A machine learning team trains a neural network and observes that training loss decreases smoothly but validation loss exhibits high variance across epochs. Which regularization technique specifically reduces variance by injecting Gaussian noise into the input layer during training?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — c is correct because adding Gaussian noise directly to the input layer during training is a regularization technique that increases data diversity and reduces overfitting, with a regularization effect equivalent to Tikhonov regularization under certain assumptions. A (L2) penalizes weights, not inputs.
Full explanation below image
Full Explanation
C is correct because adding Gaussian noise directly to the input layer during training is a regularization technique that increases data diversity and reduces overfitting, with a regularization effect equivalent to Tikhonov regularization under certain assumptions. A (L2) penalizes weights, not inputs. B (Dropout) zeros hidden activations, not inputs. D (Batch Normalization) normalizes activations and reduces internal covariate shift but is not input noise injection.