Yandex ibhengeze ukuqaliswa . Igalari iya kubonisa abasebenzisi be-4000 imizobo ekhethekileyo, eyenziwe yi-algorithm esekelwe kwi-teknoloji ye-intelligence intelligence. Nabani na unokukhetha enye yemizobo eseleyo kwisitokhwe kwaye azithathele yona simahla. Kule meko, umnini wayo kuphela uya kuba nenguqu epheleleyo yomfanekiso.
Igalari ye-neural network art gallery yahlulwe yangamagumbi ama-4 anethematics. Abasebenzisi banokujonga ukuyilwa kwenkqubo ye-AI kwiNdalo, abantu, isiXeko, kunye neendlela zeMood. Igalari ebonakalayo iya kuvumela iindwendwe ukuba zityelele umboniso opheleleyo ngaphandle kokushiya ikhaya, kwaye imisebenzi abayithandayo inokwabelwana ngayo nabahlobo kwiinethiwekhi zentlalo.
Iindwendwe zokuqala ziya kukwazi ukukhuphela umfanekiso omnye eziwuthandayo ngobukhulu obupheleleyo. Ukuze ube ngumnini wepeyinti eyenziwe yi-neural network, kuya kufuneka ungene kuyo nayiphi na inkonzo yeYandex. Imizobo efunyanwa ngabanini iya kuqhubeka ifumaneka ukuze ijongwe, kodwa kwigalari iya kuboniswa kuphela kwifom encitshisiweyo.
Imisebenzi ebonisiweyo yenziwe yinethiwekhi ye-neural ephinda i-StyleGAN2 uyilo. Kwinkqubo yokuqeqesha inethiwekhi ye-neural, iingcali zasebenzisa imisebenzi yeendlela ezahlukeneyo, ezifana ne-cubism, i-minimalism, ubugcisa besitrato, njl. Ngexesha loqeqesho, i-neural network yafunda imizobo ye-40, emva koko yaqala ukudala imisebenzi yayo. Ukukhetha imizobo ngokweendidi ezahlukeneyo, enye inethiwekhi ye-neural yasetyenziswa, esetyenziswa kwinkonzo yeYandex.Imifanekiso yokukhangela imifanekiso esekelwe kwimibuzo. Nguye owayekwazi ukubona abantu, indalo, isixeko kunye neemvakalelo ezahlukeneyo kwimizobo, ukuhlela imisebenzi ekhoyo kwiindidi.
umthombo: 3dnews.ru