Posts by Michael H.F. Wilkinson 4257 publicly visible posts • joined 24 Apr 2007
by the many many people who have depended on you for their daily or weekly dose of being annoyed about something
and of course by those who have enjoyed your articles (the intersection of these groups is non-empty, I gather from the above).
I'll buy you a pint if ever we should meet
Just thinking that will immediately draw the attention of the God of Hubris, who will instantly think of ways to make your day (or even decade) thoroughly miserable.
Never even think of saying that. As Lu Tse says: "There is no such thing as too many backups"
Wouldn't it be the height of irony if some couple kills themselves (but hopefully no innocent bystanders) whilst trying to procreate in a driving car?
The stupidity and recklessness of such an act is beyond belief. These people should be locked away in our planet's stone age and told to evolve into more responsible beings
Doffs hat at the late, great Douglas Adams.
I walked or cycled with them, this being the Netherlands, and generally shared this duty equally with the missus. Who got to take them to school on a particular day depended largely on our teaching schedules. Taking kids to school was quite an enjoyable "chore", certainly beating changing diapers.
Of course, as adolescence appeared on the distant horizon, the kids wouldn't be seen dead with their dad or mum bringing them to school. Their street cred would be out of the window! I mean, parents are SO embarrassing!
I have seen people plug a VGA cable in the wrong way round. Must have taken quite some force. They even screwed the thing in place (probably just to be sure of a good connection). I spent quite a while figuring out why the beamer didn't work, as putting the connector in the wrong way round was not on my list of plausible causes. I only found out when I tried to unplug the cable to replace it, suspecting it might be faulty. As it turned out, it was faulty by then, as quite a few pins had been bent or broken. Had to throw it out (after berating the user).
I still have an 8" floppy disk with a capacity of 128 kB (yes kB), that contains CP/M 2.0. All of it. Of course a modern OS is way more complex, but my Linux install takes up only a fraction of that space, and that is including all sorts of software packages like editors, compilers, etc.
More like a stampede than a race, actually. A race suggests a clear finish line, this sounds more like a headlong rush to a waterhole, which might be a mirage, or in which crocodiles might be lurking. It might of course also be a pristine spring of crystal-clear water, but I do not doubt some wildebeest will wee in it before most get there
of loads of robots faced with streams of lectures given by the driest, most boring physics professors, one by one turning into machines so depressed Marvin would seem cheerful company
Yes, yes, I know, that image is mental in more than one way!
I'd better get me coat
I'll see your beer, and raise you a malt whisky.* Excellent work by Katie Bouman and everyone else who contributed. It is great to see that in our computer science department the number of female faculty members is steadily rising (still quite a way to go), and in our group we already have as many female as male PhD students.
*By all means feel free to have a bourbon or other dram of your choice, no need to be snobbish about whisk(e)y, after all, as Gordon Muir put it at one of his whisky tasting session: "It's just brown stuff that tastes good!"
The issue is that although deep neural networks are based on logic, the myriad weights used to combine the complex input data make it near impossible to understand that logic. The weights given to each input item are not chosen, they are inferred automatically from the data. At the end of the training period you are left with a highly complex network in which hundreds of thousands of computations are performed and a decision is spat out. Understanding how this decision is made is near impossible for deep networks.
More subtly, the machine will learn biases from its input data. If African Americans are poorer, and are fired from their job more often due to existing racism on average, then the machine trained to decide whether or not to give a loan might well learn to discriminate against African Americans when deciding whether or not to give a loan if ethnic background is part of the training data, perpetuating an existing bias in society. This bias might not be obvious in the weights used in the final machine at all.
I have written a few review papers and chapters in books, and even edited a volume, and although this is a lot of work, the main issue is not providing a summary of existing work. That is comparatively easy. After all, even first-year students manage that quite well. Providing insight and commentaries is another matter entirely. In the volume I edited I could easily have written summaries of loads of papers for the chapters I chose not to write myself, even before the advent of Google Scholar. What I needed was experts in those fields not in my core field of expertise to provide insights, contrasting the outcomes and opinions found in different papers. While I am not suggesting AI will never be able to do that, it cannot do so now, or at least, not in a meaningful way. Likewise, well-written papers have a kind of coherent storyline, which makes them readable and ideally even memorable. Until AI knows how to keep the reader entertained (even in a scientific book), I do not fear their competition.
My only worry is that students will start using them as automatic essay-writing tools, and I have to read loads of terminally dull papers that actually do escape the vigil of the plagiarism scanner, but are still not their own work.
Smallpox is (was) a virus that shows very little mutation, and therefore vaccination was particularly effective, because all strains expressed essentially the same antigens. At the other end of the spectrum, the influenza virus and worse still HIV show huge variations in antigen expression, because they mutate (or form chimeric viruses in the case of influenza) rapidly, thus evading the immune system. That is why vaccination isn't particularly effective in those cases. Measles is somewhere in between, I would guess: Vaccines work well enough, but don't protect fully.
Words are clearly quantum objects, being a superposition of multiple meanings, and the exact meaning is only known when the waveform collapses when a human observes.
I feel an idea for a paper on quantum linguistics coming up. That would be for Annals of Improbable Research, of course. Should fit nicely with my paper on quantum homeopathy.
I'd better get me coat. It is nearly beer o'clock, after all
Precisely. I take quite a few snaps with my smartphone, but for wildlife, landscape, and sports I use my DSLR (Canon EOS 80D), and for astrophotography either an astro-modded DSLR (ageing but capable 550D), or a variety of ZWOptical planetary cameras (like webcams on steroids). I generally don't like processing my photos much, except for astrophotography, where it is absolutely necessary.
that "sEMG" is distressingly close to "smeg!", an expletive which might well be heard more often if this gets implemented in my house (I will probably end up giving it a reprogramming it will never forget using a big axe)
I'll be going. The one with the Red Dwarf DVDs in the pocket please
What they are really saying is that they need MANY more videos of white supremacist massacres shot from the point of view of the killer to train their "AI" (artificial idiot?) to detect them. Well isn't that just dandy. Any volunteers to be the victims?
</sarcasm>
I touch upon cryptography in my course "Introduction to Computing Science", and might well put up links to this code for students to have a play around with it.
Donning my tinfoil hat: this might be a decoy, without any back doors, to lead people on a wild goose chase through the code on GitHub, while the REAL back doors are quietly inserted through other means.
Interesting project though it was, it was let down somewhat by being slower at text entry than making punched cards with a nail gun.
I once saw somebody working (or at least trying to) work with that CorelDraw Java monstrosity, and that warned me to steer very clear. Earlier CorelDraw stuff (I still have the installation CD-ROMs of one version somewhere) was pretty good. Later it all went to bits.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
