CS230 Deep Learning Course—Andrew Ng teaches you how to read papers

CS230 Deep Learning Course—Andrew Ng teaches you how to read papers

One

1、compile list of papers
2、skip around the list
3、maybe you initially start off with five papers
4、go from the very efficient high information content first and then go to the harder material later

TWO—some rough guidelines

1、if you read 15 to 20 papers ,you will have a basic understanding of an area .Maybe good enough to do some work, apply some algorithms.
2、if you read 50-100 papers in an area like speech recognition and kind of understand it, then that’s probably enough to give you a very good understanding of an area.
3、where can we get ML knowledge?

• twitter
• ML subreddit
• NIPS、ICML、ICLR
• Friends

THERE—how to read a paper？

1、take multiple passes through the paper
2、steps
①read the title and the abstract and the figures
②read more carefully about the intro, the conclusions, look carefully at all the figures again and then skim the rest
③read but just skip the math
④read the whole thing but skip parts that don’t make sense.

FOUR—after reading the paper, what we should learn?

1、what did the authors try to accomplish?
2、what were the key elements?
3、what can you use yourself?
4、what other reference do you want to follow?

FIVE—about math

1、take detailed notes and then see if you can re-derive it from scratch

SIX—about codes

1、the simple lightweight version and one of learning would be to download and run the open source code if you can find it, and a deeper way to learn this material is to re-implement it from scratch.
2.it is easy to download an open sourcing and run it and say ooh, it works. But if you can re-implement one of these algorithms from scratch then that’s a strong sign that you’ve really understood this algorithm.

SEVEN—longer term advice

1、The more important thing is for you to learn steadily not for you to have a focus intense activity.
2、you are actually much better off reading two or three papers for a week.

EIGHT—advice on machine learning career

1、how to get a position?
2、selecting a position.
Recruiters look for?

• skills(ML quiz ,coding)
• meaningful work

• recruiters are not impressed by volume, having done 10 lame projects is actually not impressive, not nearly as impressive as doing one great project or two great projects.
  3、Foundation skills
• course work
• reading
• relevant project
  4、selecting a job
• work with great people/projects
• focus on the team(10-30 persons) you will interact with.
• Manager
• not on the brand

NINE—notes