I have read parts of it. It arguably was already "outdated" back then, as it mostly focused on abstract mathematical theory of questionable value instead of cutting edge "deep learning".