E6998 Robustness and Security in ML Systems

Spring 2018 -- Junfeng Yang

Reading

Read this article about how to read a research paper. The take-home message is that until you can answer a bunch of questions, you are not done reading a paper. William lists a number of important questions. I would add two: 1) What are the re-usable principles/tricks/algorithms presented in this paper? 2) What is the (authors') insight that drives the research? A system research paper often has a bunch of novel tricks. I believe the more such things you have in your toolbox, the more likely you can come up with an elegant/novel system design.

Presentation

Some advice on how to give a good presentation:

Some online advice from others:

Read this paper about things to avoid when giving a talk
Read this paper about how to give a good conference talk. Many ideas apply to the mini-talk you'll give.

Writing

Some suggested readings (to make you a better writer):
Read this paper about how to write a technical paper. Many ideas apply to writing proposals as well.
Read this paper about how to write sentences, paragraphs, etc.
And, of course, read Strunk and White. Many times.