Background on Text No. 5

In this post, I continue my parse of the Message in Carlini and Wagner's Towards Evaluating the Robustness of Neural Networks (2017 S&P). The content of the paper has not been changed, only enhanced in order to demonstrate just how that content is communicated.

Download my parse here. In it, I take Section III and I delineate just how the Message is structured clause for clause.

One quick note.

I have removed most of the punctuation, leaving only the punctuation needed inside a clause. My purpose in this has been to strip away the outer layer of sentences in order to expose to view the inner layers of clauses. The sentence is too poorly defined to serve as a unit of parsing. It's just a span of text that ends on a period. The clause, as you have seen, is a verb-based structuring of text. That's the level of precision needed to parse text.

Please email comments or questions to daniel.shea∂kit.edu

This blog is for you.