Ọmụmụ ihe jikọrọ ọnụ
Ịmụta n'otu n'otu na-azụ ụdị nkekọrịtara n'ofe ọtụtụ ngwaọrụ ma ọ bụ otu na-anakọtaghị data nkịtị n'otu ebe.
Nchịkọta
Only model updates travel to the server, so the sensitive data stays where it lives.
Ime miri emi
Na ọzụzụ nkịtị, a na-ejikọta data niile na sava etiti. Ọmụmụ ihe jikọrọ ọnụ na-atụgharị nke a: a na-ezigara ndị sonyere ihe nlereanya zuru ụwa ọnụ (ekwentị, ụlọ ọgwụ, ụlọ akụ), onye ọ bụla na-azụ mpaghara na data nke ya, naanị mgbanwe ịdị arọ ga-esi na ya pụta na-ezigaghachi. Ihe nkesa nkesa na-emelite mmelite ndị a ka ọ bụrụ ụdị zuru ụwa ọnụ emelitere ma na-emegharị ya. Google webatara echiche maka Gboard, kwalite amụma ahụigodo site na nde ekwentị na-ebugoteghị ihe ndị mmadụ dere. Ụzọ a na-enwu ebe data dị na nzuzo, ahaziri, ma ọ bụ buru ibu ịkwaga, dị ka ndekọ ahụike gbasaa n'ụlọ ọgwụ. Ihe ịma aka na-agụnye ngwaọrụ ndị a na-apụghị ịdabere na ya, data dị iche n'etiti ndị sonyere (data na-abụghị IID), yana eziokwu na mmelite raw ka nwere ike ịwụsa ozi, ya mere ejikọtara ya na usoro nzuzo.
Nghọta nka nka
Algọridim kpochapụwo bụ Federated Averaging (FedAvg): onye ahịa ọ bụla na-agba ọtụtụ usoro gradient na mpaghara, mgbe ahụ ihe nkesa na-ewere nkezi dị arọ nke ihe ọhụrụ ahụ, na-ejikarị ọnụọgụ data onye ahịa ọ bụla nwere. N'ihi na ndị ahịa na-azụ ọzụzụ maka ọtụtụ usoro tupu ịmekọrịta, mkparịta ụka na-adaba nke ọma karịa izipu gradient ọ bụla. Ka ịkwụsị mmelite sitere na ntọhapụ data, sistemu ejikọtara ọnụ na-agbakwunye mkpokọta echedoro, nke na-eme ka ihe nkesa ahụ hụ naanị nchikota agwakọta, yana nzuzo dị iche, nke na-etinye mkpọtụ emezigharị.
Mmetụta atụmatụ
Ọnụ ego na mmefu ego
Mkpebi ihe owuwu ụlọ na-akwalite arụmọrụ yana ọnụ ahịa ọrụ ruo ọtụtụ afọ.
Mkpebi doro anya
Nkà mmụta nka na-enyere ndị otu egwuregwu aka ịhọrọ nchịkọta ziri ezi, ọ bụghị naanị nke kachasị ọhụrụ.
Quality akara
Nhọrọ injinia ka mma na-ebelata ihe omume ntụkwasị obi na mmepụta.
Ọdịnihu nke mmụta zuru oke
Mmụta agbakwunyere na-esi na ahụigodo na-aga n'ihu na-aga n'ihu iji na-ahụ maka ahụike, ego, na IoT, ebe ụkpụrụ dịka HIPAA na GDPR na-eme ka nchịkọta data sie ike. Na-atụ anya njikọta siri ike na nzuzo dị iche iche yana nchịkọta echekwara, gbakwunyere usoro dị ka TensorFlow Federated, Flower, na NVIDIA FLARE na-eto maka mmepụta. Oke oke na-eto eto bụ nhazi nke ọma nke ụdị asụsụ buru ibu, na-ahapụ ndị otu dị iche iche ijikọ aka kwalite ụkpụrụ na ederede nzuzo. Ijikwa mma nke ndị na-ekesa na-enweghị ntụkwasị obi na-anọgide na-abụ isi ihe nyocha.
Mmejuputa n'ezie n'ụwa
Google Gboard na-emeziwanye amụma okwu na-esote na emoji n'ofe ekwentị na-ebugoteghị mkpịsị ugodi.
Ụlọ ọgwụ na-ejikọta ọnụ na-azụ ụdị onyogho nchọpụta nchọpụta na-enweghị ikekọrịta ndekọ ndị ọrịa echedoro.
Ụlọ akụ na-arụkọ ọrụ ọnụ na ụdị nchọpụta aghụghọ ma na-edobe azụmahịa ụlọ ọrụ ọ bụla na nzuzo.
Apple ahaziri na-ngwaọrụ atụmatụ dị ka QuickType na Siri aro iji obodo mmụta.
Ihe ize ndụ & okporo ụzọ nche
Ịkwalite otu akara ngosi nwere ike zoo adịghị ike sistemụ sara mbara.
A na-eledakarị ihe akụrụngwa na ụgwọ ọrụ anya.
Ọdịiche nchekwa na nleba anya nwere ike itolite ka sistemu na-adịwanye mgbagwoju anya.
Map mmejuputa
Kọwaa latency, ịdịmma na ebumnuche ọnụ ahịa tupu mmejuputa ya.
Benchmark n'okpuru ibu dị adị na ọnọdụ data.
Nleba anya akụrụngwa maka mperi, ịkpafu na mmetụta onye ọrụ.
Kwadebe ụzọ nzaghachi azụghachi azụ na ihe omume tupu ịchachaa.
Nọgide na-eme nchọpụta
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 Federated Learning 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
Ntuziaka na-esote
CI/CD maka mmụta igwe
Ajụjụ a na-ajụkarị
What is Federated Learning?
Ịmụta n'otu n'otu na-azụ ụdị nkekọrịtara n'ofe ọtụtụ ngwaọrụ ma ọ bụ otu na-anakọtaghị data nkịtị n'otu ebe. Naanị mmelite ihe nlereanya na-aga na sava ahụ, yabụ data nwere mmetụta na-anọ ebe ọ bi.
Gịnị bụ isi echiche nke mmụta n'otu?
Ịmụ ihe n'otu n'otu na-edobe data nkịtị na mpaghara; Onye ọ bụla so na ya na-azụ na data nke ya ma na-eziga naanị mmelite ụdị nke ihe nkesa na-ejikọta.
Kedu ihe algọridim nke Federated Averaging (FedAvg) na-eme na sava ahụ?
FedAvg na-ejikọta ùkwù ndị ahịa zụrụ azụ n'ime otu ụdị zuru ụwa ọnụ emelitere, na-ejikarị data ole onye ahịa ọ bụla na-atụ.
Kedu ngwaahịa a ma ama na-eji mmụta gbasara mmekọrịta chiri anya iji melite amụma?
Google jiri mmụta emekọrịtara ọnụ na Gboard kwalite amụma ederede site na nde ekwentị na-anakọtaghị ihe ndị ọrụ dere.
Kedu ihe kpatara mmụta mmụta jikọtara ọnụ ji mara mma maka data nlekọta ahụike?
Data ahụike bụ nke nzuzo ma hazie ya, yabụ idobe ya na mpaghara ka ị ka na-ewu ụdị nkekọrịta dabara n'iwu dị ka HIPAA nke ukwuu karịa ijikọ ya.
Kedu ihe kpatara na mmelite ụdị naanị abụghị nke nzuzo zuru oke?
Mgbanwe gradients na ịdị arọ nwere ike ikpughe nkọwa gbasara ọmụmaatụ ọzụzụ, nke mere eji agbakwunye mkpokọta echekwara na nzuzo dị iche.