Siamese Networks uye Triplet Kurasika Quiz

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

1. Chii chinotsanangudza chivakwa cheSiamese network?
   • Matavi maviri kana anopfuura akafanana anogovera huremu hwakafanana
   • Bazi rimwechete risina matinji akavanzika
   • Kuparadzanisa network ine huremu hwakazvimiririra hwakakwana
   • A recurrent loop nekufamba kwenguva matanho
2. Mukurasikirwa katatu, ndezvipi zvitatu zvekupinza?
   • Input, yakavanzika, inobuda
   • Anchor, positive, uye negative
   • Chitima, kusimbisa, kuedza
   • Encoder, decoder, classifier
3. Chinangwa chekurasikirwa katatu chinosimbisa chii?
   • Anchor uye negative zvinofanira kufanana
   • Ese embeddings inofanira kudonha kusvika pane imwe nzvimbo
   • Anchor-positive chinhambwe chinofanira kuva diki pane chinhambwe chinobatika-negative nechekumucheto
   • Positive nenegative inofanira kuva nenhambwe yakaenzana kune anchor
4. Sei 'hard-negative mining' yakakosha pakudzidzisa kurasikirwa katatu?
   • Inobvisa zvese zvakanaka kubva pakudzidziswa
   • Inoita kuti network ive shoma
   • Inotsiva kureba metric nelebel
   • Easy negatives inotogutsa muganho uye inoburitsa zero inobatsira gradient
5. Ndeupi hunyanzvi hunoita kuti maSiamese network akoshese 'pakudzidza-kamwe chete'?
   • Vanogona kuenzanisa kupinza kutsva kune imwechete referensi uye kutonga kufanana, kunyangwe kumakirasi asingaonekwe
   • Vanoda zviuru zvemienzaniso yakanyorwa pakirasi
   • Vanongoisa muzvikamu zvakatemwa, zvakafanotsanangurwa
   • Havagone kuita generalize kunze kwekuzivikanwa kwekudzidziswa
6. Ko network yeSiamese inowanzo sarudza sei kana maviri ekuisa achienderana?
   • Nekupfupikisa mapikisie maitiro avo
   • Nekuenzanisa chinhambwe chiri pakati pemavheta ekuisa pakatarisana nechikumbaridzo
   • Nekuverenga nhamba yezvikamu zvakashandiswa
   • Nekupa yega yega chinyorwa chekirasi
7. For triplet loss max(0, d(a,p) − d(a,n) + m), let d(a,p)=0.2, d(a,n)=0.8 and m=0.3. What is the loss?
   • 0.3
   • 0.5
   • 0
   • 1.1
8. What happens when a negative example is already farther from the anchor than the required margin?
   • The anchor and positive embeddings are discarded
   • The loss forces the negative closer to the anchor
   • The network switches to classification mode
   • The triplet loss becomes zero and that example gives no useful gradient