Density Estimation and Combination under Model Ambiguity



Now, the same expansion used for KI21 can be applied to KI22(fθ*, g):

ki 22(fθ* ,g) =


1 XX μ fθ* (xi) - g(xi) v b (x )
n M   g(x
i)     f


1 X <fθ* (xi) - g(xi)A2 b, ʌ 1     1 τ

- 2n Σ —g(χi)— f fn(xi) = Jn1- 2J2


(96)


E Jin(fθ* ,g)) = E


K(u) (fθ* (x) g(x)) g(x + hu)dxdu.

(97)


Applying the same steps used for S2n we can show that


lim sup E
n→∞


(Jin(fθ*,g)) У


(fθ* (x) — g(x)) g(x)dx = E (fθ* (x) g(x))


lim E ( Jin(fθ* ,g)) = E (fθ* (x) g(x)).                                 (98)

n→∞

It follows that J1n(fθ*, g) = Op(1). Repeating the same steps once more for J2n(fθ*, g) we obtain:

E А XX μ fθ(xi) - g(xi) ..^ A = E (t μ fθ- (xi) -g(xi) V2 fn^^d

n i=1         g(xi)                                       g(xi)

=E


(fθ* (xi) - g(xi))2


g(xi)


fn(xi)dxi I = (( K(u)    ( ) g^ g( g(x + hu)dxdu,

g(x)


lim sup E (J2n(fθ* ,g)) ( (fθ (x) g(x))2 dx

n→∞

lim E (J2n(fθ* ,g))= ( (fθ* (x) g(x))2 dx > 0.

n→∞

Then also J2n(fθ* ,g) = Op(1). This implies that KI22(fθ* ,g) = Jni 2 Jn2 = Op(1).
Then it is clear that given assumptions A1-A4, if h 0,nh→∞then


(99)

(100)

(101)


KI22(fθ* ,g) p E (fθ* (x) g(x)) 2 У (fθ* (x) g(x))2 dx = E [lnfθ* lng],


(102)


this implies that nh1/2 KTI 22 p , hence we need to rescale it by dn = n-1h-1/2 where dn 0 as
This is embodied in assumption A6:


n →∞.


Finally we can put all terms together:


KI =


(ln

x


KTI22 ` ah1/2Cn


(103)


—~~                           —~~                 —~~

fTn(x) lnfbθ(x))fTn(x)dx == KTI1


—-~.

KTI2


(nh1/2) 1ιNι (nh1/2) 12N2 2 Cn


KI21(fb, fθ* ) + KI22(fθ* , g)J ,


(104)


since we showed that

32



More intriguing information

1. Constructing the Phylomemetic Tree Case of Study: Indonesian Tradition-Inspired Buildings
2. The name is absent
3. Accurate and robust image superresolution by neural processing of local image representations
4. The name is absent
5. The name is absent
6. Gianluigi Zenti, President, Academia Barilla SpA - The Changing Consumer: Demanding but Predictable
7. Disentangling the Sources of Pro-social Behavior in the Workplace: A Field Experiment
8. Three Policies to Improve Productivity Growth in Canada
9. Neighborhood Effects, Public Housing and Unemployment in France
10. Problems of operationalizing the concept of a cost-of-living index