Optimizers

__________ responsible to update weights in the back propagation to minimize the loss function

• Optimizers

• PCA

• input layer

• hidden layers

___________ is the newest fact and most efficient optimizer up till now

• Gradient descent

• Stochastic Gradient descent

• Adam

• adagrad

Gradient descent needs more computational powerthan other optimizers.

• True

• False

_________________ use the exponential weighted average to smooth the noisy result of Stochastic gradient descent.

• Stochastic gradient descent with momentum

• Gradient descent

• adagrad (adaptive gradient descent)

• Ada delta and RMS prop

___________ use a NON fixed learning rate

(choose all the possible answers)

• Adagrad

• Ada delta and RMS prop

• Adam

• Stochastic gradient descent

• stochastic gradient descent with momentum

__________ is a combination of SGD with momentum for smoothing and RMS prop for efficient Learning rate.

• Adam

• Gradient descent

• Ada delta

• SGD