Kordhinta Dalabka Labaad iyo Kediska Hababka Newton

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Macluumaad noocee ah ayuu habka Newton u isticmaalo ee aanu faracyada cad ee farcanka ahi u isticmaalin?
   • Curvature, iyada oo loo marayo derivatives labaad ee shaxanka Hessian
   • Goobta saxda ah ee ugu yar ee caalamiga ah
   • Jadwalka heerka waxbarashada
   • Tirada tusaalooyinka tababarka
2. Ujeeddo afar-geesood ah oo dhammaystiran, imisa tillaabo ayuu habka Newton u baahan yahay si uu u gaadho ugu yaraan?
   • Sax toban tallaabo
   • Hal tallaabo
   • Isla lambar la mid ah soo degitaanka gradient
   • Marna iskuma xidhna
3. Waa maxay sababta buuxda ee habka Newton aan waxtar u lahayn shabakadaha neerfaha ee bilyan-parameter?
   • Jaangooyeeyaha lama xisaabin karo
   • Kala soocida labaad had iyo jeer waa eber shabakadaha neerfaha
   • Hessian-ku waa N-by-N, oo qiimihiisu yahay xusuusta N-squared iyo xisaabinta N-cubed si loo rogo
   • Waxay kaliya ku shaqeysaa sawirada
4. Maxay sameeyaan hababka quasi-Newton sida BFGS si looga fogaado kharashka Hessian?
   • Gabi ahaanba way iska indho-tiraan qulqulka
   • Waxay si aan kala sooc lahayn u qiyaasaan qalooca
   • Waxay u rogaan dhibaatada mid toosan
   • Waxay dhisaan qiyaas ahaan Hessian-ka rogan ee kala duwanaanshiyaha is daba jooga
5. Sidee L-BFGS u dhimaysaa xusuusta marka la barbar dhigo BFGS?
   • Waxa ay kaydisaa oo kaliya dhawrkii u dambeeyay ee gradient iyo tallaabo vectors halkii ay ka heli lahayd matrix buuxa
   • Waxa ay isticmaashaa sabbooyin sax ah oo keliya
   • Waxay tuuraysaa jaan-goynta tillaabo kasta ka dib
   • Waxay ku tababartaa cabbirro yar
6. Waa kuwee habka casriga ah ee qiyaasa curvature si loo dedejiyo tababbarka shabakada neerfaha ee waaweyn?
   • Hal-kulan codayn
   • Hababka Kronecker-factored sida K-FAC iyo Shaambo
   • Hoos-u-dhac stochastic ah oo cad oo aan wax ka beddel ah lahayn
   • Ka saarida
7. On an exact quadratic objective with an invertible Hessian, what assumption supports the one-step Newton result?
   • The learning rate is reduced to zero before the update
   • The Hessian is ignored and only a fixed scalar step is used
   • The full Newton update uses the exact gradient and Hessian of that quadratic
   • The objective is first converted to a linear function
8. A non-quadratic objective is optimized with Newton’s method. What role does its quadratic model play?
   • It guarantees the global minimum in one step
   • It replaces the need to evaluate a gradient
   • It makes every Hessian positive definite
   • It approximates local curvature, so later iterations may be needed as the model changes