The scheme described by Gus Gutoski will protect a parent private key from the release of a specific number of its child private keys.
That is exactly the paper I linked in the original post. My usecase needs to publish lots of private keys, so this is not usable.