Hoos-u-dhaca Stochastic Gradient oo leh Kediska Momentum

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Waa maxay ereyada firfircoonidu ku ururto muddada tababarka?
   • Tirada guud ee xilliyadii tababarka
   • Celcelis ahaan orodka (si weyn loo miisaamay) ee jaangooyooyinkii hore
   • Cabbirka xogta
   • Taariikhda saxsanaanta moodeelka
2. Tog dheer oo cidhiidhi ah, waa maxay dhibatada SGD cad ee soo gaadhay in xawligaasi gacan ka geysto hagaajinta?
   • Tiro ahaan buu buux dhaafiyaa
   • Waxay joojisaa xisaabinta gradients
   • Waxay si toos ah u kordhisaa heerka waxbarashada
   • Waxay sig-zag ka dhigtaa gidaarada dhaadheer iyadoo si tartiib ah dhulka uga sii socota
3. Qiimaha caadiga ah ee caadiga ah ee iskudarka firfircoonida waa:
   • 0.9
   • 0.0
   • 0.5
   • 2.0
4. Waa kuwee isbarbardhigga jireed ee inta badan loo isticmaalo in lagu qeexo firfircoonida?
   • Pendulum lulaya weligeed
   • Gu' ayaa dib u soo laabanaya
   • Kubad culus oo hoos u roganaysa
   • Biyo qaboojin
5. Qiyaastii intee in le'eg ayay dardargelintu kordhisaa tallaabada waxtarka leh ee jihada joogtada ah marka beta = 0.9?
   • Ilaa 2 jeer
   • Qiyaastii 0.1 jeer
   • Ma jirto wax kordhin
   • Ilaa 10 jeer
6. Wanaajiye casri ah kee ku dara qiyaasta daqiiqada koowaad ee qaabka dardargelinta?
   • Habka Newton
   • Aadam
   • K-macneheedu waa
   • Raadin random
7. In the update v = beta times v plus gradient, what happens when beta is zero?
   • The old velocity is doubled
   • Every gradient becomes zero
   • The previous velocity is discarded and the current gradient determines the step
   • The parameters stop depending on the learning rate
8. Why can repeated gradients with opposing sideways directions produce less zig-zagging under momentum?
   • Their opposing contributions can partially cancel in accumulated velocity
   • All past gradients are stored as separate parameters
   • The loss becomes convex automatically
   • Momentum ignores the current gradient entirely