Experiment codes about different estimators of Max Mean
- WEandME.m: Implementation of Weighted Estimator in the ICML paper and Max Estimator.
- CAI.m: Implementation Cai's method to estimate |\mu|.
- AbsEst.m: Call CAI.m to estimate (\mu_1+\mu_2+|\mu_1-\mu_2|)/2, i.e, so called Absolute Estimator (AE).
- MSE_CP.m: Call WEandME.m and AbsEst.m to compare approximate MSE of AE,ME,WE.
- test_CAI.m: Call CAI.m to calculate MSE of estimating |\mu|.
- test_Hermite.m: Test the performance of Hermite Polunomial's estimation for higher moments
- MeanOfAbsMean.m: Test the performance of the optimal estimator in Cai's settings, i.e., estimate 1/n\sum_{i}^{n}|\mu_i|.
- tmp.m: Calculate values of Term 1 and Term 2 in equation (38) of Cai's paper.
- ChebyshevPolyCoeff.mat/HermitePolyCoeff.mat: Coefficients of corresponding polynomials are pre-calcualted and stored here.
- BayesianEst.m: Imlementation of Bayesian Estimator with prior N(0,a^2).
- ExpMaxEst.m: Csaba's estimator that calcualtes expectation of the max.
- Teststh.m: Call other functions to organise some experiments here.