Slashdot Mirror
New Crime-Predicting Algorithm Borrows From Apollo Space Mission Tech (digitaltrends.com)
Researchers from Georgia Tech and the UK's University of Surrey have developed a new predictive policing algorithm that aims to better manage police resources and gain an upper hand in the war on crime. It reportedly uses technology that's been previously used in weather forecasting and the Apollo space missions. Digital Trends reports: The new algorithm built on previous work carried out by researchers from the University of California and police forces in both the U.S. and U.K. Their 2015 research showed how a predictive policing algorithm could accurately predict between 1.4 and 2.2 times more urban crime than specialist crime analysts. By making recommendations about where to patrol, the algorithm led to a 7.4 percent reduction in crime. However, while effective, this approach has also been criticized due to concerns about possible racial profiling and the underreporting of crime. The new algorithm has so far been demonstrated on a data set of more than 1,000 violent gang crimes in Los Angeles carried out between 1999 and 2002. Early conclusions suggest that the upgraded predictive tool could prove superior for coping with the constantly fluctuating world of real-time crime prediction. The researchers published their paper in the journal Computational Statistics & Data Analysis.
n/t
Here's your AI:
Most crimes are caused by corrupt cops or Republicans who run insurance companies, telephone companies, or branded hotels.
Go arrest them and you'll save the rest of us a lot of time and money, and not have to claim 1960s software has anything to do with AI, crime prediction, a Tom Cruise movie (Minority Report), or the price of tea in China (staying stable, but higher for us being saddled with tariffs because of orange-head.)
It's not like you can PREDICT crime. All you can do is statistically chart where crime occurs. In a Constitutional democracy where one is innocent until proven guilty, and can't be detained and search without 4th Amendment provisions, there's nothing to be done UNTIL the crime is COMMITTED.
E
To publicly put your name on something like this.
... war on dissidents who challenge corporate authority.
They reduced patrolling crime ridden areas by 7.4%?
Traitor in the open couldn't even lie about it on TV. What a joke Trump is. Hang the bitch.
This goes beyond unconscious bias, to full on patriarchal racism hiding behind the computer.
Sadm
"Crime" is fundamentally a people problem. Meaning that if you have a lot of crime you have a dysfunctional society. You can use ever more technology to go even full "pre-crime" on people, but a society that does that essentially commits suicide. So don't put your faith in the numbers. Go out and work on societal cohesion instead.
But what are the opportunity costs of people having to bear the burden of living under a predictive model? Until these can be outlined in totality can one truly make a determination whether or not to implement such measures?
What is the cost of a person complying with new regulations. Who is going to pay these costs, and how do these costs show up on the balance sheet of the proposal?
This is a novel application the Kalman filter ("technology that's been previously used in weather forecasting and the Apollo space missions") which I would expect most electrical engineers to have experience with if they've ever done any sensor applications (this is the second step to processing sensor data if simple averaging doesn't work).
What TFA and, it's associated paywalled, journal article shows that regardless of what is the situation, chances are there is a mathematical approach to dealing with it. The problem is finding somebody who can look at the problem and present it in a way that different mathematical approaches can be used.
This doesn't mean that the Kalman filter is the *best* approach for this problem, just that it is a novel way of looking at the issue of predicting crime.
Mimetics Inc. Twitter
> there's nothing to be done UNTIL the crime is COMMITTED.
If they could predict which block was likely to have a spike in crime 3 or 4 days in advance, that would be enough time to finish their donuts an drive a patrol car over there, potentially preventing the crime. If the presence of the patrol car didn't determine the criminal, the cops could actually catch a bad guy in the act for once, rather than taking a report the next day. I know, that's funny. They'll still just be taking reports after the fact.
Autocorrect / predictive text turned "deter the criminal" to "determine the criminal".
It also occurs to me my post was quite negative, and not without good reason, but I should also acknowledge I've dealt with a couple of very good cops. One in particular in Bryan, Texas went above and beyond when my wife was in crisis. On the other hand, 20 miles away in Caldwell, Texas, a different cop wasn't good at all. He shouldn't be a cop.
the article has no information about how the crime predicting algorithm borrows anything from the Apollo space missions. Article is pretty light on substantive information period. maybe someone can pay to read the paper
How is it racial profiling? Who's putting race into the prediction models?
It's violent gang related crime profiling.
I rarely see bullshit on this level. using the Apollo name to advance a police state agenda.
wait, isn't that what happened in minority report?
Never antropomorphize computers, they do not like that
the ACLU would shut anything that predicted crime down claiming it was racist or discriminated against the poor.
Based on the article, this research is intended to influence how public police services operate. It is good to have a *public* discussion on how these public services and policies could be improved. However, this research isn't really made public. The steep cost Elsevier puts on access to the research has a chilling effect on how others can contribute meaningfully to this discussion.
The article admits: "However, while effective, this approach has also been criticized due to concerns about possible racial profiling and the underreporting of crime."
The response is that a Kalman filter is used. But the specifics of how it is used is hidden behind the Elsevier paywall. Anyone that is familiar with the Kalman filter knows it is a highly complex system which can be used incorrectly to produce what initially looks like good results but do not hold up well over time. By publishing through Elsevier, the researchers are able to limit how many people will be able to discuss what they might have gotten wrong.
Put more plainly, this article reads like a script to an episode of the TV show Numb3rs. The details are just as vague on how math can be thrown at solving crime. When this is done for entertainment purposes, it is understandable the details given are extremely light. However, when it is three researchers attempting to change the location of public resources, this level of vagueness is just not acceptable.
There really is no reason for a "war on crime", except as a pretext to fight freedom and to make some people even richer.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
I know that it's the silly season - but Apollo - really it's been 50 years.
I guess that it's a nice reminder that that's what that generation of engineers built compared to increasingly intrusive ways of tracking.
Academic researchers have to publish, else lose funding, which puts them in a difficult position if the more prestigious journals they need to publish in are not open. Perhaps a subsequent paper can be published in a more open way, and they can describe things to some extent outside journals.
There is a possibility that it will be spun out into a company and become more closed, and it is attractive for UK academics to do so, as there is no such thing as tenure, as even if you are not on a short term contract if you don't attract research funding you can be let go, so an eye for a business opportunity is useful. Whether that itself is an issue depends on how the use of the research is handled.
N Korea stopping missile testing and nuke testing.
EU dropping tariffs on US goods.
Historic low unemployment for blacks and latinos.
Massive tax cuts helping middle class.
No longer punishing blue collar workers with extra tax because they can't afford health insurance.
Stopping illegal criminals from killing US citizens and not being punished (like Kate Stynly)
Showing the corruption of the Federal government under Obama on a scale we have never seen before.
Yea, complete idiot Trump must be. That OR you are still a complete moron. Yea, that's more likely, you are a moron still.
C'mon folks. We're not talking Future Crimes stuff here. They plugged in data, had statistical models run over it, and did some testing to see what happened.
If there's a certain area where there is more reported crime then increasing police presence will help in reducing crime in that area. Of course there will be secondary effects such as underreported crime and criminals moving to other areas that aren't as heavily patrolled.
The first effect - underreported crime - is something that can't be addressed since it relies on victims. That's going to happen regardless. It will skew the models used to predict where to send more patrols but in the long run it will even out.
The second effect - crime moving elsewhere - will be addressed when spikes show up and the increased patrols are moved. Criminals won't enjoy having to guess where to commit crimes without getting arrested. When their 'safe areas' aren't safe any more then it will be interesting to see what the models predict.
As for racial profiling? Is it profiling when only the crimes themselves are used as data points? If more crime is being committed in neighborhoods that are mostly purple Martians then I don't think it's profiling to say "Per the police reports more crimes are committed in neighborhoods that are mostly populated by purple Martians." It's a fact.
If this leads to better use of limited resources (police patrols) and a reduction in crime then how is it a bad thing?
Politicians also know that vilifying their opponents is counterproductive because one day they may have to form a coalition with the party they're vilifying now. That means party relations are a lot less toxic than in the USA, and coalition governments spend less time dismantling the work of the previous coalition than is common in the US.
I've also read somewhere that in the US, they used to have a time when the president and vice were the two topmost voted, no matter which party they come from (unlike nowadays, where each party sets up a candidate-president, and a candidate-vice, and the come together in the same package, depending on which top party was voted).
I'm laughing trying to imagine an alternate reality where the US kept those law around, and Trump and Clinton were forced to work together (one being the vice president). Trolling that the other candidate is unqualified might a lot less successful strategy.
Of course, it might be easier to laugh at such combo, when here around (Switzerland) we don't have a single head of state, but a seven headed hydra : the "head of executive power" is shared among a group of seven men and women.
(There exist a label "President" that is passed around, but is only a label passed around with no political power. The power is indeed evenly shared among seven).
These seven people HAVE to work together, despite different party backgrounds (usually they are from all over the spectrum)
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
Do you know what has been proven time and time again to decrease crime? police officers getting their fat, donut eating asses out of their patrol cars and walking a beat like they used to!
Now, let's hear from all the cops about how they can't do that any more. (due to the extra tonnage around their waists from the donuts no doubt)
Guess we'll have to mix it up a bit, then! ;-)
Self-importance and self-indulgence is the root of ALL evil.