In deep learning, what is “back propagation” responsible for?
Adjusting weights and biases based on mistakes
Explanation:
Backpropagation is a training method used in deep learning to improve the model’s accuracy. It works by calculating errors from the output, sending the error information backward through the network, and adjusting weights and biases to reduce mistakes.
Why not the others?
- b) Tagging friends in pictures → This is an application of AI, not the role of backpropagation.
- c) Identifying emotions in pictures → This is a task performed by a trained model, not the training process itself.
- d) Uploading data to the network → This is handled by the input layer.
