Bandit Algorithms

14.5 Exercises 187

τ∈[n] almost surely. Show that

D(P|Fτ,Q|Fτ) =EP

[τ

∑

t=1

D(Pt(·|X 1 ,...,Xt− 1 ),Qt(·|X 1 ,...,Xt− 1 ))

]

.