Yeah, that is probably what bothers me the most about this. I write C code almost everyday. I can't remember the last time I forgot a "break", but I do use fallthrough when it simplifies the logic. It also breaks even the simple case of having multiple labels for the same code.
Also, the for-loop overrides are ugly and pointless, IMO.
This would sound much less impressive if they just said 500 tons. Also, why the dead weight? 500 tons is easily doable with a (smallish) hydraulic press.
The actual title of the article is "How Do You Lift A Million Pounds Of Stainless Steel?" I feel like they didn't answer that question either. It turns out that the weights are in fact lifted by hydraulics.
It's a question of accuracy not maximum force. At some point you want to calibrate the things you use to calibrate the hydraulic presses. http://www.nist.gov/calibrations/index.cfm
Right. Something the parent article completely fails to mention. This is NIST's force standard for large forces, the place you send load cells to be calibrated.[1] If they have to use the whole million pound stack, the fee is $16,489.
>Also, why the dead weight? 500 tons is easily doable with a (smallish) hydraulic press.
Sure, but how do you know the hydraulic press is applying 500 tons? You need to have a universal standard for length (to determine the area of your piston) and for force (which doesn't exist), and consider the friction in your hydraulic system. We can generate a universal standard for the kilogram, then duplicate that log(500) times, and create a very well controlled mass that we can hang from stuff. The level of accuracy that is needed would surprise you.
Exactly. This isn't an IoT problem, it's a cloud-based systems problem. "Cloud" actually just means "someone else's computer", and that somebody will turn off their computer someday.
This is cool. The documentation is very entertaining, although not something you'd show to your boss. Looks like it implements a lot more than just neural networks.
Shameless plug: For a minimal neural network implementation in ANSI C, check out: https://github.com/codeplea/genann Sometimes lack of features is a feature.
I'm convinced to give it a try.