Halkee loo aaday: dhacdooyinka bilaashka ah ee soo socda ee xirfadlayaasha IT ee Moscow (Janaayo 14-18)

Dhacdooyinka diiwaangelinta furan:

• AI & Mobile
• NeurIPS Xaflada Sanadka Cusub
• R Moscow Kulanka #5
• Kulankii hore ee Moscow (Gastromarket Balchug)
• Tababarka barashada mashiinka ee Avito 2.0

AI & Mobile

Janaayo 14, 19:00-22:00, Talaado

Waxaan kugu martiqaadeynaa kulan ku saabsan sirdoonka macmal, codsigiisa aaladaha mobilada iyo isbeddelada tignoolajiyada iyo ganacsiga ugu muhiimsan ee tobanka sano ee cusub. Barnaamijka waxaa ku jira warbixino xiiso leh, doodo, pizza iyo niyad wanaag.

Mid ka mid ah kuwa ku hadla ayaa ah hormuudka soo bandhigida tignoolajiyada ugu dambeeyay ee Hollywood, Aqalka Cad; buugiisa "La kordhiyey: Life in the Smart Lane" waxaa lagu sheegay mid ka mid ah buugaagta tixraaca ee uu jecel yahay ee Madaxweynaha Shiinaha ee cinwaankiisa sanadka cusub.

NeurIPS Xaflada Sanadka Cusub

Janaayo 15, laga bilaabo 18:00, Arbacada

• 18:00 Diiwangelinta
• 19:00 Furitaanka - Mikhail Bilenko, Yandex
• 19:05 Xoojinta barashada NeurIPS 2019: siday ahayd - Sergey Kolesnikov, TinkoffSannad kasta mawduuca barashada xoojinta (RL) ayaa noqonaysa mid kulul oo aad loo buunbuuninayo. Sannad kasta, DeepMind iyo OpenAI waxay ku daraan shidaalka dabka iyagoo siidaaya bot waxqabadka ka sarreeya ee bini'aadamka. Ma jiraan wax runtii tan ka dambeeya? Oo waa maxay isbeddellada ugu dambeeyay ee dhammaan kala duwanaanta RL? Aan ogaano!
• 19:25 Dib u eegida shaqada NLP ee NeurIPS 2019 - Mikhail Burtsev, MIPTMaanta, isbeddellada ugu horumarsan ee dhinaca habaynta luqadda dabiiciga ah waxay la xidhiidhaan dhisidda qaab-dhismeedyo ku salaysan qaababka luqadda iyo garaafyada aqoonta. Warbixintu waxay bixin doontaa dulmar ku saabsan shaqooyinka hababkan loo isticmaalo in lagu dhiso hababka wada hadalka si loo fuliyo hawlo kala duwan. Tusaale ahaan, wada-xidhiidhka mawduucyada guud, kordhinta dareenka iyo qabashada wadahadal ku jihaysan hadafka.
• 19:45 Siyaabaha lagu fahmo nooca dusha sare ee shaqada khasaaraha - Dmitry Vetrov, Kulliyadda Sayniska Kombiyuutarka, Jaamacadda Cilmi-baarista Qaranka Dugsiga Sare ee DhaqaalahaWaxaan ka hadli doonaa dhowr waraaqo oo sahamin doona saamaynta aan caadiga ahayn ee barashada qoto dheer. Saameynahani waxay iftiiminayaan iftiinka muuqaalka kore ee shaqada luminta ee booska miisaanka waxayna noo ogolaaneysaa inaan soo bandhigno tiro fikrado ah. Haddii la xaqiijiyo, waxay suurtogal noqon doontaa in si wax ku ool ah loo habeeyo cabbirka tallaabada hababka hagaajinta. Tani waxay sidoo kale suurtogal ka dhigi doontaa in la saadaaliyo qiimaha la gaari karo ee shaqada khasaaraha ee muunadda tijaabada muddo dheer ka hor dhammaadka tababarka.
• 20:05 Dib u eegis ku saabsan shaqooyinka ku saabsan aragtida kombuyuutarka ee NeurIPS 2019 - Sergey Ovcharenko, Konstantin Lakhman, YandexWaxaan eegi doonaa meelaha ugu muhiimsan ee cilmi-baarista iyo ka shaqeynta aragtida kombuyuutarka. Aan isku dayno inaan fahanno in dhibkii oo dhan laga soo saaray dhanka akadeemiyada, in socodkii guusha ee GAN uu ku sii socdo meel walba, yaa ka hortagaya, iyo goorta uu dhici doono kacaanka aan laga warhayn.
• 20:25 nasashada kafeega
• 20:40 Qaabaynta taxanaha oo leh nidaam aan xadidnayn oo jiil - Dmitry Emelianenko, YandexWaxaan soo jeedinaynaa qaab gelin kara ereyada meelo aan loo meel dayin jumlada la soo saaray. Qaabku wuxuu si badheedh ah u bartaa hab ku habboon oo lagu kala furfuro iyadoo lagu salaynayo xogta. Tayada ugu wanagsan waxaa lagu gaaraa dhowr xog oo kala duwan: tarjumaada mashiinka, isticmaal LaTeX iyo sharaxaadda sawirka. Warbixintu waxay u heellan tahay maqaal kaas oo aan ku muujinno in nidaamka dejinta ee la bartay uu dhab ahaantii macno samaynayo oo uu gaar u yahay dhibaatada la xallinayo.
• 20:55 Tabobarka Kala Duwanaanshaha KL ee Shabakadihii Hore: Hubanti la'aanta La Soo Hagaajiyay iyo Adkaynta Cadawga - Andrey Malinin, YandexHababka isku dhafan ee qiyaasida hubaal la'aanta ayaa dhawaan lagu dabaqay hawlaha ogaanshaha kala soocida khaldan, ogaanshaha ka baxsan qaybinta iyo ogaanshaha weerarka ka soo horjeeda. Shabakado hore ayaa loo soo jeediyay hab si wax ku ool ah loogu daydo isku-dhafka moodooyinka loogu talagalay soocidda iyadoo la cabbirayo Dirichlet ka hor qaybinta wax-soo-saarka. Moodooyinkan ayaa la tusay inay ka sarreeyaan habab kale oo isku dhafan, sida Monte-Carlo Dropout, oo ku saabsan hawsha ogaanshaha gelinta ka baxsan qaybinta. Si kastaba ha ahaatee, isku-dheellitirka Shabakadda Hore ee xog-ururin adag oo leh fasallo badan way adagtahay iyadoo la adeegsanayo shuruudaha tababarka markii hore la soo jeediyay. Warqadani waxay bixisaa laba wax ku darso. Marka hore, waxaanu tusinaynaa in shuruudaha tababarka ku haboon ee Shabakado hore ay tahay kala duwanaanshaha KL ee u dhexeeya qaybinta Dirichlet. Arrimuhu waxa ay wax ka qabanayaan nooca xogta tababarka ee qaybinta bartilmaameedka, awood u siinaya shabakadihii hore in si guul leh loogu tababaro hawlaha kala soocida ee leh fasallo badan oo aan sabab lahayn, iyo sidoo kale hagaajinta waxqabadka ogaanshaha ka baxsan qaybinta. Marka labaad, iyada oo laga faa'iidaysanayo shuruudaha tababarka cusub, warqaddani waxay baadhaysaa iyadoo la isticmaalayo Shabakadihii Hore si loo ogaado werarada iska soo horjeeda waxayna soo jeedinaysaa nooc guud oo tababar iska soo horjeeda. Waxaa la muujiyay in dhismaha weerarrada sanduuqa cad ee la-qabsiga ee guuleysta, kuwaas oo saameeya saadaasha iyo ogaanshaha ka-baxsanaanta, ee ka dhanka ah Shabakadihii Hore ee lagu tababaray CIFAR-10 iyo CIFAR-100 iyadoo la adeegsanayo habka la soo jeediyay waxay u baahan tahay dadaal badan oo xisaabin ah marka loo eego shabakadaha la difaacay iyada oo la adeegsanayo caadooyinka caadiga ah. tababarka ama MC-ka saarida.
• 21:10 Dood-kooxeed: "NeurlPS, oo aad u koray: yaa eedda leh maxaase la sameeyaa?" - Alexander Krainov, Yandex
• 21:40 Xafladda danbe

R Moscow Kulanka #5

Janaayo 16, 18:30-21:30, Khamiis

• 19: 00-19: 30 "Xalinta dhibaatooyinka hawlgalka iyadoo la isticmaalayo R ee dummies" - Konstantin Firsov (Netris JSC, madaxa fulinta Engineer).
• 19:30-20:00 "Hagaajinta alaabada tafaariiqda" - Genrikh Ananyev (PJSC Beluga Group, Madaxa Automation-ka warbixinta).
• 20: 00-20: 30 "BMS ee X5: sida loo sameeyo macdanta-nidaamka ganacsi ee macdanta POS ee aan habaysanayn iyadoo la isticmaalayo R" - Evgeniy Roldugin (Kooxda tafaariiqda X5, Madaxa Waaxda Qalabka Xakamaynta Tayada Adeegga), Ilya Shutov (Media Tel, Head) Saynisyahanka xogta Waaxda).

Kulankii hore ee Moscow (Gastromarket Balchug)

Janaayo 18, 12:00-18:00, Sabtida

• "Goorma ayay mudan tahay in dib loo qoro codsiga meel eber ah, iyo sida loo qanciyo ganacsiga tan" - Alexey Pyzhyanov, horumariye, SiburSheekada dhabta ah ee sida aan ula macaamilnay deynta farsamada ee habka ugu xagjirsan. Waxaan kaaga sheegi doonaa arrin ku saabsan:
  1. Waa maxay sababta codsigii wanaagsanaa uu u noqday dhaxal xun.
  2. Sida aan u qaadanay go'aanka adag ee ah in wax walba dib loo qoro.
  3. Sida aan fikradan uga iibinay milkiilaha alaabta.
  4. Maxaa fikradan ka soo baxay aakhirkii, iyo sababta aynaan uga qoomamayn go'aanka aanu qaadanay.

• "Vuejs API majaajillo" - Vladislav Prusov, horumariye Frontend, AGIMA

Tababarka barashada mashiinka ee Avito 2.0

Janaayo 18, 12:00-15:00, Sabtida

• 12:00 "Zindi Sendy Logistics Challenge (rus)" - Roman Pyankov
• 12:30 "Data Souls Wildfire AI (rus)" - Ilya Plotnikov
• 13:00 nasashada kafeega
• 13:20 "Topcoder SpaceNet 5 Loolan & Saxiix tartanka Satelite Tellus ee 3aad" - Ilya Kibardin
• 14:00 nasashada kafeega
• 14:10 "Codalab Automated Time Series Regression (eng)" - Denis Vorotyntsev

Source: www.habr.com