Click here to consult the table of the estimation rates, and here for the testing rates. Here for the rates of some statistical problems under privacy constraints. Special thanks to Alexander Kent, who’s helping me filling the tables.
A brief summary of the main ideas in the paper. Here’s the pdf .
I gave it go to these two problems. To be honest, I probably cheated in the second one, as the constant estimator is always admissible in the frequentist framework. You can find my proposed solutions here.
This video gives an intuitive explanation why the James-Stein estimator dominates the sample average when the dimension is sufficiently large (greater than 2). Click the image to watch! Special thanks to Alexander, who shared this video with me.
I’ll let John Conway say why this book is so relevant, citing from the preface of the book: " “How to Solve It” is a wonderful book! This I realized when I first read right through it as a student many years ago, but it has taken me a long time to appreciate just how wonderful it is.