Improve README

README.md

@@ -1,12 +1,16 @@
 Coin Betting and Online Learning
 ================================
 
-Paper titled "From Coin Betting to Parameter-Free Online Learning" by
+Paper titled _From Coin Betting to Parameter-Free Online Learning_ by
 [Francesco Orabona](http://francesco.orabona.com/) and [Dávid
 Pál](http://david.palenica.com/) submitted to [COLT
-2016](http://www.learningtheory.org/colt2016/). The paper describes how to use
+2016](http://www.learningtheory.org/colt2016/).
+
+The paper describes how to use
 algorithms for repeated betting on outcomes of a coin to construct algorithms
-for online learning. As a coin betting algorithm one can use
+for online learning (learning with expert advice and online linear optimization
+over Hilbert space). As a coin betting algorithm one can use
 [Krichevsky-Trofimov
 estimator](https://en.wikipedia.org/wiki/Krichevsky%E2%80%93Trofimov_estimator),
-which is essentially optimal.
+which is essentially optimal, from which we derive two simple online learning
+algorithms.