Google Professional Machine Learning Engineer
Get started today
Ultimate access to all questions.
Ultimate access to all questions.
You are working on a Neural Network-based project that involves training a model on a dataset with columns exhibiting significantly different ranges of values. During the data preparation phase for model training, you encounter difficulties with gradient optimization, as the optimization process struggles to adjust weights effectively. What should you do to address this issue?
Explanation:
The correct answer is B: Use the representation transformation (normalization) technique. When the dataset has columns with different ranges, it can cause issues with gradient descent optimization because the step sizes for different features may differ, which can prevent the model from converging effectively. Normalization adjusts the range of the features to be on a similar scale, ensuring that the gradient descent steps are more uniform and helping the model to converge more smoothly.