Suppose you are given a data set with x (the marketing budget in TV in thousand dollars), x2 (the marketing budget in social media in thousand dollars), y (the sales of the product in thousand units). A simple neural network is constructed below to model the relationship among the variables. X1 X2 1.5 -2, 1.2 1 h1 h12 1 -1 2 2 h21 h22 -1 1 y The activation functions used for all layers are ReLU (rectified linear unit). Manually compute the predicted sales of the product for the following scenarios. Scenario 1 2 Budget in TV 3,500 1,500 Budget in social media 1,200 4,000