pyduan

18 ms·

121.
by pyduan 13y ago
I don't know the case well enough to comment on it, but my money would be the same explanation: badly coded automatic discovery systems. The DHS has been caring notoriously little about false positives and I wouldn't be surprised
122.
by pyduan 13y ago
This is probably what happened, and not some not very subtle (not to mention completely useless) scheme by the DHS as the title of the article seems to suggest. I cannot speak for what process the DHS or the TSA use, but what is done in the
123.
by pyduan 13y ago
> The flaw in your post is that deterministic does not mean showing everyone the same thing. Though to be fair, a random number generator would be used to determine which users to expose the article too (just not how many of them). Well
124.
by pyduan 13y ago
Probability matching is hardly what I would call a deterministic algorithm. What the authors likely means is that we should use a probabilistic algorithm rather than a random one. Nitpicking aside, I disagree with the usage of the origi
125.
by pyduan 13y ago
From what I understand these would be complementary: the advantage of a probiotic solution would be that you basically get the treatment from time to time then never have to care about periodontal disease (which affects almost half of the U
126.
by pyduan 13y ago
Your argument only holds if your time discounting factor is greater than the rate of deflation. In the case of Bitcoin this is clearly not the case since prices can double overnight, unless you're arguing that you're so adverse to
127.
by pyduan 13y ago
While the figure might seem shocking, keep in mind this is 17% of employees who responded to the VentureBeat survey , not of all employees, and that not only we don't even have a sample size, but on top of that the survey is only &quo
128.
by pyduan 13y ago
44% actually. But I agree that this figures seems unrealistic.
129.
by pyduan 13y ago
I wouldn't be so categorical. Regarding assessing causality, with an appropriate choice of instruments (food prices, tax on specific food categories) one could use instrumental variable regression to disambiguate the relationship. I&#x
130.
by pyduan 13y ago
Thanks for responding. I fully agree with your comment -- as I said, I too think many people are sometimes put off by the apparent complexity of machine learning, and demystifying how it works is a great thing. Unfortunately there's al
131.
by pyduan 13y ago
As someone who works in machine learning, I have mixed feelings about this article. While encouraging people to start learning about ML by demystifying it is a great thing, this article comes off as slightly cocky and dangerous. Programmers
132.
by pyduan 13y ago
I agree disambiguation is a tough problem, but I'm curious to see whether the vector representation learned by word2vec can help in that regard. Have you considered using clustering to attach a few possible cluster memberships to each
133.
by pyduan 13y ago
I don't agree with this argument. To me, it sounds a lot like the one that was used for a long time to justify Chrome showing stored passwords in plaintext without prompting for a password: "the only strong permission boundary for