I’ve read it had died off in the past and recovered afterwards. This had happened 5 times in the past 30000 years, From that it would logically follow that recovery is possible, just not for these particular specimens. Additionally, it’s hard to tell if the current bleaching is as fatal as everyone here seems to assume. FWIW scientists seem to think the situation is grave, but not as grave and hopeless as the mass media portrays it: https://www.snopes.com/fact-check/scientists-pronounce-great.... Of course the more balanced view doesn’t drive the clicks quite as well.
I'm not sure any problems would be fixed. The way things are now, politicians in the house/senate have to always worry about re-election, and at least create visibility of catering to their electoral base in the run-up to that. If their second term did not have the possibility of re-election, they'd just do what the lobbyists want them to 100% of the time (as opposed to 90%) in exchange for a job after their second term ends.
I don’t provide citations for such obvious things. If you’re interested, you can find the origin story as written by Guido himself in Python FAQ using any of the available search engines.
Even in its present state, Jupyter is strictly better than plain REPL, especially if you need visualization. It's like Python in general: it was not designed to write 10-100KLOC programs. Python was designed for scripting. Jupyter was designed for interactive data exploration, and to create a record of results which you can view without re-executing the cells. Is it error prone? Yes. But then so is REPL.
It was also a "fact", in the times of Galileo, that Earth is in the center of the universe and the Sun rotates around it. But that might be too ancient, so how about another: before the 80's there was "scientific consensus" that ulcers are caused by all sorts of things other than bacteria. Dude who discovered they were, in fact, caused by bacteria, was nearly kicked out of his field.
History is rife with "facts" that weren't actually, you know, real facts at the time, even though there was "scientific consensus".
Note that I'm not an anti-vaxxer, and I'm not a climate change denier. I also don't imbue "scientific consensus" with magical properties, seeing how often it was dead wrong.
That’s an incredibly naive and narrow view. In most of life’s circumstances “facts” can’t even fully be established, and are better referred to as “strong beliefs”. And the definition of “hate” depends on perspective as well as varies over time. Moreover, there are circumstances where hate is definitely not “wrong”.
There’s no concept of “reboot” with MCUs, since there’s usually no OS. Likewise there’s usually no concept of segfault, because segfault requires memory protection which is something most MCUs don’t use.
Proud of what, though? It's basically a 2-piece hololens that doesn't have any software. This stuff won't take off until field of view is fully immersive. In Magic Leap (and hololens) it's not.
Sure. I’m just saying worker pay issue is largely orthogonal to stock buybacks or tax cuts. It’s not like, all things being equal, the revenue would be spent on workers anyway.
It’s as though Intel is being sabotaged from inside. Delays with 10nm, suspicious exit of Krzanich, harebrained unenforceable licensing schemes. My next machine will use Threadripper and this kind of shit just seals the deal.
And therein lies a valuable lesson: don't be employee #1 if you don't know the founders personally, or can otherwise make sure they have a good track record.
Another cool thing one could build on the moon is a rail cannon to bombard Earth. Moon's gravity well makes it relatively easy to hurl stuff from there to here, and the stuff _accelerates_ as it approaches Earth, assisted by earth gravity. The scenario is explored in detail in Heinlein's "The Moon is a Harsh Mistress".
What makes YUYV more efficient is downsampling. You represent 2 pixels with 4 bytes instead of 6 due to the loss of spatial color information. If there's no downsampling, and there exists a transform between RGB and a given color space, your network will simply learn slightly differently. Remember that after the very first layer there's no notion of color space anymore.
Pretty much. And also, models almost never operate on full-size image, and if you e.g. pixel bin a bayer image you will get a de-mosaiced image, so that's not very interesting. What is interesting is, it'd be pretty cool to train on linear HDR images for instance (i.e. direct 14-stop sensor output without contrast curves and color correction), or on visible spectrum + IR + UV, but I'm not aware of any large public datasets like this. The reason why this is interesting is because a lot of vision models suffer from the lack of robustness. Their performance on the train/validation datasets might be OK, but connect them to a real live video feed and they turn to shit. To some extent this can be addressed by data augmentation during training, but you have to understand that this is an imperfect process: you're operating on the data that's already pre-cooked, so to speak, for _human_ consumption, with color correction, gamma curves, etc already applied, and there's not a whole lot you can do to make things more robust to different lighting conditions, noise in the feed when camera bumps the gain in its ADCs, color drift due to e.g. sodium lamps, over/under exposure, etc. Unsurprisingly, this is especially noticeable in efficient, quantized models, because they don't really have much leeway built in to work around the deficiencies in the input data.
For DL the colorspace is irrelevant as long as it's consistent between training and inference. I.e. if you train your model in RGB, it better be receiving RGB and not something else from the video feed or input images. Interestingly, models can also be trained in YUV (including its downsampled variants like YUYV), which can make them a tiny bit more efficient, both because there's less data, and because most camera chips can output YUV raw data directly. The net will learn whatever transform it needs to learn to make sense of the data.