Question (3): State whether (true) or (false) and justify your answer:- a- Increasing the depth of neural networks has been leading to much error reduction for any activation and loss functions design. b- If training a neural network does very well and testing is not good, then we need to add some parameter regularization. C- If training a neural network does very well, testing will do good. d- Hog feature extraction is one of the hand crafted design techniques. e- Adaptive thresholding is used for segmenting zebra-texture images