Batch represent takes a long time - can you make it incremental?

[Andrew Henson]

If I add one more person to say, a classifier with an existing 500 people, the batch represent processes all of the representations again.

Is there any way to make batch represent incremental, so that it only processes the new data?

I'd be interested to follow up any suggestions, as this could save me (and others) some time :)

Thanks everyone.