Which of the following is NOT an advantage of employing dropout in RNNs?
Preventing overfitting
Improving robustness to initialization conditions
Reducing noise in the hidden state
Enhancing computational efficiency

Machine Learning Exercises are loading ...