AWS Certified AI Practitioner

This is a classic sign of overfitting. Overfitting occurs when a model learns the noise and specific patterns of the training dataset (including spurious correlations) rather than the underlying generalizable relationships. As a result, the model achieves very high accuracy on the training data but performs poorly on unseen testing data because it does not generalize. You can detect overfitting by observing a large gap between training and validation/test performance (high training accuracy, low validation/test accuracy).

Common remedies include:

• Getting more diverse and representative training data so the model learns general patterns.
• Simplifying the model (reduce model complexity or number of parameters).
• Regularization (L1/L2, dropout, weight decay) to penalize overly complex models.
• Early stopping based on validation performance to prevent the model from continuing to fit noise.
• Cross-validation to better estimate generalization performance and tune hyperparameters.

These approaches help the model focus on the true signal in the data and improve performance on unseen test data.