u00eb dh\u00ebnat kompresohen n\u00eb kat\u00ebr filtra konvencionale 1\u00d71, dhe m\u00eb pas zgjaten n\u00eb kat\u00ebr 1\u00d71 dhe kat\u00ebr 3\u00d73 filtra konvencional\u00eb. Nj\u00eb iteracion i till\u00eb i kompresimi-zgjerimi quhet \u00abFire Module\u00bb.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/b6jhopSMit8\">DeepLab<\/a><\/noindex> (Segmentationi e Figurave Semantike me Rrjetat e Thella Konvencionale) \u2013 segmentimi i objekteve n\u00eb imazh. Karakteristika dalluese e arkitektur\u00ebs \u00ebsht\u00eb konvulucioni me hap\u00ebsir\u00eb t\u00eb zgjeruar (dilated convolution), i cili ruan rezolut\u00ebn hap\u00ebsinore. Pas k\u00ebsaj shkon nj\u00eb faz\u00eb e p\u00ebrpunimit post, duke p\u00ebrdorur nj\u00eb model probabilistik grafik (conditional random field), q\u00eb lejon eliminimin e zhurmave t\u00eb vogla n\u00eb segmentim dhe p\u00ebrmir\u00ebsimin e cil\u00ebsis\u00eb s\u00eb imazhit t\u00eb segmentuar. Pas emrit t\u00eb friksh\u00ebm \u00abmodeli probabilistik grafik\u00bb fshihet nj\u00eb filt\u00ebr i zakonsh\u00ebm Gauss, i cili \u00ebsht\u00eb aproksimuar mbi pes\u00eb pika.<\/li>\n<li>P\u00ebrpiqesha t\u00eb kuptoja struktur\u00ebn <noindex><a rel=\"nofollow\" href=\"https:\/\/arxiv.org\/abs\/1711.06897\">RefineDet<\/a><\/noindex> (Single-Shot <strong>Refine<\/strong>ment Neural Network for Object <strong>Det<\/strong>ection), por nuk kuptova shum\u00eb.<\/li>\n<li>Gjithashtu shikoja se si funksionon teknologjia \u00abv\u00ebmendje\u00bb (attention): <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>. Karakteristika dalluese e arkitektur\u00ebs \u00abv\u00ebmendje\u00bb \u00ebsht\u00eb identifikimi automatik i rajoneve me v\u00ebmendje t\u00eb lart\u00eb n\u00eb imazh (RoI, <strong>R<\/strong>egions <strong>o<\/strong>f <strong>I<\/strong>nterest) me ndihm\u00ebn e nj\u00eb rrjete neurale t\u00eb quajtur Nj\u00ebsi V\u00ebmendjeje. Rajonet e v\u00ebmendjes jan\u00eb t\u00eb ngjashme me rajonet e kufizuara (bounding boxes), por ndryshe nga ato, nuk jan\u00eb t\u00eb fiksuara n\u00eb imazh dhe mund t\u00eb ken\u00eb kufij t\u00eb mjegullt. Pas k\u00ebsaj, nga rajonet e v\u00ebmendjes, nxirren tiparet (features), t\u00eb cilat i \u00abjepen\u00bb rrjeteve neurale recurente me arkitektura <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/5lUUrREboSk\">LSDM, GRU ose Vanilla RNN<\/a><\/noindex>. Rrjetet neurale recurente din\u00eb t\u00eb analizojn\u00eb marr\u00ebdh\u00ebniet e tipareve n\u00eb nj\u00eb renditje. Rrjetet neurale recurente fillimisht ishin p\u00ebrdorur p\u00ebr p\u00ebrkthimet e tekstit n\u00eb gjuh\u00eb t\u00eb tjera, dhe tani edhe p\u00ebr p\u00ebrkthimin <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/e-WB4lfg30M\">nga imazhi n\u00eb tekst<\/a><\/noindex> dhe <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/rAbhypxs1qQ\">nga teksti n\u00eb imazh<\/a><\/noindex>.<\/li>\n<\/ul>\n<p><\/p>\n<p>Me kalimin e koh\u00ebs duke studiuar k\u00ebto arkitektura <strong>kuptova se nuk kuptoja asgj\u00eb.<\/strong>. Dhe problemi nuk \u00ebsht\u00eb se rrjeta ime neurale ka probleme me mekanizmin e v\u00ebmendjes. Krijimi i t\u00eb gjitha k\u00ebtyre arkitekturave ngjan me nj\u00eb hackathon t\u00eb madh, ku autor\u00ebt garojn\u00eb me hacks. Hack \u2013 nj\u00eb zgjidhje e shpejt\u00eb p\u00ebr nj\u00eb problem t\u00eb v\u00ebshtir\u00eb programimi. Dometh\u00ebn\u00eb, nd\u00ebrmjet t\u00eb gjith\u00eb k\u00ebtyre arkitekturave nuk ka nj\u00eb lidhje logjike t\u00eb dukshme dhe t\u00eb kuptueshme. E vetmja gj\u00eb q\u00eb i bashkon ato \u00ebsht\u00eb nj\u00eb grup hacks-i m\u00eb t\u00eb suksessh\u00ebm, t\u00eb cilat i huazojn\u00eb nga nj\u00ebri-tjetri, plus nj\u00eb operacion i p\u00ebrbashk\u00ebt p\u00ebr t\u00eb gjith\u00eb <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/Ilg3gGewQ5U\">konvolucioni me reagim<\/a><\/noindex> (p\u00ebrhapja e gabimit prapa, backpropagation). Nuk ka <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/272473\/\">mendim sistemor<\/a><\/noindex>! Nuk \u00ebsht\u00eb e qart\u00eb se \u00e7far\u00eb duhet t\u00eb ndryshojm\u00eb dhe si t\u00eb optimizojm\u00eb arritjet ekzistuese.<\/p>\n<p><\/p>\n<p>Si rezultat i munges\u00ebs s\u00eb lidhjes logjike midis haks, ato jan\u00eb jasht\u00ebzakonisht t\u00eb v\u00ebshtira p\u00ebr t'u mbajtur mend dhe p\u00ebr t'u aplikuar n\u00eb praktik\u00eb. Kjo \u00ebsht\u00eb njohuri e fragmentuar. N\u00eb m\u00eb t\u00eb mir\u00ebn, disa pika interesante dhe t\u00eb papritura mbahen mend, por shumica e asaj q\u00eb \u00ebsht\u00eb kuptuar dhe e paqart\u00eb zhduket nga memorie pas disa dit\u00ebsh. Do t\u00eb ishte e mir\u00eb n\u00ebse pas nj\u00eb jave do t\u00eb kujtohej t\u00eb pakt\u00ebn emri i arkitektur\u00ebs. Nd\u00ebrsa disa or\u00eb dhe madje dit\u00eb kohe pune jan\u00eb shpenzuar p\u00ebr leximin e artikujve dhe shikimin e videove p\u00ebrmbledh\u00ebse!<\/p>\n<p><\/p>\n<p><img decoding=\"async\" alt=\"Rrjetet nervore. Ku po shkon gjith\u00eb kjo\" src=\"\/wp-content\/uploads\/2020\/01\/86f0f24e3be1d0f1a210f3132897d981.png\" style=\"display:block;margin: 0 auto;\" \/><\/p>\n<p><\/p>\n<p>Figura 2 \u2013 <noindex><a rel=\"nofollow\" href=\"https:\/\/www.asimovinstitute.org\/neural-network-zoo\/\">Zooja e rrjeteve nervore<\/a><\/noindex><\/p>\n<p><\/p>\n<p>Shumica e autor\u00ebve t\u00eb artikujve shkencor\u00eb, sipas mendimit tim personal, b\u00ebjn\u00eb gjith\u00e7ka t\u00eb mundshme q\u00eb edhe k\u00ebto njohuri t\u00eb fragmentuara t\u00eb mos kuptohen nga lexuesi. Por p\u00ebrdorimi i gerundit n\u00eb dhjet\u00eb fjali t\u00eb rreshtuar me formula q\u00eb jan\u00eb marr\u00eb \"nga tavan\" \u2013 kjo \u00ebsht\u00eb nj\u00eb tem\u00eb p\u00ebr nj\u00eb artikull t\u00eb ve\u00e7ant\u00eb (problemi <noindex><a rel=\"nofollow\" href=\"https:\/\/en.wikipedia.org\/wiki\/Publish_or_perish\">publish or perish<\/a><\/noindex>).<\/p>\n<p><\/p>\n<p>P\u00ebr k\u00ebt\u00eb arsye u b\u00eb e nevojshme sistematizimi i informacionit rreth rrjeteve nervore dhe, n\u00eb k\u00ebt\u00eb m\u00ebnyr\u00eb, rritja e cil\u00ebsis\u00eb s\u00eb kuptimit dhe mbajtjes mend. Prandaj, tema kryesore e analiz\u00ebs s\u00eb teknologjive dhe arkitekturave t\u00eb ve\u00e7anta t\u00eb rrjeteve nervore artificiale u b\u00eb kjo detyr\u00eb: <strong>t\u00eb m\u00ebsojm\u00eb se ku po shkon gjith\u00e7ka<\/strong>, dhe jo struktura e ndonj\u00eb rrjeti nervor specifik.<\/p>\n<p><\/p>\n<p>Ku po shkon gjith\u00e7ka. Rezultatet kryesore:<\/p>\n<p><\/p>\n<ul>\n<li>Numri i startupeve n\u00eb fush\u00ebn e m\u00ebsimit t\u00eb makinerive n\u00eb dy vitet e fundit <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/company\/recognitor\/blog\/455676\/\">ka r\u00ebn\u00eb ndjesh\u00ebm<\/a><\/noindex>. Shkaku i mundsh\u00ebm: \"rrjetet nervore nuk jan\u00eb m\u00eb di\u00e7ka e re\".<\/li>\n<li>\u00c7do njeri mund t\u00eb krijoj\u00eb nj\u00eb rrjet nervor funksional p\u00ebr zgjidhjen e nj\u00eb detyre t\u00eb thjesht\u00eb. P\u00ebr k\u00ebt\u00eb, do t\u00eb marr\u00eb nj\u00eb model t\u00eb gatsh\u00ebm nga \"zooja e modeleve\" (model zoo) dhe do t\u00eb trajnoj\u00eb shtyll\u00ebn e fundit t\u00eb rrjetit nervor (<noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/yofjFQddwHE\">transfer learning<\/a><\/noindex>) mbi t\u00eb dh\u00ebnat e gatshme nga <noindex><a rel=\"nofollow\" href=\"https:\/\/toolbox.google.com\/datasetsearch\">K\u00ebrkimi i Dataset-eve Google<\/a><\/noindex> apo nga <noindex><a rel=\"nofollow\" href=\"https:\/\/www.kaggle.com\/datasets\">25,000 dataset-e t\u00eb Kaggle<\/a><\/noindex> n\u00eb <noindex><a rel=\"nofollow\" href=\"https:\/\/www.dataschool.io\/cloud-services-for-jupyter-notebook\/\">repartin falas Jupyter Notebook<\/a><\/noindex>.<\/li>\n<li>Prodhuesit e m\u00ebdhenj t\u00eb rrjeteve nervore filluan t\u00eb krijojn\u00eb <strong>\"zoopark\u00eb modelesh\"<\/strong> (model zoo). Me ndihm\u00ebn e tyre, mund t\u00eb krijoni shpejt nj\u00eb aplikacion komercial: <noindex><a rel=\"nofollow\" href=\"https:\/\/tfhub.dev\/\">TF Hub<\/a><\/noindex> p\u00ebr TensorFlow, <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/open-mmlab\/mmdetection\">MMDetection<\/a><\/noindex> p\u00ebr PyTorch, <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/facebookresearch\/Detectron\">Detectron<\/a><\/noindex> p\u00ebr Caffe2, <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/wkentaro\/chainer-modelzoo\">chainer-modelzoo<\/a><\/noindex> p\u00ebr Chainer dhe <noindex><a rel=\"nofollow\" href=\"https:\/\/modelzoo.co\/\">t\u00eb tjera<\/a><\/noindex>.<\/li>\n<li>Rrjetet nervore q\u00eb punojn\u00eb n\u00eb <strong>koh\u00eb reale<\/strong> (real-time) n\u00eb pajisjet mobile. Nga 10 deri n\u00eb 50 kadra n\u00eb sekond\u00eb.<\/li>\n<li>P\u00ebrdorimi i rrjeteve nervore n\u00eb telefona (TF Lite), n\u00eb shfletues (TF.js) dhe n\u00eb <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/19ZNz2N79u4\">objekte t\u00eb p\u00ebrdorimit t\u00eb p\u00ebrditsh\u00ebm<\/a><\/noindex> (IoT, <strong>I<\/strong>internet <strong>o<\/strong>f <strong>T<\/strong>i gj\u00ebrave). Sidomos n\u00eb telefona, t\u00eb cil\u00ebt tashm\u00eb mb\u00ebshtesin rrjetet nervore n\u00eb nivelin \"harduer\" (neuro-accelerators).<\/li>\n<li>\u00ab\u00c7do pajisje, artikuj veshjesh dhe, ndoshta, edhe ushqimi do t\u00eb ket\u00eb <strong>adres\u00eb IP-v6<\/strong> dhe do t\u00eb komunikojn\u00eb mes tyre\u00bb \u2013 <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/GG7H8Xa4m8I?t=85\">Sebastian Thrun<\/a><\/noindex>.<\/li>\n<li>Rritja e numrit t\u00eb publikimeve mbi m\u00ebsimin nga makina ka filluar <noindex><a rel=\"nofollow\" href=\"http:\/\/data-mining.philippe-fournier-viger.com\/too-many-machine-learning-papers\">t\u00eb tejkaloj\u00eb ligjin e Moore-it<\/a><\/noindex> (dyfishimi \u00e7do dy vjet) q\u00eb nga viti 2015. \u00cbsht\u00eb e qart\u00eb se duhen rrjete nervore p\u00ebr analiz\u00ebn e artikujve.<\/li>\n<li>Teknologjit\u00eb q\u00eb po fitosin popullaritet jan\u00eb:\n<ul>\n<li><strong>PyTorch<\/strong> \u2013 popullariteti po rritet me shpejt\u00ebsi dhe duket se po kalon TensorFlow.<\/li>\n<li>P\u00ebrzgjedhja automatike e hiperparametrave <strong>AutoML<\/strong> \u2013 popullariteti po rritet gradualisht.<\/li>\n<li>K\u00ebtu po ndodh nj\u00eb zvog\u00eblim gradual i sakt\u00ebsis\u00eb dhe nj\u00eb rritje e shpejt\u00ebsis\u00eb s\u00eb llogaritjeve: <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/rln_kZbYaWc\">logjika fuzzy<\/a><\/noindex>, algoritmet <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/MIPkK5ZAsms\">e boosting<\/a><\/noindex>, llogaritjet e pasakta (t\u00eb af\u00ebrta), kvantizimi (kur peshat e rrjetit nervor shnd\u00ebrrohen n\u00eb numra t\u00eb plot\u00eb dhe kvantizohen), neuroakselerator\u00ebt.<\/li>\n<li>P\u00ebrkthimi <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/e-WB4lfg30M\">nga imazhi n\u00eb tekst<\/a><\/noindex> dhe <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/rAbhypxs1qQ\">nga teksti n\u00eb imazh<\/a><\/noindex>.<\/li>\n<li>Krijimi <noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/OrHLacCDZVQ\">objekte tridimensionale p\u00ebrmes videos<\/a><\/noindex>, tani n\u00eb koh\u00eb reale.<\/li>\n<li>E r\u00ebnd\u00ebsishme n\u00eb DL \u00ebsht\u00eb se ka shum\u00eb t\u00eb dh\u00ebna, por mbledhja dhe etiketimi i tyre nuk \u00ebsht\u00eb e leht\u00eb. K\u00ebshtu q\u00eb po zhvillohet automatizimi i etiketimit (<noindex><a rel=\"nofollow\" href=\"https:\/\/youtu.be\/NcKTn4C91Yc\">automated annotation<\/a><\/noindex>) p\u00ebr rrjetet nervore me ndihm\u00ebn e rrjeteve nervore.<\/li>\n<\/ul>\n<\/li>\n<li>Me rrjetet nervore, Informatika papritur u b\u00eb <strong>nj\u00eb shkenc\u00eb eksperimentale<\/strong> dhe ndodhi <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/480348\">kriza e riprodhueshm\u00ebris\u00eb<\/a><\/noindex>.<\/li>\n<li>Parat\u00eb n\u00eb IT dhe popullariteti i rrjeteve nervore i jan\u00eb p\u00ebrgjigjur nj\u00ebkoh\u00ebsisht, kur llogaritjet u b\u00ebn\u00eb vler\u00eb tregu. Ekonomia nga e art\u00eb-depozit\u00eb po b\u00ebhet <strong>e art\u00eb-depozit\u00eb-llogarit\u00ebse.<\/strong>. Shikoni artikullin tim mbi <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\">eko-fizik\u00ebn<\/a><\/noindex> dhe arsyen e shfaqjes s\u00eb parave n\u00eb IT.<\/li>\n<\/ul>\n<p><\/p>\n<p>Gradualisht po shfaqet nj\u00eb <noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/481844\">metodologji e re e programimit ML\/DL<\/a><\/noindex> (M\u00ebsimi nga makina &amp; M\u00ebsimi i thell\u00eb), e cila bazohet n\u00eb p\u00ebrfaq\u00ebsimin e programit si nj\u00eb grumbull modelesh t\u00eb m\u00ebsuara nervore.<\/p>\n<p><\/p>\n<p><img decoding=\"async\" alt=\"Rrjetet nervore. Ku po shkon gjith\u00eb kjo\" src=\"\/wp-content\/uploads\/2020\/01\/17df76a613191522145abeba177c2685.png\" style=\"display:block;margin: 0 auto;\" \/><\/p>\n<p><\/p>\n<p>Figura 3 \u2013 ML\/DL si nj\u00eb metodologji e re programimi<\/p>\n<p><\/p>\n<p>Megjithat\u00eb, nuk ka ndodhur <strong>\u00abteoria e rrjeteve nervore\u00bb<\/strong>, n\u00eb kuadrin e s\u00eb cil\u00ebs mund t\u00eb mendohet dhe t\u00eb punohet sistematikisht. Ajo q\u00eb tani quhet \u00abteori\u00bb n\u00eb t\u00eb v\u00ebrtet\u00eb jan\u00eb algoritme eksperimentale, heuristike.<\/p>\n<p><\/p>\n<p>Referencat n\u00eb burimet e mia dhe jo vet\u00ebm:<\/p>\n<p><\/p>\n<ul>\n<li>Grupi i lajmeve p\u00ebr Data Science. Kryesisht, p\u00ebr p\u00ebrpunimin e imazheve. Kush d\u00ebshiron t\u00eb marr\u00eb, le t\u00eb d\u00ebrgoj\u00eb e-mail (foobar167gmailcom). Referencat n\u00eb artikuj dhe video d\u00ebrgohen sipas grumbullimit t\u00eb materialit.<\/li>\n<li>Lista e p\u00ebrgjithshme <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/foobar167\/articles\/blob\/master\/Machine_Learning\/courses_on_machine_learning.md\">e kurseve dhe artikujve<\/a><\/noindex>, t\u00eb cilat kam p\u00ebrfunduar dhe q\u00eb do t\u00eb doja t\u00eb p\u00ebrfundova.<\/li>\n<li><noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/foobar167\/articles\/blob\/master\/Ubuntu\/13_Keras_and_TensorFlow_how-tos.md#exercises\">Kurs dhe video p\u00ebr fillestar\u00ebt<\/a><\/noindex>, me t\u00eb cilat duhet t\u00eb filloni t\u00eb studioni rrjetet nervore. P\u00ebrve\u00e7 k\u00ebsaj, broshur\u00eb <noindex><a rel=\"nofollow\" href=\"https:\/\/foobar167.github.io\/page\/vvedeniye-v-mashinnoye-obucheniye-i-iskusstvennyye-neyronnyye-seti.html\">\u00abHyrje n\u00eb m\u00ebsimin nga makina dhe rrjetet nervore artificiale\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\">Mjetet e dobishme<\/a><\/noindex>, ku \u00e7do njeri gjen di\u00e7ka interesante p\u00ebr vete.<\/li>\n<li>Kan\u00eb qen\u00eb jasht\u00ebzakonisht t\u00eb dobishme <strong>kanalet video p\u00ebr analizimin e artikujve shkencor\u00eb<\/strong> n\u00eb Data Science. Gjeni, abonohuni n\u00eb to dhe ndani lidhjet me koleg\u00ebt tuaja dhe me mua gjithashtu. Shembuj:\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\">Blogu i Chengwei Zhang<\/a><\/noindex> si <noindex><a rel=\"nofollow\" href=\"https:\/\/github.com\/Tony607\">Tony607<\/a><\/noindex> me udh\u00ebzime hapi-hapi dhe kodin burimor t\u00eb hapur.<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<p><\/p>\n<p>Faleminderit p\u00ebr v\u00ebmendjen!<\/p>\n<p>Burimi: <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 5.0.2 - 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\/sq\/blog\/news\/nejroseti-kuda-eto-vse-dvizhetsya\" \/>\n\t<meta name=\"generator\" content=\"All in One SEO (AIOSEO) 5.0.2\" \/>\n\t\t<meta property=\"og:locale\" content=\"sq_AL\" \/>\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\/sq\/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\udd47Rrjetet neuronale. Ku po shkon gjith\u00e7ka | ProHoster","description":"Artikulli p\u00ebrb\u00ebhet nga dy pjes\u00eb: Nj\u00eb p\u00ebrmbledhje e disa arkitekturave t\u00eb rrjeteve p\u00ebr zbulimin e objekteve n\u00eb imazhe dhe segmentimin e imazheve me lidhjet m\u00eb t\u00eb qarta p\u00ebr mua n\u00eb burime.","canonical_url":"https:\/\/prohoster.info\/sq\/blog\/news\/nejroseti-kuda-eto-vse-dvizhetsya","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"sq_AL","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\/sq\/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\/sq\/wp-json\/wp\/v2\/posts\/54780","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/comments?post=54780"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts\/54780\/revisions"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/media?parent=54780"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/categories?post=54780"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/tags?post=54780"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}