Optimal estimation
Log likelihood of
M
given {(
u
i
,
v
i
)}
How do we minimize
C
?
Non-linear regression (least squares), because
û
i
and
v
i
are non-linear functions of
M
