News for hackers. Stuff that matters.

zmjjmz

Do believe that a human learning from a book is fundamentally different than training a model off of it, and thus should be regulated differently?
Re:The industry behind the industry behind AI on Tuesday July 04, 2023 @10:17AM (#36587033)
I'm curious if you feel strongly about the 'open internet' as well, i.e. that platforms and content should be publicly available by default (as opposed to e.g. Elon's recent move to restrict the viewability of tweets by not logged in users).

To me, these datasets don't come from some nefarious violation of privacy (surveillance) on the people who've written or otherwise authored their content, but from an aggregation of public content that was previously uncontroversial.

If these data were pulled from private content, it would be a different story - although I'm sure that many companies TOS allows the use of your 'private' content to train their models.

Going forward I wonder if we may see an 'AI opt-out' for your data regulated (I'm not sure if the new EU act stipulates this)

Re:A gentle introduction to vector databases on Wednesday February 23, 2022 @10:28AM (#30442054)
We've had good success with Vespa [1] for this usecase, both using it to rank by vector similarity and apply business logic oriented filters.

Search latencies are fairly low, even with constantly updating documents.

[1] https://tech.okcupid.com/vespa-vs-elasticsearch-for-matching...

Hey I meant to track one of y'all down at the MLOps conference, but didn't get the chance. I've built a very shitty version of a cached-execution DAG thing internally, and one of the design decisions I made was to have it so that parent nodes in the DAG don't need to know anything about child nodes. This allows for larger DAG builders to be more easily subclassed.

MetaFlow doesn't do that -- instead each 'step' has to know what to call next, which means that if I wanted to subclass e.g. the MovieStatsFlow in [here](https://github.com/Netflix/metaflow/blob/master/metaflow/tut...) and say, add some sort of input pre-processing before the compute_statistics call, I'd essentially end up having to either override what compute_statistics does to not match its name _or_ copy-past e that first step just to replace that last line.

I'm sure this design decision was considered and/or that use-case doesn't come up a lot at Netflix (although I've encountered that a lot), or maybe I'm missing something very obvious, but I'd love to hear your thoughts on that.

Re:Hacker news for data scientists on Wednesday December 03, 2014 @05:33PM (#8696777)
When I was last here it was mostly post-for-post consistent with /r/MachineLearning.
Re:Elon Musk: To the People of New Jersey on Saturday March 15, 2014 @03:51PM (#7406070)
Hopefully those updates are signed, in which case as long as there aren't user install-able applications it should be fine.
Re:Chicago makes Computer Science a core subject on Thursday December 26, 2013 @06:51PM (#6968703)
So having actually seen what Code.org's small hour of code assignment was (my sister in 6th grade had to do it), I can say that it's not as evil as you're making it out to be. For one, they're not taught a specific language -- they're given visual blocks to work with (and the ability to show the corresponding javascript).

I assume that without real instruction you're not going to learn the theory, and yes someone will have to advocate for OSS, but it seems that the goal is more basic -- not to pump kids full of impressions that 'this way is the only way' and 'just type code don't think no think'.

One thing I noticed more from my sister than the hour of code was that thought processes for programming that we take for granted haven't been implanted, and need to be taught. For example, my sister when faced with a path for the zombie to take would start by telling the zombie to move forwards many times then take a left... etc etc. Conditions and loops are introduced gradually and the idea that you would use a loop to do multiple things instead of just manually doing it is part of this. How should kids understand complex CS concepts without this? They need to start somewhere.

Re:Netflix: Non-'A' Players Unworthy of Jobs on Wednesday December 25, 2013 @01:02PM (#6963643)
I think the point of the article is that whether or not someone is an A-player is context dependent. Alice might not be an A-player at TechCo (for some reason e.g. the way they dole out projects) but is an excellent fit at MicroCo, where she is considered an A-player because she can work well in their context.

Whether or not this works in practice is obviously a bit iffy, since finding someone who fits exactly your context is incredibly difficult.

Re:The main trick in machine learning on Monday December 09, 2013 @06:15PM (#6877770)
>stop and think about possible sources of contamination

One great one from my Machine Learning professor was an assignment where we were required to normalize our data to [0,1]. After doing this and then going through the typical cross-validation cycle, he had us try and figure out where we contaminated our validation sets. As it turns out, we all normalized our data before splitting it up, which meant that training data influenced testing data.

It's a simple fix, but if you've done that and gone to run a large convolutional neural network for a week only to find that you made a stupid error like that, it can be pretty painful. (Especially since the bad generalization error might not be obvious until you use it the model in production)