{"id":"arxiv/2604.26926","url":"https://arxiv.org/abs/2604.26926","tags":["cs.LG","math.OC","stat.ML"],"score":0,"title":"A Note on How to Remove the $\\ln\\ln T$ Term from the Squint Bound","domain":"cs.LG","source":"arxiv","authors":["Francesco Orabona"],"reviews":0,"subject":"cs.LG","summary":"In Orabona and Pál [2016], we introduced the shifted KT potentials, to remove the $\\ln \\ln T$ factor in the parameter-free learning with expert bound. In this short technical note, I show that this is equivalent to changing the prior in the Krichevsky--Trofimov algorithm. Then, I show how to use the same idea to remove the $\\ln \\ln T$ factor in the data-independent bound for the Squint algorithm.","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-04-29","reactions":{}}