This video said about two parts, how to get to know a new field, how to read a paper,

Part one:

From paper sites such as arXiv, ieee, acm, etc to find keywords search papers, at the same time from the BBS, blog search, also get a rough impression at the same time, find you feel important N paper,

Don't intensive reading papers, each read 10%, is not important to skip, finally pick a few important intensive reading, and then choose from this article you think important for the next round of reading (similar to a wide search)

5-20 enough you understand and practice, 50-100 when you are a master,

Part two:

Read papers - Multi pass, each round to extract the information used to determine whether or not to the next round of

The first time a read-only title abstract figure
The second time the read-only type conclusion roughly sweep the rest of the figure, and sometimes can take a look at the related work
Third time to see the full text, skip formula
Fourth time skip don't understand (don 't make sense), could these places is not important