Deep Learning for Big Data

Correct Answer: To control the step size of weight updates during training

The learning rate controls how much the weights of the network are updated during each iteration of training. A larger learning rate can lead to faster convergence, but may cause the network to overshoot the optimal weights.

Correct Answer: To use knowledge gained from training one model to improve the performance of another model

Transfer learning involves using knowledge gained from training one model to improve the performance of another model. This can involve using a pre-trained model as a starting point for a new model, or fine-tuning a pre-trained model to adapt it to a new task. The other answer choices listed are not accurate descriptions of transfer learning.

Correct Answer: Binary cross-entropy

Binary cross-entropy is typically used for binary classification tasks, which involve predicting a binary output (e.g. 0 or 1). Other loss functions listed are typically used for other types of tasks. Mean squared error is often used for regression tasks, categorical cross-entropy is used for multi-class classification tasks, hinge loss is used for support vector machines, Kullback-Leibler divergence is used for probabilistic models, and absolute error is used for regression tasks.