Hanyoyin Sadarwar Jijiya Na Juyin Halitta
Hanyoyin Sadarwar Jijiya na Juyin Halitta (CNNs) sune gine-ginen dokin aiki don fahimtar hotuna.
Dubawa
They learn visual patterns by sliding small filters across a picture, which is why they power everything from face unlock to medical scan analysis.
Zurfafa nutsewa
CNN tana aiwatar da hoto ta hanyar zamewa ƙananan grid na ma'auni, da ake kira filtata ko kernels, a fadin pixels. Kowane tacewa yana duba ƙirar ƙira ɗaya, kamar gefu, ɗigon launi, ko kusurwa. Yadudduka na farko suna gano abubuwa masu sauƙi; zurfafa yadudduka suna haɗa su cikin idanu, ƙafafu, ko rubutu. Domin ana sake amfani da wannan tacewa a kowane matsayi (raba nauyi), CNN yana buƙatar ƙananan sigogi fiye da cikakkiyar hanyar sadarwar da aka haɗa kuma tana iya gano cat ko ya bayyana a sama-hagu ko kasa-dama. Yadudduka na ruwa suna rage hoton tsakanin matakai, yana sa hanyar sadarwar sauri da kuma jure wa ƙananan canje-canje. Abubuwan ƙira na ƙasa kamar LeNet, AlexNet (2012) da ResNet sun haɓaka haɓakar koyo mai zurfi, tare da nasarar AlexNet ImageNet wanda ya haifar da yanayin zamani na filin.
Fahimtar Fasaha
Babban aikin shine jujjuyawar: tacewa (ce ma'aunin nauyi 3x3) an lullube shi akan facin pixels, kowane nauyi yana ninka ta pixelsa, kuma ana tattara sakamakon zuwa lambar fitarwa guda ɗaya. Zamewa tace yana samar da taswirar fasali. Ra'ayoyi guda biyu sun sa wannan ingantaccen aiki: raba nauyi (matattarar sake amfani da ita a ko'ina) da haɗin gida (kowane neuron yana ganin ƙaramin yanki ne kawai). Matsakaicin juzu'i, rashin kan layi kamar ReLU, da haɗawa suna ba da damar cibiyar sadarwa ta gina tsarin haɓakar abubuwan gani na gani.
Dabarun Tasiri
Shawarwari masu haske
Yana taimaka muku keɓance bayyanannen da'awar fasaha daga harshen talla.
Kudin da kasafin kuɗi
Kuna iya yin mafi kyawun tambayoyin aiwatarwa kafin kashe kuɗi ko lokaci.
Ƙungiya da aikin aiki
Ƙungiyoyin da ke da fahimtar juna suna yin mafi kyawun samfura, manufofi, da yanke shawara na koyo.
Makomar Hanyoyin Sadarwar Jijiya na Juyin Halitta
CNNs sun kasance masu rinjaye a cikin ainihin-lokaci da hangen nesa mai iyaka, kamar kyamarori na waya da tsinkayen tuki, saboda suna da sauri da ingantaccen bayanai. Vision Transformers yanzu suna hamayya ko doke su akan manyan ma'ajin bayanai, don haka filin yana haɗuwa akan ƙirar ƙira waɗanda ke haɗa ingantaccen juzu'i tare da hankalin duniya. Yi tsammanin CNNs za su dawwama a cikin na'urori masu haɗawa da gefen, a cikin hoton likitanci inda bayanai ke da yawa, kuma a matsayin ingantattun abubuwan cire kayan aikin da ke ciyar da manyan tsarin multimodal na shekaru masu zuwa.
Aiwatar da Gaskiyar Duniya
Gano ciwace-ciwacen ciwace-ciwace, karaya, da ciwon suga a cikin rayuwar X-ray, CT scan, da hotunan retinal
Ƙarfafa ƙwarewar fuska don buɗe waya da yin alamar hoto a cikin ƙa'idodi kamar Google Hotuna
Karatun alamomin titi, alamomin layi, da masu tafiya a ƙasa a cikin tsarin tsinkayar mota mai tuƙi
Ƙaddamar da lahani ta atomatik akan layukan haɗin masana'anta ta hanyar binciken kyamara
Hatsari & Tsare-tsare
Ƙungiyoyi daban-daban na iya amfani da kalmar iri ɗaya daban, don haka ayyana iyaka da wuri.
Alamomi na iya yin kama da ƙarfi yayin da aikin zahirin duniya bai yi daidai ba.
Yin watsi da ingancin bayanai da tsare-tsaren kimantawa galibi yana haifar da sakamako mara ƙarfi.
Taswirar Hanya
Fara da ma'anar harshe a sarari na sakamakon da kuke buƙata.
Zaɓi ma'aunin nasara ɗaya da yanayin gazawa ɗaya kafin gwaji.
Gudun ƙaramin matukin jirgi tare da bayanan wakilci, ba saitin demo da aka goge ba.
Takaddun inda hanyoyin sadarwa na Juyin Halitta ke taimakawa kuma inda hanyoyin mafi sauƙi suka fi kyau.
Ci gaba da Bincike
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Convolutional Neural Networks quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Jagora na gaba
Graph Neural Networks
Tambayoyin da ake yawan yi
What is Convolutional Neural Networks?
Hanyoyin Sadarwar Jijiya na Juyin Halitta (CNNs) sune gine-ginen dokin aiki don fahimtar hotuna. Suna koyon tsarin gani ta hanyar zamewa ƙananan tacewa a kan hoto, wanda shine dalilin da ya sa suke sarrafa komai daga buɗe fuska zuwa binciken binciken likita.
Menene babban aikin tacewa (kernel) a cikin CNN?
Tace ƙaramin grid ɗin ma'auni ne wanda ke zamewa akan hoton, yana kunnawa lokacin da ya sami tsarin da ya koya, kamar gefuna ko rubutu.
Me yasa CNN ke buƙatar ƙarancin sigogi fiye da cikakkiyar hanyar sadarwar da aka haɗa don hotuna?
Rarraba nauyi yana nufin ana amfani da ƙaramin tacewa ko'ina a cikin hoton, don haka hanyar sadarwar ta sake yin amfani da ma'auni iri ɗaya maimakon koyan daban na kowane wuri pixel.
Menene Layer pooling yawanci ke yi?
Pooling (kamar max pooling) yana saukar da misalan taswirar fasalin, rage ƙididdigewa da sanya cibiyar sadarwa ta kasa kula da daidai inda fasalin ya bayyana.
A cikin CNN mai zurfi, ta yaya fasalin ke canzawa gabaɗaya daga yadudduka na farko zuwa yadudduka na gaba?
Yadudduka na farko suna gano ƙananan siffofi kamar gefuna da launuka; zurfin yadudduka suna haɗa waɗannan zuwa manyan ra'ayoyi kamar fuskoki ko sassan abu.
Wanne taron ne aka yi la'akari da shi da ƙaddamar da haɓakar zurfin ilmantarwa na zamani a hangen nesa?
Nasarar ImageNet na 2012 mai ban mamaki ta AlexNet ta amfani da CNN mai zurfi akan GPUs ya gamsar da masu bincike cewa zurfin ilmantarwa zai iya fin tsofaffin hanyoyin, yana haifar da saurin ci gaban filin.