
Mathematician Keith Devlin of Stanford University is co-author of the book “The Numbers Behind NUMB3RS: Solving Crime with Mathematics”:http://www.amazon.com/Numbers-Behind-NUMB3RS-Solving-Mathematics/dp/0452288576/ref=pd_bbs_sr_1/104-4707333-7765544?ie=UTF8&s=books&qid=1192610228&sr=8-1. He told Earth & Sky about using neural networks – computer programs – to fight credit card crime.
Keith Devlin: _The idea then is to feed that simulated brain, if you like – feed it with lots of data and let it see what patterns it can recognize. And simply if you set it up the right way, that device will recognize patterns. The trick is to build it in such a way that the patterns it recognizes are patterns of use._
A computerized neural network needs to undergo a process of “training” to decide whether a transaction is legitimate. One of the great advantages of neural networks, Devlin explained, is the sheer scope of data they can analyze to discern patterns. If a transaction deviates sharply from the pattern, it gets flagged as a fraud.
Keith Devlin: _So what happens is that they have neural networks that monitor every single credit card transaction that an individual makes and it looks for patterns and then if there’s a sudden change in that pattern then the neural network will recognize it as a change and will alert the fraud department there’s been a possible fraud on that card._
*Our thanks to:*
Professor Keith Devlin
CSLI, Cordura Hall
Stanford, CA
Administrative Assistant: Darla Hart
Pff just because you can make something like that doesnt mean others will make something to get passed it. The more obstacles you make the more funner it is for a person to try to get passed it. There is way more crime today then ever. NASA makes the most useful inventions I think.