Maningi amaqembu abacwaningi azama ukudala isithembiso sekhompyutha yesizukulwane esisha ekwazi ukunikela ngobuchwepheshe obusha, ukwakhiwa ... ngamafuphi, kusheshe, kusebenze futhi kudle okuncane. Izindlela okusetshenzelwa kuzo ziningi futhi, kulokhu, IBM kungenza nomahluko ngemuva kokuphumelela ukwethula i- uhlelo lwama-neurons angama-500 okufakelwa ifakwe ubuchwepheshe bokushintsha isigaba obungalingisa ukusebenza kobuchopho be-organic.
Uma sicabanga ukuthi ikhompyutha iyini, ngeke kube nzima kithi ukukhomba ukufana nanoma yiluphi uhlobo lobuchopho, hhayi ize, kunezinhlelo eziningi zokufunda nobuhlakani bokufakelwa obubeka isoftware ne-hardware ndawonye ukwenza indlela yokwenza noma yini uhlobo lobuchopho. uhlelo luyakwazi ukufunda nokwenza, kulokhu, njengomuntu. Okwamanje, ngohlelo lwama-neurons wokufakelwa, singathola ukuthi lezi zinhlelo zobunhloli bokufakelwa ziqhubekele phambili.
I-IBM ivula umnyango wesizukulwane esisha sezinhlelo zobuchwepheshe bokufakelwa
Ukulingisa ukusebenza kobuchopho bethu, lapho ulwazi luhlala emaphethini wokudubula ahlukene ama-neuron nokuxhumeka okusha okwenziwayo, abacwaningi be-IBM bakhethe ubuchwepheshe be ushintsho lwesigaba, uhlobo lwememori olukwazi ukugcina idatha ngendlela engashintshi ngoba ihlala esakhiweni nephethini yaso yobudlelwano. Inselelo manje etholwe yi-IBM kube ukuhlanganisa izakhiwo eziyinkimbinkimbi kakhulu nokusetshenziswa kwamandla okwamukelekayo, izinto ezivamile nokuziphatha okungahleliwe.
Ngokuqondene nezici ezithakazelisa kakhulu zala ma-neuron okufakelwa, gqamisa ngokwesibonelo ukuthi azilingisi nje kuphela ukusebenza kwazo, kepha nezazo isakhiwo sibonga ukufakwa kokungena, ulwelwesi, isoma ne-axon. Njengemininingwane, ukulingisa ulwelwesi lwama-neurons emvelo, kusetshenziswa amaseli enziwe nge-germanium, tellurium ne-antimony. Ngaphandle kokungabaza futhi ngenxa yalokhu kutholakala, kwa-IBM bakwazile ukuvula umnyango wesizukulwane esilandelayo sama-artificial intelligence.
Nge: Nature