Consider a linear model: g=Tx The regression problem we want to solve is =Rdargminn=1N(yng(xn))2, where D={(xn,yn)}n=1N is the training dataset. Putting the equation into the matrix form, we know that the optimization is equivalent to =RdargminEtrain()yX2. Derive the solution . State the conditions under which the solution is the unique global minimum in terms of the rank of X. Suggest two techniques that can be used when XTX is not invertible