search

ai-se / ZheYu

Zhe's works
0 stars 0 forks source link

01/10/2017 #17

Open azhe825 opened 7 years ago

azhe825 commented 7 years ago

DOING:

  1. Transfer learning in SLR

DONE:

  1. tune lda to get a stable similarity score
  2. find a way to combine UPDATE and REUSE (therefore no need to decide which method to use before review)

NOT DONE:

  1. Drafting the paper. It is really really hard to do this in China, given the difficulty to look up literatures when google scholar is not available. Therefore I prioritized the experiments over writing. Will get back to writing jobs as soon as I am back to US.

TODO:

  1. Will decide if further experiments are necessary when full result of UPDATE_REUSE comes out.
  2. Some experiments on time decaying model.

ADMIN

  1. Any feedback on UPDATE_REUSE?
timm commented 7 years ago

China, given the difficulty to look up literatures when google scholar is not available

did not know that. amazing. nothing equivalent in china?

azhe825 commented 7 years ago

not really. only thing we can do is to use really slow vpn to get to google scholar.

Anyway, I will make up the time lost on paper writing as soon as I get back to US