"Izinselelo ezinkulu - 2020" zibizwa ngokuthi enye yamaphrojekthi ebibanjelwe engxenyeni yesibili kaDisemba eSirius Educational Center. Ekuhambeni kwalolu hlelo lwephrojekthi yesayensi nezobuchwepheshe, abafundi bezikole eziphakeme baqeqeshe ubuhlakani bokufakelwa. Abafana bekufanele bafundise imoto ekuhlakanipheni okulandelayo:
- Bhala izindaba.
- Ukuthuthukisa izidakamizwa ezintsha.
- Hlaziya izithombe ze-CT.
- Amaphrofayili wokutadisha abafakizicelo ekuxhumaneni nomphakathi.
Izingane zesikole zasiza ochwepheshe abavela eyunivesithi dopolis, yandex, esikoleni esiphakeme sabaphathi be-SPBU, i-VTB Bank, BioCAD. Kuhlelwe ukuthi okunikezwa okuhle kakhulu kuzosetshenziswa ekusebenzeni.
I-Andrei rasjorgodsky, inhloko "yomqondisi we-Big Data", umqondisi weFiztech-School of Mathematics kanye ne-Informatics of MFTI, kuchaziwe ukuthi iseshini yathatha izinsuku eziyi-10. Ngalesi sikhathi, abafana baye bafunda ukusebenza ngamathuluzi obuchwepheshe bedijithali futhi manje bazokwazi ukusebenzisa amakhono abo ngokuzayo ekhaya.
Le phrojekthi ihlinzeka ngesiqondisi esihlukile "idatha enkulu, ubuhlakani bokufakelwa, ubuchwepheshe bezezimali kanye nokufunda komshini". 28 Ama-grada avela ezifundeni eziyi-18 zaseRussia amenywa ukuba azosebenza kuwo. Laba bantu baba abasondelana bomncintiswano we-Russian ongumklami wonke nomsebenzi wokucwaninga.
Kwabahlanganyeli kulolu hlelo, i-Yandex yakha i-oda elikhethekile: ukuthuthukisa uhlelo ukukhiqiza izindaba, ukuze izinhlelo zesimanje zingasetshenziswa emsebenzini wansuku zonke. I-VTB Bank neSt Petersburg State University futhi yanikeza umsebenzi weSirius Schoolckildren. Abafana kufanele bahlaziye amanethiwekhi omphakathi futhi ngokususelwa kulwazi olutholwe ukwenza isithombe soMfakisicelo wesikole sebhizinisi.
Ososayensi badida izingane zesikole ngesifiso esiyinkimbinkimbi: Abafana kufanele bafundise ubuhlakani bokufakelwa ukuze babikezele ngokunembile ngosizo lwe-computer modelling ukwakheka kwesakhiwo sezinkinga zamangqamuzana lapho kwakha izidakamizwa ezintsha. Imoto kuzofanela ithole inhlanganisela ebaluleke kakhulu naleyo ekhona ngempela emvelweni. Indlela entsha kufanele ibikezelwe ukuthi kudalwe ngempumelelo kanjani. Iziphakamiso ezivela kubafundi bezikole eziphakeme kufanele zisetshenziswe ku-algorithm ye-biocad ye-biocad.
Ubuchwepheshe bedijithali obususelwa ku-intelligence bokufakelwa selisetshenziswe ngenkuthalo embonini yezokwelapha emazweni amaningi omhlaba. Lokhu kwavuselelwa ukuqala kokuqala kobhubhane lweCoronavirus. Ubuchwepheshe be-telemedicine bukhombisile inzuzo yokushintsha izinqubo eziningi, basiza ekulahlekelweni imitholampilo kanye nodokotela abavela emithanjeni ephindaphindekayo ekhuphuke kakhulu, bakwazile ukuhlinzeka ngokunakekelwa kwempilo okukude okuphezulu ezigulini.
Ngaphakathi kohlaka lwephrojekthi yesayensi nezobuchwepheshe yeSirius "Izinselelo ezinkulu - 2020, abafundi ababamba iqhaza kulolu hlelo bathola umsebenzi onzima ovela kuma-rencegeslogists aseRussia. Odokotela bacelwa ukuthi bafundise ubuhlakani bokufakelwa ukuhlaziya izithombe zezokwelapha besebenzisa umbono wekhompyutha futhi ngokushesha bathole i-pathology emaphashini. Usizo namathiphu we-neurosette enza lula ukusebenza kwe-radiologists, wehlise isikhathi sokucubungula izithombe.
Ngezinkinga ezahlukahlukene nezinkinga eziyinkimbinkimbi, njengoba kuthola izithombe zezokwelapha, ukufunda okujulile kudinga amasethi amaningi wedatha ukufeza izinga elidingekayo lokunemba. Izincomo zobuhlakani bokufakelwa zivame ukunemba kakhulu. Imiphumela yomsebenzi izolayishwa kwilabhulali yomthombo ovulekile, ukufinyelela okuzotholwa kochwepheshe kuwo wonke amazwe. Izisombululo zesikole zingahlolwa futhi zisetshenziswe ekusebenzeni uma ikhwalithi yentuthuko izohlela umsebenzisi. Usizo lozakwethu ezinganeni zesikole luhlinzekwa yisikhungo ngobuhlakani bokufakelwa kweNyuvesi yaseNennopolis.
Ithimba lephrojekthi lisebenzisa amamaki ayisithupha asuka e-Khanty-Mansiysk Autonomous Okrug, eVevastopol, i-Khabarovsk Territory, iRiphabhulikhi yaseBashkortotostan, iTyumen naseKemerovo nasezifundeni. Basizwa ngochwepheshe abavela enkabeni ye-Innopolis.
USemen Kiselev, umcwaningi wesikhungo ubuhlakani bokufakelwa e-University of Innopolis, iphrojekthi eqondisayo, wathi abathuthukisi bethule izingane ngezindlela zokufunda komshini kanye nombono wekhompyutha. Ukuthandwa kwezibalo, izinhlelo zokuhlela kanye nezindlela zokufunda ngomshini kuzokuvumela ukuthi ushintshe isimo emathubeni wokuvula abaningi abadinga lolu shintsho. Le phrojekthi izokhuthaza izingane zesikole ukuthi zikhuthele ngokuqhubekayo ekutadisheni amathuba okuguqulwa kwedijithali komphakathi, ebakholisa ukuthi kungaholela ekuphileni okugcwele ngokwengqondo. Okuhlangenwe nakho kokusebenza namanethiwekhi we-neural kuholela ekukhetheni umsebenzi wesikhathi esizayo, kukuvumela ukuthi uthinte imisebenzi emoyeni wenkululeko, uthuthukise ukuzethemba kwakho futhi uzokunikeza izingane kanye nentsha ukuba amakhono adingayo ngokuzayo.
Abafana bakhombisa ukuthi yiziphi izindawo abangafaka ngazo ulwazi olutholwe ngalo. Ababambiqhaza bakha amamodeli anjalo nama-algorithms okuhlaziya izifinyezo zamaphaphu, lapho ukuze athuthukezelele khona ukuze aqaphele i-pneumonia nezinye izifo zokuphefumula, i-pneumothorax noma isimila. Amamodeli wokufunda ajulile angenza ukubikezela ngokunemba okucishe kube yinto engenakuphikwa, ngoba ukucabanga kwangaphakathi kwemodeli kunzima ukuvula futhi kuhumushe, izimpikiswano ezivumela ukuthi kungani lesi sinqumo silungile, kwesinye isikhathi sihlala silungile.
Njengoba izinqumo zokwelashwa zingaba nesisindo esimangalisayo, abaningi babhekisa kakhulu ethembeni lokuzenzakalela okugcwele. Kubaluleke kakhulu ukukhumbula ukuthi imoto ayibuyiseli udokotela, kepha iba ngumsizi wayo nomeluleki, empeleni kuyiso nje iso lesithathu kudokotela. I-AI iyadingeka ukwandisa imisebenzi ejwayelekile, ngokuqaliswa kwayo okunobuhlakani okuguqula ngokujulile okukhethekile ngengxenye eqinile yokubuka, njenge-radiology kanye ne-pathology. Ochwepheshe, kufaka phakathi odokotela abahlinzayo, abanesifiso esisebenzayo ekuthuthukiseni nasekusebenziseni amacebo anjalo. I-AI iyadingeka ukwenza kahle imisebenzi ejwayelekile, ngesingeniso sayo esisabalele, okukhethekile kuzoshintsha ngengxenye eqinile yokubuka, njenge-radiology kanye ne-pathology. Ochwepheshe, kufaka phakathi odokotela abahlinzayo, abanesifiso esisebenzayo ekuthuthukiseni nasekusebenziseni amacebo anjalo.
Izingane zesikole ezifundwayo ze-algorithms zihlolwe ngenani elikhulu lezithombe zangempela zezokwelapha ezivela ezibhedlela zangempela.
Abasebasha bezinhlelo kufanele bafundise ubuhlakani bokufakelwa ukubala ubukhulu kanye nendawo yezinqubo ze-pathological emaphashini, zikhombise udokotela kuma-anomalies wesitho. Izinto zocwaningo zabahlanganyeli bephrojekthi zitholakala emithonjeni evulekile, imininingwane yalezi zindatshana zesayensi zamanyuvesi asekhaya nawakwamanye amazwe. Bahlola ama-algorithms abo kumakhulu ezithombe zezokwelapha ezivela kwi-LPUS yangempela, ngoba ukufinyelela idatha efanelekile nenokwanele kubaluleke kakhulu ekuqeqesheni ama-algorithms we-AI.
UDanila Pechenev, oneminyaka eyishumi nanye usuka eKemerovo, uthi wakhetha lo msebenzi, ngoba usesebenze ngombono womshini. Iphrojekthi yakhe yenethiwekhi ye-neural, ebona umbhalo wesandla, waba ngumncintiswano we- "Big Wezinselele" umncintiswano. Futhi umsebenzi kumsebenzi omusha uzokuvumela ukuthi ufunde izindlela zokufunda zomshini ezithuthukile. Ukuhlaziywa kwezithombe zezokwelapha namuhla kungumsebenzi ophuthumayo wokusebenzisa amanethiwekhi we-neural. Ubuchwepheshe bukhona, ezimweni eziningi ezibonisa ukunemba okuphezulu kumasheke ukuze kuqinisekiswe.
Okwenziwayo wephrojekthi, kuvulwa amathuba amahle ukuze azenzele ukuthuthukiswa nokuzibona. Abahlanganyeli abaningi be- "BIG GIC" edlule "babhalise emanyuvesi amakhulu amazwe futhi basebenza ngenkuthalo ngochwepheshe be-MFTI Laboratories. Iziqu zaseNyuvesi zihola amaphrojekthi kuyo yonke isikhathi sonke sokuqeqeshwa futhi zikwazise ngezimpumelelo kanye nentuthuko emaphephandabeni wesayensi womhlaba. Imibono eminingi yababambiqhaza be- "Izinselelo ezinkulu" sezisetshenzisiwe zokuzijwayeza izinkampani zozakwethu.