Kediska Gradients Burburinta iyo Qarxaya

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Hawlgalkee xisaabeed ee dib-u-faafintu waa sababta asaasiga ah ee luminta iyo qarxinta gradients?
   • Si isdaba joog ah ugu dhufashada qodobbada lakabka kasta iyada oo loo marayo xeerka silsiladda
   • Ku-daridda gradients ee hal lakab
   • U qaybinta khasaaraha iyadoo loo eegayo cabbirka dufcadda
   • Qaadashada xidid laba jibaaran ee miisaanka
2. Why can sigmoid or tanh units contribute to vanishing gradients when they are saturated?
   • Their derivatives are always larger than one
   • Their derivatives become small in saturated regions, so repeated chain-rule products can shrink
   • They remove the chain rule from backpropagation
   • They force every weight to zero
3. Nashqadadee ayaa sida ba'an u saamaysay dhibaatooyinka isjiid jiidka ee taxanaha dheer?
   • Dareenka laba lakab
   • Dib u noqosho toosan oo keli ah
   • Shabakad neerfaha ah oo soo noqnoqda ayaa furfurmay marar badan
   • Miis wax laga eego
4. Aragtida luminta tababbarka si lama filaan ah isu rogtay NaN waxay u badan tahay inay tusinayso dhibkee?
   • Qaybaha baaba'aya
   • Isku dhafka saxda ah
   • Aad u yar heerka waxbarashada
   • Jaangooyooyin qarxa
5. Sidee bay isku xirnaanta hadhaaga (boodboodka) u caawiyaan jajabinta baaba'a?
   • Waxay siinayaan jaangooyooyinka dariiq gaaban oo toos ah gadaal, iyagoo ka gudbaya isku dhufashada sii yaraanaya
   • Waxay meesha ka saaraan dhammaan waxyaabaha aan tooska ahayn
   • Si toos ah ayay u kordhiyaan heerka waxbarashada
   • Waxay tirtiraan lakabyada hore
6. In a simplified analysis of a recurrent linear transformation repeated over many steps, what matrix quantity is a rough signal of growth versus decay?
   • The matrix trace relative to the batch size
   • The spectral radius relative to one
   • The number of hidden units alone
   • The determinant relative to the learning rate
7. A backpropagated gradient is multiplied by a factor below one at each of many layers. What trend should you expect?
   • It must become exactly one
   • It grows exponentially regardless of the factor
   • The signal can shrink exponentially with depth
   • It stops depending on earlier layers after one step
8. Why can a recurrent network be especially vulnerable over a long sequence?
   • The network has no hidden state
   • The output layer is used only once
   • The sequence is processed without a chain rule
   • The recurrent transformation is applied repeatedly across timesteps, compounding its gradient effect