Downloaded the app, searched for mexican restaurants in my area and it missed a few. I tried searching for the restaurants specifically and they were in the DB and categorized as mexican. Any recommended way to accomplish this?
To add another example, say a model learned that ice cream sales correlate well to forest fire rates. Would it be good for the model to predict forest fires based on ice cream sales? The answer is no, because there is no causal link.
Yes, but the question is not whether that's true, but whether that's useful.
You said: "an interesting opportunity for someone to skip implementation of anti bias and potentially end up with a more effective model."
Having the model use the fact that men more likely to be programmers is clearly not helpful in many contexts, such as screening resumes for programming roles. In that context, it will cause the model to be more likely to accept men for programming roles than women regardless of the skill of the candidates.
I agree to some extent. I'm not sure whether AI should be used for resume screening, but I'd lean towards no until biases are proven to not be an issue (if that's possible). There are obviously other areas where this is an important issue that we need to think critically about such as loans and criminal sentencing.
I think that you don't quite understand how these models pick up these biases. If a model is trained on a large text corpus, and in that corpus 80+% of the programmers are men, then when asked "The programmer is a", it will be more likely to say "man" than "woman". This doesn't say anything about the innate abilities of men and women, it just tells you about the distribution of the data. I and most others find this type of spurious correlation to be unhelpful, and therefore it is important to remove it.
Making sure that NLP algorithms are unbiased is important not just from a social justice perspective, but from a perspective of how useful the algorithms are. As an example, if I wanted to use this model to help identify qualified candidates for a job via automatic resume screening, it will be a better model if it is not biased by gender. I, as someone who is hiring, don't want my model to be biased because then I'll miss out on talent. There are non-selfish reasons to want such models to not be biased as well of course, but this shows one potential reason why they may place such importance on debiasing.
I actually really like apple's screen time limit capabilities. Being able to limit my time on twitch/youtube has been really helpful for me. However, one thing I would like to see is limiting the number of times I can open an app. Even if I limit email to 3 minutes, I can still open it 15 times a day :D
Figure this is as good a place as any to ask this: I have been looking for good books to read for learning about foreign policy. Didn't see any foreign policy lists on the linked cite, wanted to see if any users had recommendations :)
As a data scientist I often run code that takes hours, and getting a slack message when it finishes is super helpful, and it has often helped me catch when a script finishes too early due to some error.
Does anyone else reject the premise of the article? I understand that polling definitely missed big in some areas, but isn't that normal? Polling is really hard. You have to make a lot of assumptions, adjust for demographics, predict whether a voter will turn out, while using a sample of usually ~1000 people. Of course some polls are going to be wrong. I don't think the polling was generally wrong though. Polling averages had Biden +8 or so, when all votes are counted he'll likely be +5, which is within margin of error. Polls were extremely accurate in Georgia, where conventional wisdom would say Biden shouldn't win.
I get that most of the polling error benefited Trump, and the industry needs to reflect on why that happened and how to prevent underestimating republicans again in the future, but I don't yet buy the narratives that "the polls got it wrong."
I made this slack bot to track processes that take a while (I'm in the ML space), and get a notification when they finish. I thought this might be useful to others, and would love feedback on it.
I think that all the pain points of the article are a result of not using notebooks for their purpose. In my opinion, notebooks are good for:
1. POC/MVP: Showing that what you want to do will work before making a full structure.
2. Creating PDF/HTML documents with code and output.
3. Exploratory data analysis and visualization.
I think many of the data scientists in the article go well beyond what a notebook is. A notebook is where you start, but should never be a production tool.
I wouldn't call it artificially building something, but I suppose yes. I also find this code to be much more readable than generator code. Once you understand the function, all you have to do is pass it a function that returns a boolean.
Thanks so much! Want to get started with it before I decide to buy a book on it, but seems like a great language to program functionally in without all the hurdles :)