APPLICATION OF SUPERVISED LEARNING
NEURAL NETWORK
|
Question
[CLICK ON ANY CHOICE TO KNOW THE RIGHT ANSWER]
|
Backpropagation calculates the ____ and propagates it back to earlier layers.
|
|
value
|
|
|
error
|
|
|
data
|
|
|
None of the above
|
Explanation:
Detailed explanation-1: -Answer. Answer: Backpropagation calculates the algorithm and propagates it back to earlier layers.
Detailed explanation-2: -Backward pass at the output layer The error on weight w is calculated by differentiating total error with respect to w.
Detailed explanation-3: -The backpropagation algorithm is an efficient way to compute the error ℓ iteratively using error ℓ+1, so we proceed by computing the error from the last layer L, again, using the chain rule:(20.8)jL=∑k∂Remp∂akL∂akL∂zjL, and, when k≠L, the terms vanish and we obtain:jL=∂Remp∂ajL∂ajL∂zjL=∂Remp∂ajL′(zjL).
There is 1 question to complete.