How do you mathematically define the objective function of a linear regression model, and how does it relate to maximum likelihood estimation?