AWS Certified Machine Learning – Specialty — Question 365

A machine learning (ML) specialist is developing a model for a company. The model will classify and predict sequences of objects that are displayed in a video. The ML specialist decides to use a hybrid architecture that consists of a convolutional neural network (CNN) followed by a classifier three-layer recurrent neural network (RNN).

The company developed a similar model previously but trained the model to classify a different set of objects. The ML specialist wants to save time by using the previously trained model and adapting the model for the current use case and set of objects.

Which combination of steps will accomplish this goal with the LEAST amount of effort? (Choose two.)

Answer options

• A. Reinitialize the weights of the entire CNN. Retrain the CNN on the classification task by using the new set of objects.
• B. Reinitialize the weights of the entire network. Retrain the entire network on the prediction task by using the new set of objects.
• C. Reinitialize the weights of the entire RNN. Retrain the entire model on the prediction task by using the new set of objects.
• D. Reinitialize the weights of the last fully connected layer of the CNN. Retrain the CNN on the classification task by using the new set of objects.
• E. Reinitialize the weights of the last layer of the RNN. Retrain the entire model on the prediction task by using the new set of objects.

Correct answer: D, E

Explanation

To adapt a pre-trained model using transfer learning with the least effort, only the final classification layers should be reinitialized and retrained while preserving the feature extraction layers. Reinitializing only the last fully connected layer of the CNN (Option D) and the final layer of the RNN (Option E) allows the model to learn the new classes efficiently. Retraining entire networks or large subsections (Options A, B, and C) requires significantly more computational effort and training time.