Kongera amakuru
Kwiyongera kwamakuru kwagura muburyo bwamahugurwa yashizweho mugukora kopi zahinduwe zingero zihari - nko guhinduranya cyangwa guhinga amashusho.
Incamake
It matters because more varied data reduces overfitting and helps models generalize to inputs they have not seen.
Kwibira cyane
Kwiyongera kwamakuru bitanga ingero nshya zamahugurwa ukoresheje label-izigama impinduka kumakuru usanzwe ufite. For images, that means rotations, flips, crops, color shifts, blurring, and adding noise — changes that alter pixels but not the correct answer (a flipped cat is still a cat). For text, techniques include synonym replacement, back-translation (translate to another language and back), and random word deletion or swapping. Kubijyanye n'amajwi, urashobora kongeramo urusaku rwimbere, guhinduranya umwanya, cyangwa clips-igihe. Intego nukwigisha icyitegererezo kudahinduka bifite akamaro - ko ikiranga ikintu kidashingiye kumwanya wacyo, kumurika, cyangwa interuro. Ibi bituma moderi irushaho gukomera kandi ifite agaciro cyane mugihe yanditseho amakuru ari make, kuva buri rugero nyarwo ruba rwinshi. Imiyoboro igezweho ikunze guhitamo kwiyongera kuguruka muri buri gihe cyamahugurwa.
Ubushishozi
Augmentation works because it injects prior knowledge about invariances directly into training: by showing the model many transformed versions of one example, you encourage it to learn features that ignore irrelevant variation. Icy'ingenzi, impinduka zigomba kubika ikirango - guhinduranya '6' muri '9' byigisha ibitari byo. Advanced methods go beyond simple edits: Mixup blends two images and their labels, Cutout masks regions, and learned policies like AutoAugment search for the best transformation combinations for a given dataset.
Ingaruka z'Ingamba
Ibyemezo bisobanutse
Iragufasha gutandukanya ibyifuzo bya tekiniki bisobanutse nururimi rwo kwamamaza.
Igiciro na bije
Urashobora kubaza ibibazo byiza byo gushyira mubikorwa mbere yo gukoresha amafaranga cyangwa igihe.
Itsinda hamwe nakazi
Amakipe asangiye ibitekerezo akora ibicuruzwa byiza, politiki, nibyemezo byo kwiga.
Ejo hazaza h'amakuru
Imipaka irabyara kandi yize kwiyongera: ukoresheje moderi yo gukwirakwiza cyangwa GAN muguhuza ingero nshya rwose, zifatika zifatika aho guhindura gusa izishaje. Gushakisha byikora byikora (AutoAugment, RandAugment) bigabanya guhuza intoki, kandi kwiyongera ubu nibyingenzi mubyigenzurwa no kwigira, aho abiga biga mukumenya ko ibitekerezo bibiri byongeweho ibitekerezo bimwe bigomba guhura. Tegereza kwiyongera kugirango ukomeze uhuze umurongo hamwe namakuru yubukorikori, cyane cyane kumasomo adasanzwe hamwe na domaine-yerekeye ubuzima bwite aho gukusanya amakuru nyayo bigoye.
Gushyira mu bikorwa Isi
Ishusho itondekanya ishusho itoza kumafoto atunguranye, asaruwe, hamwe namabara-amabara kugirango amenye ibintu atitaye kumpande cyangwa kumurika.
Itsinda rya NLP rikoresha ibisobanuro-byinyuma (Icyongereza kugeza Ikidage ninyuma) kugirango bisobanure interuro no kwagura amarangamutima-isesengura dataset.
Imvugo yerekana yongeramo urusaku rwa café kandi ihindura ikibanza kumajwi yafashwe kugirango igume neza mubihe byuzuye urusaku rwisi.
Ubuvuzi bwa AI bukoresha disformasiyo ya elastique kandi igahinduka kumurongo ntarengwa wa scan ya MRI kugirango ugwize ingero nke zanditseho nta barwayi bashya.
Ingaruka & Kurinda
Amakipe atandukanye arashobora gukoresha ijambo rimwe muburyo butandukanye, sobanura intera hakiri kare.
Ibipimo birashobora kugaragara bikomeye mugihe imikorere-yisi-itaringaniye.
Kwirengagiza ubuziranenge bwamakuru na gahunda yo gusuzuma akenshi bitanga ibisubizo byoroshye.
Igishushanyo mbonera
Tangira nururimi rusobanutse rwibisubizo ukeneye.
Toranya intsinzi imwe hamwe nuburyo bumwe bwo gutsindwa mbere yo kwipimisha.
Koresha umuderevu muto hamwe namakuru ahagarariye, ntabwo ari demo yashizweho.
Inyandiko aho Data Augmentation ifasha kandi nuburyo bworoshye bworoshye.
Komeza Ubushakashatsi
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 Data Augmentation 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
Ubuyobozi bukurikira
AI & Data
Ibibazo bikunze kubazwa
What is Data Augmentation?
Kwiyongera kwamakuru kwagura muburyo bwamahugurwa yashizweho mugukora kopi zahinduwe zingero zihari - nko guhinduranya cyangwa guhinga amashusho. Nibyingenzi kuko amakuru menshi atandukanye agabanya gukwirakwira kandi agafasha moderi guhuza inyongeramusaruro batabonye.
Niyihe ntego yibanze yo kongera amakuru?
Augmentation ikora amakopi atandukanye, yahinduwe yamakuru ariho kugirango agabanye gukabya no gufasha icyitegererezo gukora inyongeramusaruro zitagaragara.
Ninde muribi nubuhanga busanzwe bwo kongera amashusho?
Guhinduranya, guhinga, kuzunguruka, no guhinduranya amabara nibisanzwe byongerewe amashusho bihindura pigiseli mugihe ubitse ikirango.
Niki-gusubira inyuma-gukora iki mukwongera inyandiko?
Inyuma-yubusobanuro ikoresha inyandiko ikoresheje urundi rurimi ninyuma, itanga verisiyo yasubiwemo ibika ibisobanuro.
Kuki kwiyongera bigomba 'kuranga-kubika'?
Guhinduka ntibigomba guhindura igisubizo cyukuri - guhinduranya '6' muri '9', kurugero, byandika nabi urugero.
Tekinike ya mixup ikora iki?
Kuvanga gukora ibyitegererezo bishya muguhuza umurongo winjiza nibirango byabo, bigatera imipaka imipaka yoroshye.