Just to give a little context for people who do not live around Atlanta:
Gwinnett County has been a traffic ticket mill for decades. This is not Atlanta - it is an exurb of Atlanta where all the racist, faux-Christian whitebread people moved a couple decades back. All it really has going for it is two major highways running through the county (I-85 and GA-316). Several people I know have had bad experiences there, and I literally do not stop in Gwinnett County anymore.
This is also the place where Larry Flynt was brought to court on obscenity charges, and shot by a white supremacist who confessed but never had charges brought against him (http://en.wikipedia.org/wiki/Larry_Flynt#Shooting).
Please don't confuse Gwinnett with Atlanta. Gwinnett is a shithole that has only strip malls, a thriving prison industry, a growing international population, and an entrenched set of racist white people who don't like the new international population and want to throw them all in jail.
Worst of all, this place tattooed "GWINNETT IS GREAT" and "SUCCESS LIVES HERE" on their ugly ass water towers. Here's a picture I found via Google, ironically on a blog called "stuff black people don't like": http://stuffblackpeopledontlike.blogspot.com/2012/09/success-used-to-live-here-what-fall-of.html (I haven't read the blog posting, but I'm white and don't like Gwinnett, either).
Make no mistake, this cop was ticketing people for texting solely so he could take all the brown ones to jail. This is the Gwinnett MO: enforce traffic violations heavily, and try to turn every traffic stop into a drug bust or driving-without-a-license bust.
Here, take a look at what Gwinnett has to offer: http://www.gwinnettmugs.com/
As long as Nvidia keeps crippling double-precision performance on their (non-Tesla) cards, I'll keep buying AMD.
One of the highlights of the GTX Titan was that the card did double-precision floating point at full speed, just like one of Nvidia's Tesla products. That's no longer the case here - the GTX 780 performs double-precision at 1/24 of normal rate, just like a standard desktop GPU.
The only interesting exploit is one that hasn't been patched, right? So anyone who discovers, sells, or buys an exploit knows of a vulnerability and is choosing not to disclose it.
By not disclosing a vulnerability, you are allowing others to be vulnerable. It's hard to argue that this is ethical behavior...
Here's an analogy: what if, for every nuke the U.S. destroyed, a nuke disappeared from every other nuclear arsenal in the world? That's what it's like.. by keeping a vulnerability secret, it can be used against anyone using the software. By disclosing the vuln, everyone can patch, disable, or protect the vulnerable software.
If I had mod points, I'd mod this up. Rich0 is right, we have a legacy system of shareholder voting that does not work so well now that the biggest shareholders are institutional investors. The votes of mutual fund managers dwarf those of individual stockholders.
Sounds kinda like how the wealthy Saudi nationals were able to fly out of the country after 9/11...
âoeSomebody brought to us for approval the decision to let an airplane filled with Saudis, including members of the bin Laden family, leave the country,â Clarke says.
That would be Richard Clarke, who was Terrorism czar at the time. Funny thing is, nobody can remember who requested the authorization for that flight. I guess billionaires, oligarchs, and their other wealthy friends have a common bond they don't share with their fellow countrymen...
In my ML class, we used WEKA. Of course, there is also Matlab. Problem is, neither of these are free, and they are both slow as hell. I would not use either one outside of class/prototyping.
Ideally there would be a free, open source toolkit written in a compiled language. The toolkit should have a variety of ML techniques that can be switched around with little pain. Only toolkit I know of like this is the ML part of OpenCV, and the documentation for OpenCV is... lacking.
Another poster linked to mloss.org. I hadn't seen that site.. Looks like a great resource, but it also looks like 400+ fragmented tools that do not play well together, and are probably mostly dead projects by now.
I dispute that auditory function in the brain is fairly well understood. *Some* of the fundamentals are fairly well understood.
As an example, there is the Olivocochlear system that feeds back from Superior Olives to the cochlea. We think it may contribute to active amplification of sounds in the cochlea. See wikipedia for a list of PROPOSED functions.
What we do know is that cutting the olivocochlear connection impairs sensitivity in the cochlea. We do know what neurons connect to what other neurons, and have some idea of the types of connections.
What we don't know is "how the thing works".
This system is key to human hearing; it's not just a lump of cells with no known function. So, we are a long ways off from any human- (or even cat-) level auditory models.
These fields you mention (computer vision, speech recognition) are good examples of the state of intelligent machines.
We can make these things work pretty well for very specific tasks (e.g. recognize faces in a picture), but we are nowhere near having general, human-level intelligence. It's hard to see how we are even close to having human-level vision capabilities.
Kurzweil is a "futurist" and "technologist" which means he has some technical background and a big mouth with a lot of hot air.
Sure, he accomplished some research VERY early on with sound and OCR. He has not been on the cutting edge for decades.
Kurzweil's career is summed well with this quote (text copied from Wikipedia's article on Kurzweil):
In the cover article of the December 2010 issue of IEEE Spectrum, John Rennie criticizes Kurzweil for several predictions that failed to become manifest by the originally predicted date. "Therein lie the frustrations of Kurzweil's brand of tech punditry. On close examination, his clearest and most successful predictions often lack originality or profundity. And most of his predictions come with so many loopholes that they border on the unfalsifiable."
So, right angle (aka t-bone crashes) are down, but rear- end collisions are up? That doesn't sound so bad.
Right angle crashes can kill people. Rear-end collisions are fender benders.
There's more detail here, including links to papers: http://voipsecurityblog.typepad.com/marks_voip_security_blog/2012/09/dtmf-telephony-denial-of-service-tdos-issues-for-ivrs.html