Ikibazo gihuza ibibazo · Hagati Urwego

Kubura no Guturika Gradients Ikibazo

Kugenzura urunigi-amategeko agenga ibicuruzwa, kwiyuzuzamo, uburebure bwikurikiranya, inzira zisigaye nimbibi za spécran-radius heuristics.

Inzira zijyanye no kuyoboraKurimbuka no Guturika Gradients
Ikibazo 1 Bya 8

Ni ubuhe buryo bw'imibare muri backpropagation nintandaro yo kuzimira no guturika gradients?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Ni ubuhe buryo bw'imibare muri backpropagation nintandaro yo kuzimira no guturika gradients?
    • Kugwiza inshuro nyinshi kumurongo ukoresheje itegeko ryumunyururu
    • Ongeraho gradients murwego rumwe
    • Kugabanya igihombo nubunini bwicyiciro
    • Gufata kare kare imizi yuburemere
  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. Ni ubuhe bwubatsi BURUNDU bugira ingaruka zikomeye kubibazo bigenda bikurikirana?
    • Ibice bibiri
    • Gusubira inyuma kumurongo umwe
    • Umuyoboro uhora usubiramo udafunguye hejuru yigihe kinini
    • Imbonerahamwe
  4. Kubona igihombo cyamahugurwa gihinduka NaN birashoboka cyane byerekana ikibazo ki?
    • Kurimbuka
    • Guhuza neza
    • Ntoya cyane igipimo cyo kwiga
    • Ibikoresho biturika
  5. Nigute amasano asigaye (skip) afasha mukuzimya gradients?
    • Batanga gradient inzira itaziguye inzira isubira inyuma, kurenga kugabanuka kugwira
    • Bakuraho ibintu byose bidafite umurongo
    • Bongera igipimo cyo kwiga mu buryo bwikora
    • Basiba ibice byambere
  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