Quiz collegato alla guida · Medio Livello
Quiz sulla propagazione inversa
Testare i gradienti di backpropagation, l'ordine di dipendenza inversa, i valori forward memorizzati nella cache, la moltiplicazione delle regole di catena, le somme dei percorsi condivisi e la separazione dell'ottimizzatore.
Domanda 1 di 8
What quantity does backpropagation compute for a neural network's relevant weights?
Anteprima di ogni domanda8 domande
Le risposte e le spiegazioni vengono visualizzate mentre rispondi al quiz.
- What quantity does backpropagation compute for a neural network's relevant weights?
- Derivatives of the loss with respect to those weights
- A causal percentage of total error uniquely assigned to each weight
- The architecture drawing without any calculation
- A permanently updated set of weights without an optimizer
- Quale regola matematica è al centro della propagazione inversa?
- Il teorema di Pitagora
- Teorema di Bayes
- La regola della catena del calcolo
- La formula quadratica
- Why can reverse-mode backpropagation compute many parameter derivatives efficiently?
- It avoids all local derivatives
- It reuses graph structure and relevant intermediate results
- It calculates only the output prediction again
- It guarantees every implementation has exactly one forward pass of runtime
- In quale ordine la propagazione inversa calcola i gradienti?
- Dal livello di input verso il livello di output
- Tutti gli strati contemporaneamente senza ordine
- Solo per lo strato nascosto centrale
- Dal livello di output indietro verso il livello di input
- Does computing backpropagation gradients by itself establish that model weights were updated?
- Yes; every backward computation automatically applies a learning rate
- Yes; gradient values always replace the weights directly
- No; the optimizer update is a separate step
- Only if the network has no parameters
- In a conventional cached-backpropagation workflow, what does the forward computation provide for later derivative calculations?
- Relevant intermediate values that can be retained for the backward computation
- A completed optimizer update without any update rule
- A guarantee that all later evaluation errors disappear
- A permanent deletion of the training objective
- On one composed backpropagation path, local scalar derivatives are three and four. What is their chain-rule product?
- Sette
- Uno
- Three quarters
- Twelve
- A shared parameter has loss-gradient contributions two and five through two paths. What combined contribution is accumulated?
- Dieci
- Sette
- Tre
- Two fifths