As far as I know, none of these 3 work specifically in NLP, most of their work is in image processing and to the best of my knowledge none of them have any background in linguistics.
Their pricing page says that it costs them around 1$ to serve 80 searches. I really wonder how they arrive at that number. That seems shockingly expensive somehow.
> This seems like such a distant yet specific relationship type
I don't know much about it (never stepped foot outside EU/NA) but apparently some Asian countries have a pretty intricate system of kinship terms. For example Diagram IV in [1] shows different (Mandarin?) Chinese terms for what we would just call a "cousin" in English.
I also found the video by NativLang [2] on the topic pretty interesting.
scikit-learn (next to numpy) is the one library I use in every single project at work. Every time I consider switching away from python I am faced with the fact that I'd lose access to this workhorse of a library.
Of course it's not all sunshine and rainbows - I had my fair share of rummaging through its internals - but its API design is a de-facto standard for a reason.
My only recurring gripe is that the serialization story (basically just pickling everything) is not optimal.
Currently on course 2/4 in the series and it's great. Every week starts with a reading assignment from the RL book followed by a series of videos (re-)explaining stuff. The videos themselves are very nicely structured, with clear outlook and summary at the start and end of them.
Sutton himself appears in a couple of videos and there are other great guest lectures with interesting insights about applications.
I live in Switzerland too and especially Youtube never seems to know whether to serve me ads in German, French or Italian (even though Google should know I almost exclusively consume English content).
So I can sometimes watch several videos in a row, where I get served the same ad in a different language.
One of the most eye-opening moments of my CS studies was, when we discussed the Gibbard-Satterthwaite Theorem [1] in a class about Algorithmic Game Theory. I always like to bring it up in discussions about voting. In brief it states that every voting system is either
1. dictatorial (only 1 vote really counts)
2. limiting the number of outcomes to 2
3. is not strategy-proof (you have to vote tactically, for example ranking your favourite lower to achieve a better result)basically people have to 'game' the system to get what they want)
It would be interesting to know how different proposed system tackle this. [Iirc one of the more reasonable approaches seemed to be relaxing No. 3 to make strategic voting computationally hard].
GP explicitly stated a case where the US did the same. So for all intents and purposes the law does propagate 'magically' over the internet. Well, at least US law does.
I have to agree again. What you point out is correct and I hope to be able to act accordingly.
My personal take-away is still that these kinds of ethical ponderings are a bit like 'first-world problems'. I have the luxury that I can accept or deny any job based on my personal moral values. If the alternative was to have to go to some place where my life is significantly worse, I'd probably think twice too.
Yes, I did. That's when he explained his situation to me. I mean I was aware of his background, but up to that point I didn't really grasp how strongly that influenced his decision making. In the end I didn't really agree with his POV but I surely did understand and appreciate why he was thinking the way he did.
You know, I'm inclined to agree with your sentiment. I just think it might be more nuanced. At least to me my friend's suffering is more immediate. Whatever work he would have done there is multiple layers removed from me. I'm not saying that this is ok, but I feel it's just human.
Luckily the story ends with him getting a nice internship at another company that's less problematic.
I have a good friend who just finished his Master's in CS and considered taking an internship at Palantir. I was very apalled at first how he could even consider to work for a company that are obviously the 'bad guys'.
He comes from one of those countries your president considers a sh*thole and scoring that internship polishes his CV enough so that he can get a stable and high paying job which will allow him to stay and not have to go back there.
Can't really bame him....
> Additionally if you prefer lectures ETHZ has recordings of their ML class[1].
I took that class back in 2015. I found that the lectures were sometimes hard to follow, unless you already know the concepts (which creates a bit of a catch-22). For me the most valuable moments were, when Prof. Buhmann got sidetracked by some anecdotes. Would absolutely recommend, but maybe not as a starting point.