(7) TensorFlow and ANN (6 marks) (a) Why the softmax function is suitable for multiple classification but not regression? (1 mark) (b) Modify the build_model() function and param_distribs at Pages 30 and 31 of Lecture 7 slides to implement a process of fine-tuning an MLP with two possible activation functions (i.e., "RELU" and "sigmoid" functions). In other words, treat the two activation functions as hyperparameters and use the grid search method to determine the best hyperparameters. (All other hyperparameters are up to you.) Present the Python code to illustrate your implementation