Open Problem: Better Bounds for Online Logistic Regression

   Abstract