Starting from:
$35

$29

HOMEWORK 1 – V3 Solution

1. Gaussian mean estimation.

1.1. Optimal shrinkage factor [10pts]. Let X1; X2; :::; Xn 2 Rd be i.i.d. multivariate Gaussian random vectors, i.e., Xi N ( ; 2I). Denoting the sample mean estimator with ^ , 1 Pn Xi,
n    i=1


consider an estimator of the form ^s =  1

^: Find the optimal  that minimizes the risk

2

R(^s;  ) = E[k^s    k22].
k^k2




1.2. Generalizing Stein’s lemma [10pts]. Let X p (x) and g : Rd ! Rd where p (x) and g (x) are di erentiable w.r.t and x, and let E[g (X)] = ( ) for some function . Show that

    (a) E[rx log p (X)g (X)>] + E[rxg (X)] = 0,
    (b) E[r log p (X)g (X)>] + E[r g (X)] = r  ( ).

1.3. Generalizing SURE [10pts]. Let X N ( ; ) where 2 Rd and 2 Rd d. If ^(X) 2 Rd is an estimator of the form X + g(X) where g : Rd ! Rd is di erentiable. De ne the functional

S(X; ^) = Tr(  ) + 2 Tr(  rxg(X)) + kg(X)k22:

Then show that S(X; ^) is an unbiased estimator of the risk, i.e., E[k^(x)    k22] = E[S(X; ^)].

2. Exponential families.

2.1. Second moment [10pts].    For a random variable X    p (x) = exp(h ;  (x)i    ( )), let

E[ (X)] =  . For    2 Rd,  nd Tr(E[( (X)    )( (X)    )>]) in terms of    and ri    ( ) for i    0.

2.2. Score function [10pts]. Assume that X p (x) where p is not necessarily in the expo-nential family form. Denote the log-likelihood by ‘ (x) = log p (x), show that

    (a) E[r ‘ (X)] = 0.

    (b) E[r ‘ (X)r ‘ (X)>] =  E[r2‘ (X)] (Problem 1.2 may be helpful).

2.3. Maximum entropy principle [Bonus 2pts].  Assume that p(x) is a probability mass function of a discrete random variable taking values from a  nite set X . Entropy of p is de ned as H(p) = Px2X p(x) log p(x). For   : X ! Rd, show that the maximum entropy distribution satisfying
Ep[ (X)] =    2 Rd is a member of exponential family. That is, show that the solution to
maximizeH(p) subject to: Ep[ (X)] =    ;
p

is an exponential family. (Hint: Write the Lagrangian associated with the above optimization problem. Since X is nite, think of p(x) as a vector and maximize over it.)
1

More products