When you observe that your model is "overfitting", there is a large gap between training accuracy and test accuracy. Which of the following approach(s) is/are considered as a good practice to avoid overfitting? Note: Please select all that apply.

If we are using a K-NN model, we can reduce the value of KK. If we are using a decision tree model, we can reduce the maximum depth of the tree. Collect more data and use them in the training dataset. Change the data in the training and validation set without adding any new data. Fine-tuning the parameters using the test set.