Dense layer does the below operation on the input and return the output. output = activation(dot(input, kernel) + bias)