Kahneman-Tversky
Kahneman-Tversky Optimization (KTO) xeetu méngale la buy jàngee ci etiketu baraam-ci kaw wala baraam-ci suuf ci barabu méngale ñaar.
Résumé
It matters because binary feedback is far easier and cheaper to collect than the ranked pairs most methods demand.
Plongeur bu xóot
KTO, Ethayarajh ak ay naataangoom ci Stanford ak juntuwaay yu bees yi ñoo ko dugal ci 2024, mungi leble ci gis gis bi, ligeeyu Daniel Kahneman ak Amos Tversky ci ni nit ñi di jappe ngëneel ak ñakk. Pexe yiñ miin lu melni DPO dañu soxla ñaari tànneef: tontu buñ tànn ak tontu buñ bañ ngir benn laaj. KTO lu moy loolu dafay liggéey ak done yuñ boolewul, muy génne bu nekk ñu màrkee ko bëgg wala bëggul. Dafa tabax benn ñàkka xam-xam bu nit ñi di jëfandikoo ngir gëna mëna jëfandikoo model bi ci misaal bi ni ngëneel wala ñàkkaale bu méngoo ak benn poñ de referansi, di jëfandikoo ñàkka bañ, suko defee génne yi ñu bëggul ñu gëna yar ñi ñu bëggul ñuy neexal. Loolu dafay may ekip yi ñu jëfandikoo siñaal yu bari yu am baraam ci kaw/wàcci yiñ jota dajale ci aplikaasioŋu defar yi.
Gis-gis xarala
KTO dafay màndargaal fonction valeur buñ modele ci theorie prospect, di natt ba fu tontu bi neexal bi nekk ci kaw wala ci suufu benn baseline (dafay faral di nekk moyenne KL-divergence ci politiku royuwaay). Misaal yiñ bëgg dañuy yokk valeur bi, yi ñu bëggul dañu koy wàcce, te koeffisien buy baña ñàkk dafay tax deviation negatif yi di gëna diis. Li gëna am solo mooy limu soxla du lenn ludul benn etiket ci misaal bu nekk, du ñaar yu méngoo.
njeextalu pexe
Gaawaay ak yaatuwaay
Liggéeyukaay yi ci làkk yi mën nañu gëna gaaw te duñu yàq deggoo gi.
Dugg ak yegg
Dafay yaatal jëfandikoo gi ci làkk yi ak ci anam yi ñuy jokkoo.
dogal yu gëna leer
Ekip yi mën nañu gëna yàgg ci àtte ci jamono ji otomatisation di liggéey ci baamtu.
Ëlëgu Kahneman-Tversky
KTO méngoo na bu baax ak produit dëgg yi, fu jëfandikukat yi klike bëgg wala bëggul waaye bariwul luñuy rang ñaari tontu ci wetu. Xaarandi jëfandikoo bu gëna yaatu ngir yokkute yuy wéy di rësikle feedback production, boole ci gëstu tuning ratio done yiñ bëgg ak yiñ bëggul ak poids perte-aversion. Teg KTO's behavioral-economics framing ak yeneen mébet, ak jëfandikoo ko ci feedback multimodal, yoon yu am solo lañu ndax ekip yi dañuy wut yoon ci siñaal yu dëggu yi.
Doxal ci àdduna dëgg
Jëfandikoo baraam-ci kaw/baraam-ci suuf klike ci chatbot biñ dugal ngir gëna suqali ko te doo musa tabax ñaari tànneef
Defar benn xeetu tontu soo amee ay tontu yu 'baax' ak 'yu baaxul' waaye amul benn méngale bu méngoo ak benn laaj bi
Ekipu produit biy rësikle ay bandera moderaasioŋ (baaxul) ak tontu yuñ denc (baaxul) ci tàggat KTO
Liggéeyu feedback bu jaxasoo fu bëgg-bëgg yi gëna néew bëgg-bëgg yi ci defar KTO's loss-aversion ak poids classe
Risk yi ak balustrade yi
Lépp lu jaarul yoon mën na dugg ci rapoor yi, jàppale ci liggéey bi, wala ci njariñu gëstu bi.
Sensibilite bu gaaw mën na jur njariñ yu wuute ci laajte yu noonu mel.
Done yu am solo mën nañu feeñ sudee seytu jëfandikoo gi néew doole.
Roadmap ngir samp gi
Mandargal formaa génne gi, melokaan bi, ak standard kalite yi laata ngay dugal ko.
Tontu yu am solo ak balluwaay yu wóor saa yu dëggu bi di am solo.
Fexeel am barabu xool nit ñi ngir am njariñ yu am solo.
Toppal anami gacce yi ak di faral di tàggataat ay laaj wala def-liggéey.
Weyal di banneexu
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 Kahneman-Tversky Optimization 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
Gis bi ci topp
Optimisation taamu direct
Laaj yi ñuy faral di laaj
What is Kahneman-Tversky Optimization?
Kahneman-Tversky Optimization (KTO) xeetu méngale la buy jàngee ci etiketu baraam-ci kaw wala baraam-ci suuf ci barabu méngale ñaar. Dafa am solo ndax feedback binär moo gëna yomba te gëna yomb ngir dajale, moo gën ñaari ranked yi anam yu bari di sàkku.
Ban xeetu done la KTO soxla?
KTO dafay jàng ci misaal bu yomb baraam-ci kaw/ci suuf etiketu ñaari tànneef yu méngoo.
Ban liggéey moo inspiré KTO?
KTO dafa lebal xalaatu theorie prospect biy xayma njariñu benefiis ak perte ci anam wu asimetrik, moo tax ñu tuddee ko.
Luy 'bañ ñàkk' ni ñu ko jëfandikoo ci KTO?
Teori prospect dafa wax ni perte yi ñoo gëna rëy benefiis yi, moo tax KTO dafay gëna diis deviation negatif yi.
Buñu ko méngale ak DPO, KTO dafa am njariñ soo amee lan?
KTO dafay leer sudee feedback bi amul siñaal binär, produit yi ñoo koy dajale ci anam wu gëna yomb yeneen yuñ rang.
Ci KTO, deviation yi ñu ngi leen di natt ci lan?
Fonction valeur bu KTO dafay àtte ndax tontu bi mingi ci kaw wala ci suufu baseline de referanse, diko jàppee ni benefiis wala perte.