Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Note: You can take 10% off all Slashdot Deals with coupon code "slashdot10off." ×

Comment Re:The problem with neural networks (Score 1) 44

It's not impossible. More complex, but not impossible. You take the input from that caused that (car's black box), see which parent and child neurons fired indicating the undesired action and then check those neurons against the training set to see what aspects of the training set conditioned that behavior (aka, made those neurons respond to that image). You know now what caused that. If you want, you can then modify the network. Or you can take the inputs that caused the incorrect response, make some small changes to it to so that it is representative of other similar situations to come up with a 1,000 item set, and then add it to your training data as a negative set (aka what not to do). Then re-verify after training that those neurons no longer fire in that situation and cause it to drive off of a cliff. Actually, it's almost more intuitive than finding some threading issue, race condition, buffer overflow or multitude of other small, nasty bugs that you have to deal with on a normal basis. It just takes a lot longer to run the tools to see what the issue is and retrain the network.

Comment Re:quickly to be followed by self-driving cars (Score 1) 904

Your advice made sense years ago; these days it does not. 2-5 year old cars with low mileage don't cost much less than brand-new models these days, unless it's some unpopular model (and they're unpopular for a good reason).

My wife's 2 year old Honda Pilot with low mileage cost ~$12k less than new. That's a hunk of change. Now, the payments were a bit higher since you don't get 0% and you don't get to stretch it out over 7 years or whatever crazy long term that they have for new cars now, but you do still save significant chunks of change buying used. If we had gone even a couple of years older, the savings to be had were over 50% versus new for cars that are reliable, popular and have >100,000 miles likely left in them.

Comment Re:Just another case.... (Score 1) 184

Have you seen the Linux ATA/SATA and other code bases like Audio, Video, etc (likely AHCI also)? They're chock full of work arounds for various chipsets, drivers and firmwares. Acting like workarounds aren't effectively industry standard is a little silly. Linux has adapted to its fair share of odd hardware that doesn't work quite as expected.

Comment Re:Crying wolf (Score 5, Informative) 184

What makes you think that? Samsung is one signature away (PIA -- Proprietary Information Agreement) from viewing the vendor's source code and advising them. It's pretty damn routine and uncontroversial. I don't understand why people think that just because something is not open source that no one outside of the company ever, ever, under any circumstances can see a hunk of the code. Just sign a PIA and over the code in a secure manner, or give them remote VPN access to the test box. Pretty damn simple and routine.

Comment Re:No More Bennett (Score 3, Insightful) 187

Exactly. This is pretty tripe. He admits up front that the bug bounty program says "No brute forcing of other users account" and then assumes that brute forcing is ok. There's also the possibility that they meant that brute forcing in general is not ok, so just tossed his submission when it arrived because it was a brute force attack. My guess is that they already knew it could be brute forced and were looking for other potential security issues to find and implement as a group before they push the next update -- that they were actually looking for a little more in depth security issues than that.

I have to say that I'm not honestly surprised that Bennett didn't think of that conclusion, because it would require more than a strict literal interpretation of something and navel gazing, which really are his two specialties.

Comment Re:Diversity (Score 1) 287

Yup. But Google's point (not in this article, but in an interview I believe) is that right now there's only a certain number of minority software engineers. There's a finite quantity at this instant. ALL of the SV tech companies are fighting over them -- trading them back and forth essentially as the employee hops due to better job offers. So, Google's numbers may go up, but Facebook's may go down as a result. This is why Google, Facebook, et al all have huge programs supporting STEM for minority elementary, middle and high school students. That's the only way that it can be fixed, and why these reports are silly. The only report we really need is unemployment %'s in the industry by minority status. My totally uninformed guess is that that statistic looks pretty fair, if not in favor of minorities...

Comment Re:Is that even correct ? (Score 1) 185

The error is that the mirror absorbs a photon and then emits one. That's how mirrors work -- they don't physically reflect photons, they absorb and emit new ones. Thus, the mirror would have to be capable of handling a 5kW instantaneous flux without degradation. That's hard to do on an external surface that's prone to getting a bit mucked up. I mean, the mirror helps, but there are practical considerations with respect to making one good enough to handle that level of incoming power flux.

Comment It would have worked if they had the right team (Score 1) 123

They had a good idea that generated a ton of interest. They got a ton of money to do it. But the team that they put together just didn't have the right skillset mix to pull off something so ambitious. Some of their team posting in their forums and their official updates showed a pretty serious lack of knowledge in some crucial areas. Their original UI and framework was a train wreck (haven't checked back in a year). A number of people bought it to serve as a kind of media aggregator -- run Plex, XBMC, some emulators and original Indie or other content. Then they panicked that lots of people were so interested in getting XBMC/Plex onto it that let out some updates that borked the ability to do that, and really burnt a lot of people

I was ok with the media center parts of it being worthless. I understood that they had a vision for gaming and were focused on it (although executed it poorly), and so was begrudgingly ok with the fact that they were throwing up a walled garden focused on gaming, rather than nurturing a vibrant hacker community. They killed that community, which it turns out was a lot of their customers and things withered. They really could have been a Raspberry Pi with a controller. But their controller just absolutely sucked. Their kickstarted called it "a tribute to all classic controllers out there. This will be the best controller ever." or something to that effect, which was one of the key reasons I bought it. They made it sound like they had spent some very serious effort building an awesome controller. The controller was really, really bad. If you can't play games well with the system, then it's not going to succeed.

Comment Re:Why such crap? (Score 2) 263

No, but mis-printed pages, illegible pages, pages falling out of binders, getting out of order, having coffee spilled on them, etc. all have happened. My guess is that the incidence rate might actually be lower with iPads. Printing 35lbs of paper for each and every flight of which there are thousands of a day, which is different for each and every flight is not something that happens error free. In fact, it's something that absolutely screams automation and computer-based workflow. I've seen numerous bad flight books, or having to rush new flightbooks to the plane because the other ones were wrong. I bet that the total incidence rate is lower with the iPad version. It's just that this makes news.

Comment Re:Wow ... (Score 2) 263

This. So much this. I don't get why people don't understand staggered roll outs. Do an update on one, wait two weeks, then update the other. Or heck, do a tick/tock update where they're always on slightly different versions.

Google does this for a reason with all of their updates in the Android store, and lots of major devs do it also. It's built into the deployment tool, where you can specify all at once or how to dole it out so that you see major bugs before they affect you're entire group. I can only assume that they have a similar mechanism that will allow it to happen simply for their custom iPad app.

"Why should we subsidize intellectual curiosity?" -Ronald Reagan