And the remainder of that quote: > Wilmore said he had no reason to doubt Musk’s claims about an offer to bring them home earlier, though Wilmore said he was not privy to such an offer. Which pretty much amounts to "I…
I think you are reading too much into my response. The list of events isn't an attempt to justify anything. It is mainly to establish that what happens in 1948 isn't coming out of nowhere, but rather has been brewing…
Tbh starting the history doesn't at 1948 doesn't really help much. Around 1948 is when the Jewish and Palestinian narratives are already in full swing and have some of the strongest evidence for each side. On the Jewish…
I disagree. In most subjects, recapitulating the historical development of a thing helps motivate modern developments. Eg 1. Start with bag of words. Represent words as all zero except one index that is not zero. Then a…
Yes! And that’s an oversight on my part — word embeddings are interesting but I usually deal with documents when doing nlp work and only deal with word embeddings when thinking about how to combine them into a document…
Eh, I disagree. When I began working in ML everything was about word2vec and glove and the state of the art for embedding documents was adding together all the word embeddings and it made no sense to me but it worked.…
Heh -- my explanation isn't the clearest I realize, but yes, it is BoW. Eg fix your vocab of 50k words (or whatever) and enumerate it. Then to make an embedding for some piece of text 1. initialize an all zero vector of…
> I think the key property of embeddings is that the dimensions each individually mean/measure something, and therefore the dot product of two embeddings (similarity of direction of the vectors) is a meaningful…
Yes! And the follow up that cosine similarity (for BoW) is a super simple similarity metric based on counting up the number of words the two vectors have in common.
You're right that what I described isn't what people commonly think about as embeddings (given we are more advanced now the above description), but broadly an embedding is anything (in nlp at least) that maps text into…
One straightforward way to get started is to understand embedding without any AI/deep learning magic. Just pick a vocabulary of words (say, some 50k words), pick a unique index between 0 and 49,999 for each of the…
Depends on what "AI" means here. There is a spectrum of "we have a bunch of data in a database and some folks hand tuning queries" to "we built a deep learning network to predict XYZ." In the middle of that spectrum lie…
> Which required me to learn "direct response marketing and copywriting". How’d you go about getting good at these things?
If by "conventional neural network" you mean a stack off fully connected layers, then yes, in theory one of those could learn a similar mechanism because of the universal approximation theorem. However, training one…
If anything, it seems like this was an effective means to introduce a lot of people to possible liabilities they have under GDPR/CCPA (or why they are not applicable to them).
This is partly because the first course in real analysis (and usually the first upper division linear algebra course) is the first proof oriented course most students face. The material isn't necessarily the hardest,…
The linked graph shows Sweden at 7x norway's rate, 5x finland's rate and 3x denmarks rate. Am I reading that right? How are you concluding that Sweden's strategy was clearly superior?
That's a fair criticism. I always rewrite & revise the important notes as I go forward, so I never end up looking back. I can imagine that it is super frustrating to find things that aren't near the end or on a known…
> The basic problem with the Remarkable is that it's useless without the $8/month "cloud" plan because that's the only plan that enables its primary feature: handwriting recognition. Are you saying the key feature for…
> All those problems were present in 1957 flu epidemic, the 1968/69 epidemic the 1977/79 flu epidemic Those were all influenza A sub-types viruses which we've had experience with. The closest we've come to COVID is…
The most important thing here is that we're still in the middle of the COVID pandemic and don't know quite a bit about it whereas the Spanish flu has ended. We don't, for example, know about the longterm effects of…
We definitely noticed the 2009-2010 swine flu pandemic, it just wasn't as deadly as COVID. Most pandemics in past memory have been minor in nature. We likely were aware of them, but didn't feel their impact because they…
For a double blind randomized trial, no one can know who's in what group. If an individual isn't willing to get the vaccine then they'd be excluded by default.
The author is misunderstanding why the "illiberal left" is discussed nowadays and tries to debunk a number of examples but fundamentally fails. The most telling is an attempt to debunk a claim by the Economist --…
What really clicked confidence intervals for me is the explanation that the '90%' in a 90% confidence interval indicates that 90% of the 90% confidence intervals will contain their true parameter. There is no reason to…
And the remainder of that quote: > Wilmore said he had no reason to doubt Musk’s claims about an offer to bring them home earlier, though Wilmore said he was not privy to such an offer. Which pretty much amounts to "I…
I think you are reading too much into my response. The list of events isn't an attempt to justify anything. It is mainly to establish that what happens in 1948 isn't coming out of nowhere, but rather has been brewing…
Tbh starting the history doesn't at 1948 doesn't really help much. Around 1948 is when the Jewish and Palestinian narratives are already in full swing and have some of the strongest evidence for each side. On the Jewish…
I disagree. In most subjects, recapitulating the historical development of a thing helps motivate modern developments. Eg 1. Start with bag of words. Represent words as all zero except one index that is not zero. Then a…
Yes! And that’s an oversight on my part — word embeddings are interesting but I usually deal with documents when doing nlp work and only deal with word embeddings when thinking about how to combine them into a document…
Eh, I disagree. When I began working in ML everything was about word2vec and glove and the state of the art for embedding documents was adding together all the word embeddings and it made no sense to me but it worked.…
Heh -- my explanation isn't the clearest I realize, but yes, it is BoW. Eg fix your vocab of 50k words (or whatever) and enumerate it. Then to make an embedding for some piece of text 1. initialize an all zero vector of…
> I think the key property of embeddings is that the dimensions each individually mean/measure something, and therefore the dot product of two embeddings (similarity of direction of the vectors) is a meaningful…
Yes! And the follow up that cosine similarity (for BoW) is a super simple similarity metric based on counting up the number of words the two vectors have in common.
You're right that what I described isn't what people commonly think about as embeddings (given we are more advanced now the above description), but broadly an embedding is anything (in nlp at least) that maps text into…
One straightforward way to get started is to understand embedding without any AI/deep learning magic. Just pick a vocabulary of words (say, some 50k words), pick a unique index between 0 and 49,999 for each of the…
Depends on what "AI" means here. There is a spectrum of "we have a bunch of data in a database and some folks hand tuning queries" to "we built a deep learning network to predict XYZ." In the middle of that spectrum lie…
> Which required me to learn "direct response marketing and copywriting". How’d you go about getting good at these things?
If by "conventional neural network" you mean a stack off fully connected layers, then yes, in theory one of those could learn a similar mechanism because of the universal approximation theorem. However, training one…
If anything, it seems like this was an effective means to introduce a lot of people to possible liabilities they have under GDPR/CCPA (or why they are not applicable to them).
This is partly because the first course in real analysis (and usually the first upper division linear algebra course) is the first proof oriented course most students face. The material isn't necessarily the hardest,…
The linked graph shows Sweden at 7x norway's rate, 5x finland's rate and 3x denmarks rate. Am I reading that right? How are you concluding that Sweden's strategy was clearly superior?
That's a fair criticism. I always rewrite & revise the important notes as I go forward, so I never end up looking back. I can imagine that it is super frustrating to find things that aren't near the end or on a known…
> The basic problem with the Remarkable is that it's useless without the $8/month "cloud" plan because that's the only plan that enables its primary feature: handwriting recognition. Are you saying the key feature for…
> All those problems were present in 1957 flu epidemic, the 1968/69 epidemic the 1977/79 flu epidemic Those were all influenza A sub-types viruses which we've had experience with. The closest we've come to COVID is…
The most important thing here is that we're still in the middle of the COVID pandemic and don't know quite a bit about it whereas the Spanish flu has ended. We don't, for example, know about the longterm effects of…
We definitely noticed the 2009-2010 swine flu pandemic, it just wasn't as deadly as COVID. Most pandemics in past memory have been minor in nature. We likely were aware of them, but didn't feel their impact because they…
For a double blind randomized trial, no one can know who's in what group. If an individual isn't willing to get the vaccine then they'd be excluded by default.
The author is misunderstanding why the "illiberal left" is discussed nowadays and tries to debunk a number of examples but fundamentally fails. The most telling is an attempt to debunk a claim by the Economist --…
What really clicked confidence intervals for me is the explanation that the '90%' in a 90% confidence interval indicates that 90% of the 90% confidence intervals will contain their true parameter. There is no reason to…