3
$\begingroup$

I often come across Keras code that adds GaussianNoise to the input, however its not clear to me what advantages does it offer to the learning.

input_img = layers.Input(t_x.shape[1:], name = 'RGB_Input') 
pp_in_layer = layers.GaussianNoise(GAUSSIAN_NOISE)(input_img)
$\endgroup$
3
$\begingroup$

Adding noise in the input data is equivalent to adding a regularization term to the objective function.

See Training with Noise is Equivalent to Tikhonov Regularization and Adding noise to the input of a model trained with a regularized objective.

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.