You built a bird classification model by randomly splitting the dataset into training and validation sets. Training accuracy is very high but the model performs poorly on the validation set. You discover the original dataset is imbalanced. What should you do to improve validation performance?

The model's poor validation performance despite high training accuracy, coupled with an imbalanced dataset, indicates that the random split likely resulted in an uneven distribution of classes between the training and validation sets. Stratified sampling ensures that the proportion of each class is approximately the same in both the training and validation sets as it is in the overall dataset. This provides a more representative validation set, allowing for a more accurate evaluation of the model's generalization ability. Collecting more data for majority classes would worsen the imbalance. Training on a smaller subset of data would likely lead to underfitting. Systematic sampling does not guarantee class distribution preservation.