Pre-release part 2 videos


(Jeremy Howard) #1

See below for pre-release part 2 videos. Please do not distribute this to public channels such as twitter, blogs etc; I don’t want to publicly announce this until it’s complete. Notebooks and python modules are available at https://github.com/jph00/part2 .

We did not use wiki.fast.ai for this part of the course, but instead created forum wiki threads. Each lesson’s thread is linked below next to each video.


Any lessons talk about how to remove vocal from a song by deep learning?
(Jeremy Howard) #2

(Kent) #3

Since Part 2 is now visible to the public, this thread is also available to everybody. If we don’t want to distribute this yet, maybe @jeremy you want to list it somewhere else?


(Kouassi Konan Jean-Claude) #4

Thank you @jeremy to share it with the forum users !
I learned a lot from them.
I have written a publication on medium relatively to a previous post where I made a comparison of basic built-in functions per framework (necessary for paper replication).
I titled it : Basic neural networks patterns it is worth for a researcher to know, by platform

Any comment or suggestion are welcome.

Thanks.


(Jeremy Howard) #5

I posted on #part1 that the pre-release videos are now available to all. It’s a trial run so we can get feedback before the official launch.


(Jeremy Howard) #6

Thanks for sharing, @Kjeanclaude! Congratulations on creating this. It’s definitely useful information. I think there’s two things that could make this more effective:

  • Get a native English speak to edit your post. Current it is clear what you’re saying, but it sounds a little awkward to a native ear. Although it shouldn’t matter, unfortunately most people tend to associate non-native speech with lower credibility, so this is an important issue to deal with
  • Move your table from the spreadsheet into a standard web page (e.g. blog post). You could, for instance, have a paragraph for each section, and the list the comparative approaches as a standard <ul> or <dl> list.

(Kouassi Konan Jean-Claude) #7

Thank you very much @jeremy for your advice, I will take it into account and bring some improvements. I am firstly a french native and suggestions from native English are welcome.
I would also appreciate if you could give some examples of expression to review.
Thank you.


(Jeremy Howard) #8

I’m afraid I don’t have time at the moment to provide specific editorial help - maybe one of your fellow international fellows could help?


(Kouassi Konan Jean-Claude) #9

OK @jeremy, thank you very much. I will review it and ask for the opinion of native English on, then publish it on my personal website.


(magicly) #10

this course is so great. thanks @jeremy.
just finishing part1, fighting for part2:grinning:


(Prasad Chalasani) #11

Hi @jeremy in the intro of part 2 you mentioned you would be discussing Sequential models (RNN, etc) for structured sequential data , e.g. irregular time-series. I’m not able to find which lecture-video covers that. Could you please point me to it? Thanks!


(Jeremy Howard) #12

I don’t recall saying “irregular” time series. Lesson 14 covers time series in general. For irregular data, see http://arxiv.org/abs/1701.06675


(Prasad Chalasani) #13

Thanks @jeremy yes “irregular” was my addition :slight_smile: and thank you for that additional reference!


(Jeremy Howard) #15

(adhamh) #16

@jeremy why did you decide to not use the wiki? For those that missed the signed up and are following along after the forums can be difficult to navigate…


#17

I agree with @adhamh.
The course so far is great (as expected!) but it using the forum is not very practical.

If @jeremy is (understandably) busy maybe we can split the job among a few of us of creating those pages on wiki.fast.ai?


Are the Part 2 lesson notes available on the wiki as well?
(Jeremy Howard) #18

It seemed easier to create wiki pages on the forum. There didn’t seem to be many people using the original wiki. If some folks are open to creating a set of wiki pages for part 2, I’d be thrilled!


(Jeremy Howard) #19