Minor update to multi-label description in readme

README.md

@@ -68,4 +68,4 @@ CAP and RAP can be combined, of course, by applying both a kernel constraint and
 
 Multi-label
 -----------
-AutoPool directly generalizes to multi-label settings, in which each instance x receives multiple positive labels. In this setting, a separate auto-pooling operator is applied to each class. Rather than a single parameter α, there is a vector of parameters α_c where c indexes the output vocabulary. This allows a jointly trained model to adapt the pooling strategies independently for each category. Please see the [paper](http://www.justinsalamon.com/uploads/4/3/9/4/4394963/mcfee_autopool_taslp_2018.pdf) for further details.
+AutoPool directly generalizes to multi-label settings, in which multiple class labels may be applied to each instance x (for example "car" and "siren" may both be present in an instance). In this setting, a separate auto-pooling operator is applied to each class. Rather than a single parameter α, there is a vector of parameters α_c where c indexes the output vocabulary. This allows a jointly trained model to adapt the pooling strategies independently for each category. Please see the [paper](http://www.justinsalamon.com/uploads/4/3/9/4/4394963/mcfee_autopool_taslp_2018.pdf) for further details.