Which action is recommended to provide the HIGHEST accuracy model for the company’s test and validation data?
Increase the randomization of training data in the mini-batches used in training
Allocate a higher proportion of the overall data to the training dataset
Apply L1 or L2 regularization and dropouts to the training
Reduce the number of layers and units (or neurons) from the deep learning network
Explanations:
Increasing the randomization of training data can help the model generalize better. However, it does not directly address overfitting, and does not guarantee a significant improvement in test accuracy.
Allocating a higher proportion of data to training will improve the model’s ability to learn, but it does not address the issue of overfitting, which is the main concern here.
Applying L1 or L2 regularization and dropouts reduces the model’s complexity, helping to prevent overfitting and boosting generalization, leading to better performance on test data.
Reducing the number of layers and neurons might reduce overfitting, but it also risks underfitting and losing the model’s ability to learn complex patterns, which could lower accuracy on both training and test data.
I figure that the answer is:
Apply L1 or L2 regularization and dropouts to the training