Except for some of us who come from the time you had to get permission to get an ID on the Arpanet. In that case, the WHO is not saying mathematically nice things about us. :-(
Ehh, nature, she always drags some series into the mix - the mortality rate is a function of distribution across age and other health factors - it doesn't seem to drive the immune system bonkers, so WHO is calling out higher risks to older people, esp those with additional factors such as cardiac or pulmonary impairment. Probably due to the fact this is fundamentally a respiratory issue. Abstractly, more data always helps in spite of the cost to get it, but for us in the USA, the last official action has been to delete the tests_administered count from daily CDC reporting. So that makes any projection more guesswork than not.
Given my first terminal was an ASR-33, this is of interest to me - yes I have insurance, and yes I have friends who got staggering bills because the anesthesiologist wasn't in-network (not to bash on anesthesiologists, I have counted some as good friends and they're the greatest geeks in medicine). - will this keep in-network hospitals using in-network providers for services ?
Common LISP could certainly generate assembler for Arduino and other MCUs, but there's no way you're fitting that fat old beast (whom I do still love) on one . :-)
If this was reddit, I'd award gold for that comment. Only thing I can add is that one should only value paid experience, i.e. the person working a day job doing a banking app and a personal job of adding high quality code to a complex OSS project, well hell, they only know a banking app
The 'problem' with modern science is that a large finite amount of stuff has been discovered, and new discoveries are most often built on that stuff. I splash around in the backwaters of machine learning, however I have to pick my targets carefully and maintain a very tight focus - the defined knowledge base has already grown to more than anyone can ingest in detail. Given the amount of knowledge that has to be absorbed to achieve mastery in a scientific discipline, and the fact that knowledge base keeps growing, it takes longer and longer to lay a foundation for just understanding what the hell people are talking about. Now, there are opportunities, for example the use of genetic algorithms as part of ML solutions, because for any 'real' researcher, GA's aren't the cool kid in the opinions of reviewers or funding agents. It's easier when you are the PI and the funding authority, even if the budget is quite a bit smaller
Don't get me wrong, I like your cites - but at the end of the day, math is math. Either it works or it doesn't. If it works, apply it where you can! I've got a sneaking suspicion a PCA based analyzer for lidar generated point clouds is going to be a lot faster.......
Heh - had to go look up storage protect, aka 'red light errors' . Kept getting in trouble in the beginning of my career when I didn't realize snooping about lit up lights on the front panel . :-)
This has pretty broad applicability across a wide range of algorithms. The common failure mode when the machine fails to recognize the otherwise normal real face and body indicates that the whole face/skeleton relationship has fallen apart. Defeating this is interesting, as we have enough trouble just trying to recognize faces. To add to this "yes these are faces too but they are not faces too" is probably going to drive some researchers to drink. At the end of the day, this is a common flaw in a lot of deep learning systems, they're very brittle.
Not as a single tool, but if I'm using numerical libraries I need to be accurate, and not have another unpleasant experience with numerical methods. That said, I can usually get by with the prebuilt libs
Much of the last 2 decades of my career has been as much principal work as I can get, and everything I know says you smacked this ball out of the park.
First, if the title is real, then the term Principal carries serious weight, and hews as closely as possible to the best definitions of Principal Investigator (yah, academia has been busy making a mess of that)
More often than not, a good PI knows that all problems don't need the most powerful solution, some just need to go away. If a claimed principal is always applying the most complex solution everywhere, then they're bad at their job.
A good 1/3rd of my benefit is not going 'zomg - a hard problem' to my co-workers, its the exact opposite. And even when it is a hard problem, at least half the time then it's 'Don't worry - here's what Djikstra/whomever did to solve it'
I love advanced algorithms, and relish the chance to actually go try and make ones - that said, my main value is not that, its that some horrible new problem erupts, my job is to say, 'no, calm down. That's an old problem in a new suit'
That said, switch isomorphic to homomorphic - I'll give the seniors credit, they usually tag the isomorphic cases . :-)
Got 9/10 but it was a challenge - between the way real Trump mangles English and fake Trump strains to flagrantly fail the Turing test (h/t @JanelleCShane) it was a struggle - nice job, train it more, and hook it up to it's imaging equivalent.....then let them all sort that mess out . :-D
I just hate n-grams is the short answer - tear apart the input with link, use arc and modifier info as input to the statistical processes - so far, too early to say it works better, but sure as hell its more interesting . :-) . Sadly I have to keep my instances running all the time, they eat all of the public Reuters news feeds - side note - additional info can be obtained by taking the top N parse candidates for a given sentence and collapsing them into a graph - I have a pretty good signal out of that for 'the parser has gone insane'
Gotcha - yah, I can containerize, it's not like I'm screwing with drivers or whatnot on the AMI - I'll keep an eye on you, not ready for GPU yet, as I can't even rationally define the vectors I'm extracting from the Link stuff. Best of luck to you, lot of people piling into that battleground, and the Dunning-Kreuger effect is rampant. :-)
Side question, if you're willing to entertain it. Tired me tells a notebook to checkpoint, wanders off to bed, comes back next day and wonders why its taking an aeon to open the notebook..... oh yeah, damn, I've got gigs of images and panda crap in it . -- do you wrangle this problem, i.e. please don't save your notebooks as gihugic files representing a mental state you can't possibly remember ?
OK, did a quick scan on free and paid tiers and wondering about ability to bring in specialized python libraries- in my case, https://github.com/opencog/link-grammar which almost always requires hand building and migrating, and some custom packages on top of my own even more horrible C++ . - I have to run a jupyterhub ami on an ec2 instance right now
Ran this on the Ithaca Intersystems Z-8000 based rig in the early/mid '80s - it was remarkably command compatible with V7 for the basics, even built a MicroAngelo based 3D graphics system on it.....the biggest problem the box had wasn't Coherent, it was the fact the socketed logic on the custom MMU S100 card kept trying to walk out of the sockets because of thermal issues
All the experience I had was theorem proving for crypto apps - that said, I think there is only the appearance of reversibility at the symbolic level, underneath its the same old mess - I vaguely recall that at full tilt, the room with all the Suns would get quite toasty . :-)
Grrrr. This could have been interesting if Umberto Ecco was arguing semiosis, instead of this poor exercise in semantics. As a concrete example, consider "The AI therapy app is not a mind. It cannot emote. It cannot feel the warmth behind its praise or the sense of urgency behind its criticism. It is passive." Yah, no. I can also do sentiment analysis on the users comments, and the behavior of the system can be driven by a goal to change that sentiment analysis, so I've got me an active participant in a feedback loop - yes, there is no their there, its just a numbers game - but this article is still complete twaddle