Kediska Joogista iyo Joogtada Stochastic

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Muxuu sameeyaa ka tagista xilliga tababarka?
   • Si joogto ah u tirtira miisaanka
   • Si aan kala sooc lahayn ayaa qayb ka mid ah wax soo saarka neerfaha u dejisa eber baas kasta oo hore
   • Laban laabma heerka waxbarashada
   • Wuxuu ku darayaa lakabyo badan shabakada
2. Maxay tahay sababta ka tagista u wanaajiso guud ahaan?
   • Waxay kordhisaa tirada xuduudaha
   • Waxay meesha ka saaraysaa baahida xogta tababarka
   • Waxay dammaanad qaadaysaa khaladka eber
   • Waxay ka hortagtaa neerfayaasha inay la qabsadaan oo ay xoogaystaan, matalaad adag
3. Maxaa ku dhacaya ka-tagidda wakhtiga imtixaanka?
   • Waa la damiyay oo shabakada buuxda ayaa la isticmaalay
   • Waxay hoos u dhigtaa xitaa neurons badan
   • Waxay tirtirtaa kala badh miisaanka
   • Waxay kala soocaysaa sumadaha
4. Heerka ka-tagidda p = 0.5 ee lakabka macneheedu waa qiyaastee inta lagu jiro tababarka?
   • Dhammaan neerfayaasha waa la saaraa
   • Wax neurons ah oo saameeya
   • Qiyaastii kala badh neuron-yada ayaa eber laga saaray baas kasta
   • Lakabka waa la tirtiray
5. Maxaa lagu tilmaamaa ka-tegidda inta badan?
   • Dib u noqosho toosan oo keli ah
   • Geed go'aan
   • A Fourier transformer
   • Isku dhafka shabakado badan oo khafiif ah oo wadaaga miisaanka
6. Maxaa Monte Carlo ka-tagidda loo isticmaalaa?
   • Dardar gelinta xogta
   • Qiyaasida hubanti la'aanta moodalka iyadoo la ilaalinayo ka tagista firfircoonida fikradda
   • Cadaadiska shabakada si loo yareeyo miisaan yar
   • Calaamadaynta xogta tababarka si toos ah
7. With inverted dropout, why is no extra activation scaling normally needed at ordinary inference?
   • The keep probability becomes exactly zero
   • Inference always averages new random masks
   • Surviving activations were already scaled during training
   • Every weight is divided by the number of layers
8. The same input yields varying predictions because dropout remains active during evaluation. What explains the variation?
   • New random masks sample different subnetworks on each pass
   • The input changes its label by itself
   • The network has no trained weights
   • Batch size must be one