acadien

Your 2nd point is critical, you can filter these images easily before even running them through the DNN. However researchers are also interested in why it is possible to spoof NN's in general. The typical response of 'overfitting' is being questioned.

Also the question is raised as to whether or not new methods of spoofing are possible that aren't so easily detectable.

Its the citation chain game! When you first start reading papers in a new field you essentially have to read every citation present to understand whats going on. As you accrue knowledge you'll need to read fewer citations and you'll understand more of the jargon. That, to me, is the primary challenge.

It took me about 2 years before I could read a paper end to end and understand it. Of course by that point I realized that reading most papers was a waste of time. You often only need the figures, conclusion and methods section.

Come on, its 20 lines long.

Time and time again I see people arguing on HN as if they haven't read the article. Is it just bad reading comprehension or is it that people read the headline and jump the gun?

So what if they're capable? The tech is not available, end of story.

"self-driving car" and "car capable driving itself on highway under optimal conditions" are 2 totally different things. The real excitement comes when I can tell my car to go pick up the kids from school or drop me off at the airport and then drive itself back home. Anyways this is a great step in the right direction but its not the game changing tech we're all waiting for.

Please check my edit above.

Thanks for the extra information. It would have been nice if the article author brought up some statistics that could show whether programs like these are under used.

I wasn't clear and I think you missed my point, please see my edit.

The Ivy Leauge admits roughly 20,000 students every year [1]. Lets say they give one fifth of those get free admission. That is a total of 4000 kids which is practically nothing!

Roughly 5 million people are admitted to University every year in the US[2]! So maybe 0.08% of all University students get a free ride at an ivy league. Really this makes no difference except for padding the Ivy League's statistics, its of little to no help for most 'talented low income' highschool students.

[1]http://theivycoach.com/ivy-league-admissions-statistics/

[2]http://nces.ed.gov/fastfacts/display.asp?id=98

Edit: To address most of the responses below. I'm implying that even if all appropriate students applied for these programs, as the article suggests, it wouldn't make a difference because some tiny fraction of them would be admitted. There are very few opportunities for low income students unless they are willing to take on the burden of debt. Some lottery that gives a tiny fraction of them free tuition is not a solution, it is offensive to suggest they have opportunities they simply do not have. Fix the broken system, correct tuition costs!

I'm going to have to agree with the top post, sometimes really smart people say stupid things. I wouldn't read too much into it.

It cracks me up how that is the go to 'application' for all robots, drones and swarm-bots. It is used so often it has become a cliche.

You can get down a layer two but that's about it. There is no method for capturing a full 3D model in experiment, although we can get projections from X-Ray diffraction and other scattering methods. Also there are single photon X-Ray methods being developed that will get us extremely close to complete 3D models. But that aside you also have to take into account that these materials are often not at 0K so they're vibrating or even moving at the atomic scale. This is a huge problem for example, when studying Proteins.

>For one, the absolute leading researchers are at the top universities, and there is a big gap between these people and merely very good researchers.

I have noticed the opposite effect. Typically the best researchers are the worst instructors, prepare the least for their courses and couldn't care less about their teaching quality. Those that are both solid instructors and researchers are exceptionally rare.

While leading researchers may have a better grasp of their specialized field this is often not applicable in graduate courses. The exception being when those professors develop an entire course around their research and (often) force many of their research students into that course.

Is it really so terrible that they try out new projects and abandon those that don't either don't work or aren't profitable?

Location: DC area

Remote: Prefer not

Willing to relocate: Yes

Technologies: Python (numerical stack) C/C++, MPI/openMP, some GPGPU and VHDL experience

Résumé/CV: https://www.dropbox.com/s/dshyukh4g39hxqu/ASCResume2015.pdf?...

Email: adamcadien@gmail.com

I'm 2 months away from finishing my PhD and I have several internships under my belt. Self driven and highly motivated learner that takes pride in his work. Hoping to join a team, expand my skill set and work on interesting problems.

Thanks, exactly the kind of information I was trying to elicit.

What is it that the firmware can achieve? Is the firmware capable hijacking data, communicating with the NIC and transmitting data? Or is it somehow injecting harmful code? I feel like I'm missing something here.

Can't you just write all 0's to the drive or just reformat it? Genuine question here, why would you need to physically replace the drive to ensure security when you can write to the whole thing?

Great point, very true.

White rice is off by 50 calories for example (135 on chipotlefan and 185 on chipotle).

That's odd some of their values differ greatly from the ones provided by Chipotle themselves. This strikes me as a significant problem, nutritional values are provided and are totally inaccurate. Which one is accurate how can we judge?

http://www.chipotle.com/en-US/menu/nutrition_calculator/nutr...

Looks like the information provided by chipotlefan is from 2009.

Isn't part of the problem that production costs for batteries are so high? I suspect this is Tesla staking claim in a market that will grow as the price of large rechargeable batteries drops. And that price drop will in large part be due to Tesla.

There are lots of apartments that don't offer parking as well. Cities will need to offer some way to charge cars on the street, this will certainly be an interesting challenge specifically to figure out who pays for the installation and the power.

There was a project several years ago to map a University's wifi using a self driving RC car with a wifi detector on it. The goal was to identify weak spots in the signal. I can't remember the name of the project and its not popping up on google, anyone else remember this?

Yeah the marijuana community always seems to be on the cutting edge of growing technology, much the same way the gaming community often leads use of cutting computing technology.

Anyways, a 'thousands of acres equivalent' solution would be absolutely fascinating to see implemented.

The downside being its not scalable. You need an entirely artificial setup to get aquaponics up and running, motors, pipes, and chemical cleaning. Don't get me wrong aquaponics is a great technology, I think it needs some changes in order to make it feasible for the large scale.

Are you seriously suggesting pay should be determined by how fun a job is and not how much skill it requires or how much it contributes to society?

For one thing Adderall slightly increases your blood pressure and heart rate. That can have serious cardiovascular side effects in the long run.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2690554/

The other possibility is the irresponsible over use as with drones. If they aren't too expensive you could just strap a bomb on one of these and run it towards any possible threat. I agree with your assessment and look forward to the onset of more nonlethal warfare. I'm just saying the opposite effect is also possible.

It seems like using these devices for delivery is one possible end goal (besides the obvious military applications). A self driving truck alone cannot deliver packages to your front door. However put one of these robot dogs inside and figure out some way for it to pick up and drop off packages and suddenly you can take on FedEx/UPS/etc.

Side note, anyone else really want to ride one of these?

I'm not sure that mach and pH are units per say. pH for example is on a nonlinear scale and mach itself is unitless. Maybe that's totally irrelevant.

I guess its fruitless to try and apply logic to language and better to just accept the general usage?