Krichevsky–Trofimov estimator

Template:One source In information theory, given an unknown stationary source Template:Pi with alphabet A and a sample w from Template:Pi, the Krichevsky–Trofimov (KT) estimator produces an estimate p_i(w) of the probability of each symbol i ∈ A. This estimator is optimal in the sense that it minimizes the worst-case regret asymptotically.

For a binary alphabet and a string w with m zeroes and n ones, the KT estimator p_i(w) is defined as:

$\begin{array}{cc}{p}_0(w)& =\frac{m+1/2}{m+n+1},\\ p_1(w)& =\frac{n+1/2}{m+n+1}.\end{array}$

This corresponds to the posterior mean of a Beta-Bernoulli posterior distribution with prior $1/2$. For the general case the estimate is made using a Dirichlet-Categorical distribution.

See also

• Rule of succession
• Bayesian inference using conjugate priors for the categorical distribution

References

Template:Probability-stub