Meta-Jàng
Meta-jàng, wala 'jàng ngir jàng,' dafay tàggat xeetu nit ñi ñu gaaw ci mëna ànd ak liggéey yu bees ci misaal yu néew.
Résumé
It matters because it pushes AI toward the human-like flexibility of mastering something new without huge datasets.
Plongeur bu xóot
Meta-jàngat dafay fexe génne ay xeetu jàng liggéey yu bees ci lu gaaw ci tàggat ci liggéey yu bari te wuute, du benn. Duñu def benn ensemble done bu baax, waaye model bi dafay séddale ay liggéey ci diiru 'meta-training', fu liggéey bu nekk am ensemble ndimbal bu ndaw (ngir jàngee ci) ak ensemble laaj (ñu wara jàngat ci). Luñu bëgga mooy wut barab bu ñuy tàmbalee wala pexe buy yamale, kon su liggéey bu bees dëgg-dëgg yegsee, yenn jéego wala misaal yu néew kese lañu soxla. Mën-mënu 'few-shot' bi nekk na lu am solo ci terrain bi. Xeetu xam-xam yiñ gëna xam ñooy MAML, muy jàng tàmbali bu yomb defar, ak pexe yu lalu ci metrik yu melni Reseau Prototypical, yuy xaaj ci méngale ak prototype class yiñ jàng.
Gis-gis xarala
Model-Agnostic _AIU_AAR_13__-Jàng (MAML) dafay jëfandikoo ab bouclage buñ boole. Boucle bi ci biir dafay méngale model bi ak benn liggéey ci ay jéego gradient yu néew; boucle bi ci biti dafay yeesal parametre yi njëkk, suko defee, ginaaw biñu ko méngale, performance bi dafay yéeg ci liggéey yu bari. Ci dëgg defay optimiser ngir gaaw ci mëna ànd ak liggéey bu jaar yoon, yenn saa muy laaj gradient yu ñaareelu rang.
njeextalu pexe
dogal yu gëna leer
Daf lay jàppale nga tàqale kàddu yu leer ci wàllu xarala ak làkku fësal njaay.
Njëgg ak budget
Mën nga laaj laaj yu gëna baax ci samp gi balaa ngay dugal xaalis wala sa jotu liggéey.
Ekip ak def liggéey
Ekip yi bokk xam-xam ñoo gëna mëna jël yenn dogal ci wàllu produit, politik ak jàng.
The Future of Meta-Learning
Meta- xalaati jàng dañuy gëna jaxasoo ak jàng ci biir xeetu làkk yu yaatu, yuy méngoo ak misaal yi ci saa si te du am benn coppite ci diisaay bi. Xaarandi lëkkaloo bu gëna dëgër ak xeetu fondasioŋ yi, robotik yu gëna am njariñ ci done ak personaalisasioŋ, ak gëstu ci meta-jàngat bu gëna xéewale te gëna dëgër, wàññi njëg yu bari yi ñuy laaj ci pexe yu yàgg yi.
Doxal ci àdduna dëgg
Xeetu nataal yu néew, di barab bu benn model di xàmmee kategori mbir yu bees, daale ko ci benn ba juróomi misaal yuñ etikete.
Robotik, muy robot buñ tàggat ci liggéey yu bari, mu mëna ànd ak liggéey bu bees ci diir bu gàtt.
Tegtal buñ personaalise bu baax wala waxtaanu klaweer buy gaawa méngoo ak jëfandikukat bu bees bu am done yu néew.
Gis-gis drog, fu ay model di méngoo ngir mëna wax luy waaja am ci xeetu molecule bu bees ci ay misaal yuñ natt.
Risk yi ak balustrade yi
Ekip yu bari mën nañu jëfandikoo benn baat ci anam wu wuute, kon teela leeral yaatuwaayam.
Benchmark yi mën nañu nuru lu am doole waaye performance yi ci àdduna bi duñu tolloo.
Bëgg kalite done ak palaŋu jàngat dafay faral di jur njariñ yu yomba dagg.
Roadmap ngir samp gi
Tàmbaleel ci joxe leeral ci làkk wu leer ci njariñ li nga soxla.
Tannal benn metric bu baax ak benn anam bu baaxul balaa ngay saytu.
Doxal ab pilote bu ndaw ak ay done yu representatif, du ab demo bu leer.
Bindal barab yi Meta-Jàng di jàppale ak barab yi gëna yomba jëfandikoo.
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 Meta-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
Gis bi ci topp
Jàng bu xóot bu Bayesian
Laaj yi ñuy faral di laaj
What is Meta-Learning?
Meta-jàng, wala 'jàng ngir jàng,' dafay tàggat xeetu nit ñi ñu gaaw ci mëna ànd ak liggéey yu bees ci misaal yu néew. Dafa am solo ndax dafay puus IA ci neexaayu nit ñi ngir xam lu bees te du am ay done yu bari.
Lan mooy meta-learning, wala 'jàng ngir jàng', di njëkka am?
Meta-jàng dafay jàngale liggéey yu bari suko defee benn model mëna gaaw ci ànd ak liggéey bu bees ci ay misaal yu néew.
Bu ñuy meta-training, naka lañuy faral di defaree done yi?
Liggéey bu nekk dafay joxe benn xeetu ndimbal bu ndaw ngir mëna ànd ak benn xeetu laaj ngir jàngat mën ànd ak moom, ñu baamtu ko ci liggéey yu bari.
Luy MAML jàng?
MAML dafay gis paramet yu njëkk yi, ginaaw ay jéego gradient yu néew ci liggéey bu bees, di joxe performance bu am doole.
naka lañuy xaajalee misaal yu bees yi ci reso prototypique yi?
Reseau prototypique yi dañuy xayma benn prototype (moyenne embedding) ci klaas bu nekk ba noppi ñu jox laaj bu nekk prototype bi gëna jege.
Lan mooy wareefu bouclage bi ci biir MAML?
Boucle bi ci biir dafay def adaptaasioŋ bu jëm ci liggéey bi, waaye boucle bi ci biti dafay yeesal paramet yi ngir gëna mëna ànd ak lépp.