amụma nkwekọrịta
Amụma nkwekọrịta na-ekpuchi ihe atụ ọ bụla iji wepụta setịpụ ma ọ bụ nkeji oge nke ekwenyere na ọ ga-enwe azịza ziri ezi yana ohere ahọpụtara, dị ka 90%.
Nchịkọta
It turns a single guess into a trustworthy range with a mathematical coverage promise.
Ime miri emi
Ọtụtụ ụdị na-enye gị amụma isi ma ọ bụ akara softmax nke dị ka ntụkwasị obi mana ọ naghị adịkarị. Amụma kwekọrịtara na-edozi nke a. Ị na-ewere ụdị a zụrụ azụ, tụọ ka 'iju' ihe atụ ọ bụla si eji usoro enweghị nkwekọrịta (dịka ọmụmaatụ, njehie ma ọ bụ otu na-ewepụ ihe gbasara puru omume), wee gbakọọ akara ndị ahụ na nhazi nhazi ejidere. Iji buo amụma maka isi ihe ọhụrụ, ị gụnyere akara ọ bụla nke akara ekwenyeghị na ya adịghị njọ karịa pasentị 90 nke akara nhazi. Nsonaazụ bụ atọrọ amụma, ikekwe ọtụtụ aha maka nhazi ọkwa ma ọ bụ nkeji oge maka nlọghachi azụ. Nkwenye nke isi okwu enweghị nkesa: ọ bụrụhaala na enwere ike ịgbanwe data gị, setịpụ ahụ na-ekpuchi ezigbo uru na ọnụego ahọpụtara, n'agbanyeghị ụdị nke ị na-eji.
Nghọta nka nka
Isi aghụghọ bụ mgbanwe mgbanwe yana quantile. N'iji akara n calibration, ọnụ ụzọ bụ elu ụlọ nke (n+1)(1-alpha)/n quantile nke akara ndị ahụ. N'ihi na akara ọhụụ ọhụrụ nwekwara ike ịdaba n'ọkwa ọ bụla n'etiti akara nlegharị anya, ihe gbasara nke puru omume ọ gafere n'ọnụ ụzọ bụ n'ọtụtụ alfa. Arụmụka ahụ achọghị echiche gbasara ihe nlereanya ma ọ bụ nkesa data, naanị na isi ihe na-agbanwe agbanwe.
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 amụma nkwekọrịta
Nchọpụta na-agabiga ihe achọrọ mgbanwe na usoro oge yana nkesa na-agbanwe, na-eji usoro ngbanwe na nha nha nke na-edozi ọnụ ụzọ n'ịntanetị. Mkpuchi ọnọdụ, na-ekwe nkwa na ọnụego a na-ejide maka obere otu ọ bụla karịa na nkezi, bụ oke oghere mepere emepe. Na-atụ anya ka ụgbọ mmiri dị n'ime pipeline LLM, ngwaọrụ nleba anya ahụike, yana sistemu kwụụrụ onwe ya ka ndị na-achị achị na-achọwanye nke ọma, ejighị n'aka na-enyocha kama ibu amụma efu.
Mmejuputa n'ezie n'ụwa
Ihe nhazi ọnya akpụkpọ ahụ na-eweghachite {melanoma, nevus} setịpụrụ mgbe ọ na-ejighị n'aka, na-akpali onye ọkachamara n'ihe gbasara anụ ahụ kama iji otu akara nwere obi ike.
Ụdị ọnụ ahịa ụlọ na-ewepụta oge $310k-$365k na-ekwe nkwa na ọ ga-enwe ọnụahịa ire ere 90% nke oge maka mkparịta ụka ndị na-azụ ahịa.
Sistemụ azịza ajụjụ LLM na-ejikọ obere azịza nke nwere nkwa mkpuchi, na-egosipụta nnukwu usoro dị ka ikpe chọrọ nyocha mmadụ.
Pipeline nyocha nke ọgwụ na-egbu egbu na-ewepụta oge amụma ka ndị na-ahụ maka chemist mara ogige ndị nwere atụmatụ dị warara nwere ntụkwasị obi na ndị na-ejighị n'aka.
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 Conformal Prediction 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
Ntụgharị egwu egwu na amụma Multi-Token
Ajụjụ a na-ajụkarị
What is Conformal Prediction?
Amụma nkwekọrịta na-ekpuchi ihe atụ ọ bụla iji wepụta setịpụ ma ọ bụ nkeji oge nke ekwenyere na ọ ga-enwe azịza ziri ezi yana ohere ahọpụtara, dị ka 90%. Ọ na-atụgharị otu ịkọ nkọ ka ọ bụrụ oke ntụkwasị obi nwere nkwa mkpuchi mgbakọ na mwepụ.
Kedu ihe usoro amụma conformal na-ewepụta kama ịbụ otu akara?
Amụma nkwenye na-eweghachite ntọala (nhazi) ma ọ bụ etiti oge (regression) emebere ka ọ nwee ezigbo uru na ọnụego ahọpụtara dị ka 90%.
Kedu echiche na-akwado nkwa mkpuchi mkpuchi ọkọlọtọ?
Nkwa ahụ na-adabere na mgbanwe mgbanwe, nke pụtara na nkesa nkwonkwo adịghị agbanwe agbanwe site n'ịhazigharị isi ihe. Ọ dịghị mkpa echiche banyere ihe nlereanya.
Kedu ọrụ nke nhazi nhazi?
A na-agbakọ akara ekwenyeghị n'usoro n'usoro nhazi ejidere, na ọnụọgụ ha na-ekpebi akara ndị a ga-etinye maka isi ihe ọhụrụ.
Ọ bụrụ na ịtọlite alfa = 0.1, kedu mkpuchi ị na-ezubere iche?
Mkpuchi nhata 1 mwepu alfa, yabụ alfa = 0.1 lekwasịrị anya mkpuchi 90% nke ezigbo uru.
Kedu ihe na-emekarị na amụma atọrọ mgbe ihe nlereanya na-ejighị n'aka maka ntinye?
Ntinye aka siri ike na-ewepụtakwu aha akara nke akara ya dabara n'okpuru ọnụ ụzọ, yabụ na ntọala buru ibu na-egosi ejighị n'aka ka ukwuu.