Don't be surprised if those predictions are heavily biased against minorities and poor people. Do you care if they do?
It's a similar problem to using ML to give people credit scores.
If the training data includes a lot of minorities and poor people breaking laws / delinquent payments, then your ML will simply key on race/economic status as a predictor.
So you've built a system that simply targets those groups.
But you might object and say that this race/economic status targeting gives the highest accuracy! It was only learned in the training data, after all. You can make a great classifier that is extremely unfair.
So you have to realize there is a conflict here between accuracy and fairness. This means there is a conflict between observational data (training), and using that data to produce decisions/outcomes.
If you make decisions/outcomes that reinforce the training data, you do not give racial groups/low economic status people a chance to improve their lives.
Racism is morally wrong but not mathematically wrong. P(criminal|black) > P(criminal), but if you observe that someone has black skin and treat them poorly because of it, you've done a bad thing. It doesn't matter that you were just following Bayesian reasoning because you're still hurting someone on the basis of something they can't control.
Lady Justice doesn't wear a blindfold as a fashion accessory. Discarding information is a key factor in nearly every established system of justice / morality. Refusing to do so (i.e. "just" running a ML algorithm) places you directly at odds with society's hard-earned best practices.
Take a look at crimereports.com. You might get lucky and find a good source on a per city or county basis, it's too fragmented overall too try this. Different countries might have different documentation standards and publishing guidelines for this kinda of data, might be worth a shot to look.
The first sentence of the article has more detail but is still false. Then the second sentence of the article contradicts the first sentence, adding the phrase "of the DC city police". Shall we believe that version?
The hardware is nice. The mini PCIe appears to be temperamental depending on which board you've used and there's little to no support from the vendor - they seem to have released everything, then disappeared.
All of them AFAIK. Even the Zero W can (finally) be had with relative ease, I just ordered one from SparkFun and Adafruit also has 'em in stock. Been wanting to make one into a security camera but they have been hard to get until just recently.
I think it depends on what is meant by "readily available". It's hard to do something like buy more than a couple Zero-W's at $10 apiece, unless things have changed very recently. In the past, I've seen people wanting to use a few dozen in a project having trouble finding that many for uninflated prices.
I can't see why they would though. I don't think the RPF has the resources to sell so many units at a loss. I suspect it's probably just barely breaking even on both. The Wifi chipset the Zero W uses costs exactly $5 in quantity according to Digikey; the board layout is slightly different but otherwise the bill of materials is the same so that's the main cost difference. They're probably making some profits off of the accessories like the case or the camera (especially the camera, which is pretty expensive and I expect fairly popular).
Very true. It's definitely hard to get more than a couple still. But they aren't perpetually sold out anymore, at least. That is definitely an improvement in the availability that has occurred in the past couple months, so maybe their supply is finally settling down.
"The Marvell® ARMADA® 3700 SoC family incorporates rich high-speed I/Os including USB 3.0, SATA
3.0, Gigabit Ethernet (1 GbE) and 2.5 GbE (NBASE-T) off ering two confi gurations: 88F3720 for
dual-core and 88F3710 for single-core, both confi gurations have industrial temperature grade
support. In addition, these devices feature a wide set of security and data acceleration engines
suitable for innovative networking, storage, and computing applications. The ARMADA 3700
supports advanced power management technologies for switching the CPU cores, as well as
per-core dynamic voltage and frequency scaling. This solution off ers a signifi cant reduction in power
consumption under diff erent workloads and delivers an optimal Performance-per-Watt in the embedded markets."
