What are skip connections in NN models (I assume that layers are numbered in increasing order starting from the input layer up to the output layer)? Skip connections skips some of the layers in the neural network architecture and feeds the output of layer k to the input of layers of index at least k+2. Skip connections are connections that run backwards Skip connections skips some of the layers in the neural network architecture and feeds the input of layer k to the output of layers of index at most k+2. What does the term 'dropout' refer to in the training of NN models? Dropout removes a fraction of the neurons and connections from the neural network in order to increase the complexity of the network. Dropout removes a fraction of the neurons and connections from the neural network in order to decrease the complexity of the network. Dropout inserts new neurons and connections in the neural network in order to decrease the complexity of the network. Consider the following code: nnetmode/