It isn't just the detection algorithm that is tuned to benefit big corporations. The end to end process is asymmetric with automated issuance of copyright claims which need to be manually disputed which appear to be automatically rejected. Then the appeal process has a three strikes rule against channel owners but no apparent punishment for corporations.
I can understand tuning the detection algorithm but to balance the dispute process needs to take this into account. The fact that it doesn't shows where Youtube stands on the issue.
This is well understood by pain researchers and the article explicitly mentions that they are trying to find solutions that address chronic pain without impacting the pain responses you are referring to.
"The team bred genetically engineered mice from embryos that had HCN2 excised from their DNA. Subsequent experiments showed that these mice did not develop neuropathic pain (the kind that affects the nervous system and is often caused by long-term conditions such as cancer or diabetes). Not only that, the mice with HCN2 cut out were still able to feel acute pain – the necessary, protective jolt that tells us to remove our finger from a drawing pin. “That’s the holy grail,” McNaughton told me"
While not networked to the national police I expect that many commercial security cameras do have facial recognition to identify suspected thieves and VIPs along with tracking movement habits through the store which is shared with advertisers and product manufactures to justify cost for higher profile shelf space.
My understanding is that was a passing fad a few years ago. It was hyped by some startups but commercial users found it less than helpful compared to traditional metrics. I met a grocery store consultant (a real job) at a security conference. He said they used cameras to deal with traffic issues (even load between aisles) and predict cashier/customer ratios. They had abandoned the face/eye-tracking tech. Eye-level shelf space is better. They don't need an algorithm to keep telling them that.
They don't need CV for this. They can use beacons embedded in smart devices to tell which part of the building you're in. Worked with one of these guys.
They need cameras for theft. Tossing a person detecting algorithm on top is basically free (especially if you don't need it to do fancy things like track faces).
"While their model is far from uncovering the full mystery of vision, it is a step in the right direction"
This penultimate paragraph directly contradicts the headline. I know that writers don't have much control over their headlines but this is endlessly frustrating as a reader.
The most interesting tidbit for me was how similar the news breakdown was between the two sources: New York Times and The Guardian. The largest deviation was 3.4% for suicide coverage but almost everything was within 1%.
Perhaps these examples are too similar to get a good distribution but if news organizations are all covering basically the same items perhaps there is an opportunity for differentiation. The google search trends shows that what people are interested in knowing doesn't match what media is interested in telling.
I wish they had chosen a different, or a third source though. I rate NYT & The Guardian (though both of decent quality) as fairly similar in editorial outlook.
Even as the person who wrote the scraping code for the original project, I'm a little suspect myself of the news data because of how similar the two distributions turn out to be.
I think the strong similarity is an artifact of our data collection process, rather than reflecting some very deep truth about the similarity of the two sources. Or of sources in general. My priors are that the distribution should have looked more different, but I just didn't do extra verification at the time.
Meta-analysis is a sound principle but undermined by selection bias. This is why every meta-analysis includes selection criteria for which studies are incorporated into the analysis. Unfortunately publication bias is systemic and biases the results.
The quote referenced is poetic and pithy but I fail to understand how it applies to meta-analysis and selection bias. What is the cow pie in this scenario?
The root problem with using meta-analysis is that it just wasn't designed to work with real-world science. It generally assumes that it's being applied to an unbiased sample of unbiased results. It's now pretty well understood that the published literature is really a biased sample of (oftentimes) biased results. No amount of selection criteria can fix that; the best you can hope for is that they will yield a biased sample of unbiased results.
I'm no expert on health science, I'm just taking potshots from the peanut gallery, but I'd guess it's pretty much always better to ditch the shiny mathematical bauble and its false promise of providing an easy, simple, objective answer to an inherently subtle and complex problem, roll up your sleeves, and get to work on a proper systematic review.
This should only happen for user requested auto-completion. For example typing "g" <tab> "g" <tab> would result in Google in the provided example.
This is very beneficial in code completion or navigating file directories. It is horrible when typing an email (though maybe when searching email contact). Definitely needs to be context aware.
"The girl and boy each have male and female sex chromosomes — in other words, each twin has some cells that carry an XX pair (female) and some that carry an XY pair (male)."
They are both genetically ambiguous and present differently due to hormone generation rates.
Contact binaries are fairly common. Expected to make up about 10-15% of NEOs. My layman's understanding of how Kuiper object contact binaries develop is mutual capture during the early life of the solar system and angular momentum decay until they come into contact.
Can you say more about how angular momentum decays in orbital mechanics? I understand that, because the Earth rotates faster than the moon orbits, our tidal bulge will be ahead of the sublunar point, and will accelerate the moon in its orbit. Are there any other ways to get rid of angular momentum?
Tidal forces are one way - either between the two binary components, or from a close encounter with a third, larger body.
Another is the "YORP Effect". Sunlight falling on an asteroid produces a slight thermal radiation pressure (push). If, due to asymmetries in asteroid shape/albedo, the net radiation pressure force is not aligned with the asteroid's center of mass, it will produce a torquing force which will cause the asteroid to spin faster (or slower) over time. Applying the idea of YORP Effect to binary asteroids yields the "BYORP [Binary YORP] Effect", by which the orbital dynamics of the binary system are modified by this asymmetric radiation pressure over time, in a way that either pushes them together into a contact binary or apart into two unbound asteroids.
It's even hypothesized that some asteroids may be in a binary/contact-binary cycle on long timescales! There are solutions to the above in which the BYORP effect causes a loss of angular momentum in a binary pair, causing them to merge into a contact binary - but the contact binary may settle into a state where the YORP Effect actually causes the newly merged asteroid to spin faster, eventually flinging them apart due to centripetal forces... back into a binary state where the BYORP Effect may again cause them to merge someday.
I can understand tuning the detection algorithm but to balance the dispute process needs to take this into account. The fact that it doesn't shows where Youtube stands on the issue.