1) Every net user would have to get an official 'net passport' where they've been authenticated and biometrically IDed.
2) Every time you log onto the net, you'd have to authenticate again, using your biometric ID. New hardware would have to be added to every computer to do that. (It's really easy to spoof a photo ID if you're the one taking the picture.)
3) Every country would have to use an equally trustworthy and inviolable authentication technology.
4) Every country would have to share their user database with every other.
5) All databases would have to be kept up-to-date.
6) The majority of citizens of every country would have to agree to national ID cards (the first step in creating a net database).
Computers were fun back when the reward was worth the effort. Poking data into the display buffer, writing short bits of code in machine language to open the door of CD drive -- the direct connection between software and hardware -- that's what I liked.
Today the best way to do that is probably to build a robot or some other sort of embedded system. Watching your Lego-bot roll around the floor and respond to input according to your rules is a lot more engaging than calling Qt to put up a button or OpenGL to draw a square.
It's obvious pretty quickly that 'Hello World' isn't exactly the door to Narnia.
At this point I could not care less what's up with TBird. TB3 was so badly mangled that after years of using TB, their 'latest greatest' convinced me to I give up and (ugh) switch to Apple Mail. It's the only Apple app I use routinely, which should say something about how much of an albatross TB3 has become.
I scanned the original Sophos report and found several problems with its authenticity.
There is no attribution at all for the claim (on page 16) that 42% of the malware sites are in the US. Where *did* they get this important number?
Reference #63 is supposedly the source that 'Mom and Pop websites' were overcome by a FTP vulnerability, but the source, a Sophos blog entry, says nothing of the kind. It discusses only the vulnerability, not the target or victims.
I'd like to point out that since Sophos sells antiviral software, their promotion of any report that paints a dire picture of web threats (whether from crackers or gov'ts) may well be colored by ulterior motives.
Outstanding advice. I have a BS in bio and an MS in CS plus 20 years of experience, 80% in R&D (supercomputing/sci programming, DoD C^4I, AI). Presently, I do medical image analysis R&D in a giant pharma. My experience confirms mbkennel's advice. But I would avoid scientific programming. It's a support job that leads only to more of the same. You will likely work beneath postdocs and remain employed only as long as long as your current project remains funded.
More generally, without a PhD you will never lead an R&D team. You will always be a subordinate. This is worst in pure sciences, in academia and at large east coast corporations, and probably best in engineering and at small startups.
My recommendation: look at jobs in bioinformatics (or even comp. bio) that 'require' a MS. Talk to others who are working in such a role to learn whether they really are in a leadership position (and not just extolled the potential of one).
Also: consider a MS in one of the engineerings -- EE, ME, Mat Sci, or Eng Sci. Then find work in industry. Licensed professional engineers are recognized by most for-profit employers as first string players and team leaders. The folks who lead engineering teams, no matter how large (like space shuttles or 787s), usually are pro engineers w/ MSs, and not PhDs. The exceptions are, again, east-coast giant corporations who are more afraid of failing than excited about winning.
Finally, avoid a degree in the sciences unless it's a PhD + postdoc(s). There's a perpetual glut of PhD physicists (and soon, chemists & biologists). When competing for a science job, a MS in science will lose out to these folks every time (since the project manager will also have a PhD, and will see you as 'one of *them* and not 'one of *us*').
Revolutionary IT is an oxymoron. IT is all about deep infrastructure and you can't revolutionize the status quo; you can only evolutionize it after first understanding it thoroughly and then chipping away at the edges. No prof is willing to immerse him/herself in that level of ritual embowelment just to win a $100k contract.
Now robotics might be different. A new robot can serve an isolated niche for DARPA which doesn't require the professor to first understand the workings of a huge and complex hierarchical organization like the US military.
The same is true of spy technology. Though IMHO DARPA should be prohibited from funding spy tech since it's so likely to be misguided to civilian targets and then misused domestically, info fusion and surveillance are readily served by a clever prof with a better mousetrap.
BTW, the earlier oft-mentioned 1998 study of scientist faith published in Nature magazine defined 'scientist' rather differently. Their sample included only biologists, physicists, mathematicians, and astronomers. They identified merely 7% of scientists as religious, summarizing:
"Our chosen group of "greater" scientists were members of the National Academy of Sciences (NAS). Our survey found near universal rejection of the transcendent by NAS natural scientists. Disbelief in God and immortality among NAS biological scientists was 65.2% and 69.0%, respectively, and among NAS physical scientists it was 79.0% and 76.3%. Most of the rest were agnostics on both issues, with few believers. We found the highest percentage of belief among NAS mathematicians (14.3% in God, 15.0% in immortality). Biological scientists had the lowest rate of belief (5.5% in God, 7.1% in immortality), with physicists and astronomers slightly higher (7.5% in God, 7.5% in immortality)."
The best potential of tech in government is to turn their spy technology on themselves. We the people can be Big Brother.
Put cameras and microphones in every pol's office and videorecord and mike their entire day. Then store the record in a publicly web accessible read-only vault. Forever. Mike them and their staff 24x7, at all their off-site social engagements too.
As a servant of the people, no pol could refuse to play by the New Transparency and still hope to be reelected. It would so reduce any opportunity for quid pro quo that we wouldn't need campaign finance reform. The losers would quickly reveal their dark side no matter how hard they tried to conceal it.
I became aware of Martin Gardner as an adolescent (in the 1970s), and I've often looked to his work as the best evidence that a self taught amateur can ably and artfully pursue a love of science, math, and a rational life without the aid (or hindrance) of advanced university degrees or a lab full of expensive equipment.
Mr Gardner's infectious enthusiasm and excellence in recounting his explorations played a big role in shaping my sense of self and my appreciation for the positive role model that an amateur mathematician can play in the life of another person, namely me.
You're mistakenly equating invention (engineering) with discovery (science). There are many examples of home invention, not to mention the many tech businesses that sprang from garages or small numbers of founders (e.g. HP, Apple, Oracle, Google, Sun, SGI, iRobot, Dean Kamen's firm, Kurzweil Synthesizers, etc).
Many very successful inventions sprang from the minds of individuals (intermittent wipers, Breathe Right Nasal Strips, Gatorade, Ronco kitchen products), few of which were technologically sophisticated.
The road to success takes no more than a single good idea followed by a *lot* of development. (e.g. Edison's many trials to find the right filament material for his light bulb.). What you lack in genius or education can certainly be surmounted with a bit of cleverness and a lot more persistence.
If Israel's WiFi standard is like Europe's, this begs several questions:
0) Isn't it likely that Apple has already explored this ground? If not prior to the iPad's rollout, then after introducing the same technology in the iPhone when they introduced it over a year ago throughout europe and Israel?
1) Why haven't european regulators also rejected the iPad? Since they apparently have not, they must have tested the iPad (or grandfathered it as comparable iPhone tech) and accepted it.
2) Since the europeans tested and accepted the iPad, why haven't the Israelis accepted the european test results since they're supposedly equivalent?
Sounds like the Israelis are waving a red herring. Either they're protecting an in-country product or license, or they're punishing Apple for something. Either way, this kind of pissy petulence makes them sound like a snotty child.
It's just a matter of time before BB data is subpoenaed in a lawsuit, thereby opening the floodgate to BB data inclusion in *all* car liability lawsuits and probably most traffic violation litigation. Frankly I'm surprised BB's haven't become a mainstay in traffic court already. It'd be an unbiased improvement on the he-said-she-said prosecutions of today's grievously-flawed law enforcement process. Turn on those cameras!
When the cost of producing one more of your product is zero (e.g. on-line media), your pricing model has to change. You're not delivering a rag to peoples' doorsteps any more. If priced right, it's feasible to sell a billion more of your product. If priced wrong, you're overlooked and irrelevant. And bankrupt. That's mistake #1.
Of course, if you charge a truly nominal fee ($1-$5/month) you'll attract a far greater number of readers, and wield a great deal more influence politically and culturally. For a newspaper like the NYT, that's especially important if you want mindshare and if your columnists are to win Pulitzers. Increasing the number of eyes on your prize "gets out The Word".
Your price also needs to reflect the way people use your product. Unlike readers of snail newspapers, most net-denizens follow many media sources. But they scan them, not read them closely they way you read a newspaper. If you price the product like a newspaper, you force the consumer into *your* model of consumption, not theirs. That's mistake #2.
I'm a big fan of the NYT, and would happily subscribe to continue my access, but only at the right price. That said, I'd be hard pressed to pay for ANY daily newspaper these days, so a subscription fee over $60/year just ain't gonna happen. I'd go back to taking The Economist on paper and RSS the Reuters/AP websites instead.
Wake up, NYT. In the e-conomy you either publish MORE or you perish.
You can't hide from Big Brother, but you can confuse the hell out of him.
Do this by behaving inconsistently, in ways that complicate spammeisters from slotting your into a standard bucket.
Leave the TV tuner box set to a channel you hate (e.g. Country Music TV, Fox News, MSNBC, TLC, Family Channel) and then turn off the set. Choose a different odious channel each time. Or choose channels randomly.
Lie on the shopper discount card questionaires. In time, most places will disambiguate you (if you use a credit card), but your misbehavior will probably flag you as a spoil sport who won't be receptive to spam.
Even if this stuff doesn't protect you, it'll make you feel like you're taking arms against being stamped, indexed, briefed, and debriefed.
Aside from logistics, there is no excuse for not doing this. In my experience, software innovations are notoriously sensitive to subtleties in input data (e.g. data mining, AI, image processing). Posting both code & data (and a test driver, of course) should be mandatory for all publications that claim to have found a signal in data, better or faster.
The question is, how to maintain code & data long after the publication publishes? IMHO, any peer-reviewed publication should be required to maintain such a repository for perhaps 20-30 years, ideally under a GPL (or its kin) so access to it would be free in perpetuity.
Maybe such a service would finally justify peer reviewed pubs' exorbitant fees for non-subscriber access.
Before about 1985, I would have agreed. Since then research AI has devolved into developing a subset of NP-hard algorithms, which seem to either recognize or anticipate patterns (machine learning) or use feedback toward behaving adaptively. Good Old Fashioned AI (GOFAI or 'strong AI') passed away peacefully some 25 years ago, and given the ephemeral timeline of modern academic funding ('publish ASAP or perish'), that's unlikely to change any time soon.
On the other hand, the rising demand for espionage/surveillance tools and no-person-in-the-loop warfare systems may breathe life back into GOFAI. In the US's never ending quest to make warfare ever more painless for the voters, it's likely that DARPA will soon throw big money at building fleets of killer robots.
So the good news is, maybe 'strong AI' isn't dead. The bad news is, 'strong' has an ominous new meaning.
Your math knowledge depends on two questions: 1) What domain do you intend to work in (continuous or discrete), and 2) Do you plan to do research.
If you might deal with data coming from a continuous domain like the Real World, you need linear algebra, calculus, and statistics, i.e. engineering math. This spans professions like signal processing, image processing, graphics, data mining, machine learning, simulation, financial engineering, and AI (where you need to recognize and manipulate complex patterns).
However, if your data will arise entirely from discrete domains, i.e. the mainstream world of computing, you'll still benefit from probability (to understand and characterize the range of variation in input data), but you won't need more than the first term of calculus (differential) or linear algebra, and probably not even statistics. This domain covers 95% of the work of programming: web technology, databases, networking, GUIs, mobile, etc.
In any case, you *will* need discrete math (to understand the theory of computing). And in no case will you ever use real analysis (the theory of calculus).
If you want to do research, plan to take more math regardless (like queuing theory and statistics). But which math you need will depend on which area of research you choose.
Thing is that the person in question did not depart from either a UK or a US airport. Schiphol Arirport already had 15 such scanners and both the Airport's management and the Dutch Interior Minister announced yesterday they intend to get 60 more this year.
Exactamundo. Remember the terrorist was screened only in Nigeria. No new scanner technology in US or Netherland airports would have made any difference. The obvious reaction must be to fix the weakest link in the chain (in Nigeria), not the strongest (in US or Netherlands).
What's more, if a candidate screening technology doesn't improve the strongest link both in a usable and reliable manner, then the 'fixed' chain is no stronger; it just gets more expensive. Bruce Schneier correctly observed that a new scanner with a hit rate of only 85% correct (and 15% incorrect), will generate so many false positives as to be totally unusable in a high-throughput production environment like an airport. To be useful, any new scanner technology has to be damned-near perfect.
> If the government wanted to learn something about you, they would just go to your friendly neighborhood telecom oligopoly.
And if someone wants to break into your house, they'll just kick in the door. Knowing that, do you leave your house unlocked?
Of course not. Likewise, if you make it harder for someone to spy on you, they'll have to try harder. And because most e-spies are lazy and count on your e-door being unlocked, when it's not, they'll break into your e-home less often.
First of all, the authors have chosen the wrong metaphor. HTS is about sending a large number of different organisms through the same assay, looking for candidate targets that produce a positive response to that one test. The folks at Rowland are doing the inverse. They're sending a large number of assays at a single target (many candidate algorithm variations (assays) seeking to recognize an image or a specific pattern therein).
Second, they're generating candidate algorithms, not just algorithm parameters. That's more like genetic programming than genetic algorithms.
Third, the kind of algorithms they describe sound more like neural nets with a large combinatoric variation of units, layers, feedback components, and sundry other possible variants. These seem to me to be both a biologically plausible mechanism for a biologically inspired vision model, not to mention a viable candidate for the pattern recognition of a wide range of vision targets. GA-based mechanism would be too symbolic and boolean to be biologically plausible, and too rigid in matching specific input patterns.
Fourth, a better metaphor for the parameter sweep they describe in the video is monte carlo methods (generating and testing many many parameter combinations using a supercomputer), not HTS, HCS, GP, or GAs.
And no, I haven't read their research papers yet. But I definitely will. Interesting stuff. I wonder if they have/will tap-ped into David Marr's bio-inspired vision models & theories?
No degree or college or professor or course can make you great programmer. All great coders are self-made -- they learn by reading the code of the best programmers they know and then write the best code they can. Training is irrelevant.
Of course, this is true of all great people, not just programmers.
I've taken college courses from the 1975-1982 pre-slide era and from 2000-2009 with slides. Believe me, having slides is ALWAYS better than not. A fast prof is far less of a problem than an illegible or incoherent one.
As to profs with slides who move too fast... The solution is simple. Print the slides before class and then write your notes on them. Any reputable prof makes their slides available (which is a damned sight better than your having to transcribe their illegible chalkboard scribblings, ESP. in math classes). Just ask that they be posted before class so you can print 'em.
FYI for the younger crowd... Back before the 'net, the better departments/profs actually would designate a student to transcribe and rewrite each class's lecture notes for redistribution to the rest of the class, just so note takers didn't have to struggle to keep pace with a fast lecturer. Today's presentation slides do this for us.
Final advice... Get used to it. You're going to see slides in every single presentation you attend as a pro. Bad presenters are inevitable, but don't blame the medium for a message that sucks.
Slashdot Mirror
For 'Schmidt World' to happen:
1) Every net user would have to get an official 'net passport' where they've been authenticated and biometrically IDed.
2) Every time you log onto the net, you'd have to authenticate again, using your biometric ID. New hardware would have to be added to every computer to do that. (It's really easy to spoof a photo ID if you're the one taking the picture.)
3) Every country would have to use an equally trustworthy and inviolable authentication technology.
4) Every country would have to share their user database with every other.
5) All databases would have to be kept up-to-date.
6) The majority of citizens of every country would have to agree to national ID cards (the first step in creating a net database).
No time soon.
Computers were fun back when the reward was worth the effort. Poking data into the display buffer, writing short bits of code in machine language to open the door of CD drive -- the direct connection between software and hardware -- that's what I liked.
Today the best way to do that is probably to build a robot or some other sort of embedded system. Watching your Lego-bot roll around the floor and respond to input according to your rules is a lot more engaging than calling Qt to put up a button or OpenGL to draw a square.
It's obvious pretty quickly that 'Hello World' isn't exactly the door to Narnia.
E's bleedin' demised!
At this point I could not care less what's up with TBird. TB3 was so badly mangled that after years of using TB, their 'latest greatest' convinced me to I give up and (ugh) switch to Apple Mail. It's the only Apple app I use routinely, which should say something about how much of an albatross TB3 has become.
I scanned the original Sophos report and found several problems with its authenticity.
There is no attribution at all for the claim (on page 16) that 42% of the malware sites are in the US. Where *did* they get this important number?
Reference #63 is supposedly the source that 'Mom and Pop websites' were overcome by a FTP vulnerability, but the source, a Sophos blog entry, says nothing of the kind. It discusses only the vulnerability, not the target or victims.
I'd like to point out that since Sophos sells antiviral software, their promotion of any report that paints a dire picture of web threats (whether from crackers or gov'ts) may well be colored by ulterior motives.
Outstanding advice. I have a BS in bio and an MS in CS plus 20 years of experience, 80% in R&D (supercomputing/sci programming, DoD C^4I, AI). Presently, I do medical image analysis R&D in a giant pharma. My experience confirms mbkennel's advice. But I would avoid scientific programming. It's a support job that leads only to more of the same. You will likely work beneath postdocs and remain employed only as long as long as your current project remains funded.
More generally, without a PhD you will never lead an R&D team. You will always be a subordinate. This is worst in pure sciences, in academia and at large east coast corporations, and probably best in engineering and at small startups.
My recommendation: look at jobs in bioinformatics (or even comp. bio) that 'require' a MS. Talk to others who are working in such a role to learn whether they really are in a leadership position (and not just extolled the potential of one).
Also: consider a MS in one of the engineerings -- EE, ME, Mat Sci, or Eng Sci. Then find work in industry. Licensed professional engineers are recognized by most for-profit employers as first string players and team leaders. The folks who lead engineering teams, no matter how large (like space shuttles or 787s), usually are pro engineers w/ MSs, and not PhDs. The exceptions are, again, east-coast giant corporations who are more afraid of failing than excited about winning.
Finally, avoid a degree in the sciences unless it's a PhD + postdoc(s). There's a perpetual glut of PhD physicists (and soon, chemists & biologists). When competing for a science job, a MS in science will lose out to these folks every time (since the project manager will also have a PhD, and will see you as 'one of *them* and not 'one of *us*').
My point: one proposal by one junior CS DARPA researcher isn't going to revolutionize the US military's software, not even a little.
And IIRC data centers *evolved*. The transition from COBOL and VMS to PHP and Apache didn't happen overnight.
Revolutionary IT is an oxymoron. IT is all about deep infrastructure and you can't revolutionize the status quo; you can only evolutionize it after first understanding it thoroughly and then chipping away at the edges. No prof is willing to immerse him/herself in that level of ritual embowelment just to win a $100k contract.
Now robotics might be different. A new robot can serve an isolated niche for DARPA which doesn't require the professor to first understand the workings of a huge and complex hierarchical organization like the US military.
The same is true of spy technology. Though IMHO DARPA should be prohibited from funding spy tech since it's so likely to be misguided to civilian targets and then misused domestically, info fusion and surveillance are readily served by a clever prof with a better mousetrap.
With the aid of Google Books, I found the composition of 'scientists' in Ecklund's survey to be:
241 physics
214 chemistry
289 biology
228 sociology
207 economics
225 political science
205 psychology
BTW, the earlier oft-mentioned 1998 study of scientist faith published in Nature magazine defined 'scientist' rather differently. Their sample included only biologists, physicists, mathematicians, and astronomers. They identified merely 7% of scientists as religious, summarizing:
"Our chosen group of "greater" scientists were members of the National Academy of Sciences (NAS). Our survey found near universal rejection of the transcendent by NAS natural scientists. Disbelief in God and immortality among NAS biological scientists was 65.2% and 69.0%, respectively, and among NAS physical scientists it was 79.0% and 76.3%. Most of the rest were agnostics on both issues, with few believers. We found the highest percentage of belief among NAS mathematicians (14.3% in God, 15.0% in immortality). Biological scientists had the lowest rate of belief (5.5% in God, 7.1% in immortality), with physicists and astronomers slightly higher (7.5% in God, 7.5% in immortality)."
The best potential of tech in government is to turn their spy technology on themselves. We the people can be Big Brother.
Put cameras and microphones in every pol's office and videorecord and mike their entire day. Then store the record in a publicly web accessible read-only vault. Forever. Mike them and their staff 24x7, at all their off-site social engagements too.
As a servant of the people, no pol could refuse to play by the New Transparency and still hope to be reelected. It would so reduce any opportunity for quid pro quo that we wouldn't need campaign finance reform. The losers would quickly reveal their dark side no matter how hard they tried to conceal it.
Rats will flee a glass ship of state.
I became aware of Martin Gardner as an adolescent (in the 1970s), and I've often looked to his work as the best evidence that a self taught amateur can ably and artfully pursue a love of science, math, and a rational life without the aid (or hindrance) of advanced university degrees or a lab full of expensive equipment.
Mr Gardner's infectious enthusiasm and excellence in recounting his explorations played a big role in shaping my sense of self and my appreciation for the positive role model that an amateur mathematician can play in the life of another person, namely me.
Goodbye Mr Gardner and thank you.
You're mistakenly equating invention (engineering) with discovery (science). There are many examples of home invention, not to mention the many tech businesses that sprang from garages or small numbers of founders (e.g. HP, Apple, Oracle, Google, Sun, SGI, iRobot, Dean Kamen's firm, Kurzweil Synthesizers, etc).
Many very successful inventions sprang from the minds of individuals (intermittent wipers, Breathe Right Nasal Strips, Gatorade, Ronco kitchen products), few of which were technologically sophisticated.
The road to success takes no more than a single good idea followed by a *lot* of development. (e.g. Edison's many trials to find the right filament material for his light bulb.). What you lack in genius or education can certainly be surmounted with a bit of cleverness and a lot more persistence.
Good luck.
If Israel's WiFi standard is like Europe's, this begs several questions:
0) Isn't it likely that Apple has already explored this ground? If not prior to the iPad's rollout, then after introducing the same technology in the iPhone when they introduced it over a year ago throughout europe and Israel?
1) Why haven't european regulators also rejected the iPad? Since they apparently have not, they must have tested the iPad (or grandfathered it as comparable iPhone tech) and accepted it.
2) Since the europeans tested and accepted the iPad, why haven't the Israelis accepted the european test results since they're supposedly equivalent?
Sounds like the Israelis are waving a red herring. Either they're protecting an in-country product or license, or they're punishing Apple for something. Either way, this kind of pissy petulence makes them sound like a snotty child.
"I'll take my ball away and play with myself."
It's just a matter of time before BB data is subpoenaed in a lawsuit, thereby opening the floodgate to BB data inclusion in *all* car liability lawsuits and probably most traffic violation litigation. Frankly I'm surprised BB's haven't become a mainstay in traffic court already. It'd be an unbiased improvement on the he-said-she-said prosecutions of today's grievously-flawed law enforcement process. Turn on those cameras!
When the cost of producing one more of your product is zero (e.g. on-line media), your pricing model has to change. You're not delivering a rag to peoples' doorsteps any more. If priced right, it's feasible to sell a billion more of your product. If priced wrong, you're overlooked and irrelevant. And bankrupt. That's mistake #1.
Of course, if you charge a truly nominal fee ($1-$5/month) you'll attract a far greater number of readers, and wield a great deal more influence politically and culturally. For a newspaper like the NYT, that's especially important if you want mindshare and if your columnists are to win Pulitzers. Increasing the number of eyes on your prize "gets out The Word".
Your price also needs to reflect the way people use your product. Unlike readers of snail newspapers, most net-denizens follow many media sources. But they scan them, not read them closely they way you read a newspaper. If you price the product like a newspaper, you force the consumer into *your* model of consumption, not theirs. That's mistake #2.
I'm a big fan of the NYT, and would happily subscribe to continue my access, but only at the right price. That said, I'd be hard pressed to pay for ANY daily newspaper these days, so a subscription fee over $60/year just ain't gonna happen. I'd go back to taking The Economist on paper and RSS the Reuters/AP websites instead.
Wake up, NYT. In the e-conomy you either publish MORE or you perish.
You can't hide from Big Brother, but you can confuse the hell out of him.
Do this by behaving inconsistently, in ways that complicate spammeisters from slotting your into a standard bucket.
Leave the TV tuner box set to a channel you hate (e.g. Country Music TV, Fox News, MSNBC, TLC, Family Channel) and then turn off the set. Choose a different odious channel each time. Or choose channels randomly.
Lie on the shopper discount card questionaires. In time, most places will disambiguate you (if you use a credit card), but your misbehavior will probably flag you as a spoil sport who won't be receptive to spam.
Even if this stuff doesn't protect you, it'll make you feel like you're taking arms against being stamped, indexed, briefed, and debriefed.
Aside from logistics, there is no excuse for not doing this. In my experience, software innovations are notoriously sensitive to subtleties in input data (e.g. data mining, AI, image processing). Posting both code & data (and a test driver, of course) should be mandatory for all publications that claim to have found a signal in data, better or faster.
The question is, how to maintain code & data long after the publication publishes? IMHO, any peer-reviewed publication should be required to maintain such a repository for perhaps 20-30 years, ideally under a GPL (or its kin) so access to it would be free in perpetuity.
Maybe such a service would finally justify peer reviewed pubs' exorbitant fees for non-subscriber access.
> It goes back and forth in waves, though.
Before about 1985, I would have agreed. Since then research AI has devolved into developing a subset of NP-hard algorithms, which seem to either recognize or anticipate patterns (machine learning) or use feedback toward behaving adaptively. Good Old Fashioned AI (GOFAI or 'strong AI') passed away peacefully some 25 years ago, and given the ephemeral timeline of modern academic funding ('publish ASAP or perish'), that's unlikely to change any time soon.
On the other hand, the rising demand for espionage/surveillance tools and no-person-in-the-loop warfare systems may breathe life back into GOFAI. In the US's never ending quest to make warfare ever more painless for the voters, it's likely that DARPA will soon throw big money at building fleets of killer robots.
So the good news is, maybe 'strong AI' isn't dead. The bad news is, 'strong' has an ominous new meaning.
The probability should be p=2^-14. A p value of 214 would be an amazingly low probability.
This is why we computer scientists need to study more math.
Your math knowledge depends on two questions: 1) What domain do you intend to work in (continuous or discrete), and 2) Do you plan to do research.
If you might deal with data coming from a continuous domain like the Real World, you need linear algebra, calculus, and statistics, i.e. engineering math. This spans professions like signal processing, image processing, graphics, data mining, machine learning, simulation, financial engineering, and AI (where you need to recognize and manipulate complex patterns).
However, if your data will arise entirely from discrete domains, i.e. the mainstream world of computing, you'll still benefit from probability (to understand and characterize the range of variation in input data), but you won't need more than the first term of calculus (differential) or linear algebra, and probably not even statistics. This domain covers 95% of the work of programming: web technology, databases, networking, GUIs, mobile, etc.
In any case, you *will* need discrete math (to understand the theory of computing). And in no case will you ever use real analysis (the theory of calculus).
If you want to do research, plan to take more math regardless (like queuing theory and statistics). But which math you need will depend on which area of research you choose.
Thing is that the person in question did not depart from either a UK or a US airport. Schiphol Arirport already had 15 such scanners and both the Airport's management and the Dutch Interior Minister announced yesterday they intend to get 60 more this year.
Exactamundo. Remember the terrorist was screened only in Nigeria. No new scanner technology in US or Netherland airports would have made any difference. The obvious reaction must be to fix the weakest link in the chain (in Nigeria), not the strongest (in US or Netherlands).
What's more, if a candidate screening technology doesn't improve the strongest link both in a usable and reliable manner, then the 'fixed' chain is no stronger; it just gets more expensive. Bruce Schneier correctly observed that a new scanner with a hit rate of only 85% correct (and 15% incorrect), will generate so many false positives as to be totally unusable in a high-throughput production environment like an airport. To be useful, any new scanner technology has to be damned-near perfect.
> If the government wanted to learn something about you, they would just go to your friendly neighborhood telecom oligopoly.
And if someone wants to break into your house, they'll just kick in the door. Knowing that, do you leave your house unlocked?
Of course not. Likewise, if you make it harder for someone to spy on you, they'll have to try harder. And because most e-spies are lazy and count on your e-door being unlocked, when it's not, they'll break into your e-home less often.
Google: 'Our motto is, "Don't be evil"'.
'What? You thought that meant US? Silly boy. It means YOU!'
First of all, the authors have chosen the wrong metaphor. HTS is about sending a large number of different organisms through the same assay, looking for candidate targets that produce a positive response to that one test. The folks at Rowland are doing the inverse. They're sending a large number of assays at a single target (many candidate algorithm variations (assays) seeking to recognize an image or a specific pattern therein).
Second, they're generating candidate algorithms, not just algorithm parameters. That's more like genetic programming than genetic algorithms.
Third, the kind of algorithms they describe sound more like neural nets with a large combinatoric variation of units, layers, feedback components, and sundry other possible variants. These seem to me to be both a biologically plausible mechanism for a biologically inspired vision model, not to mention a viable candidate for the pattern recognition of a wide range of vision targets. GA-based mechanism would be too symbolic and boolean to be biologically plausible, and too rigid in matching specific input patterns.
Fourth, a better metaphor for the parameter sweep they describe in the video is monte carlo methods (generating and testing many many parameter combinations using a supercomputer), not HTS, HCS, GP, or GAs.
And no, I haven't read their research papers yet. But I definitely will. Interesting stuff. I wonder if they have/will tap-ped into David Marr's bio-inspired vision models & theories?
No degree or college or professor or course can make you great programmer. All great coders are self-made -- they learn by reading the code of the best programmers they know and then write the best code they can. Training is irrelevant.
Of course, this is true of all great people, not just programmers.
I've taken college courses from the 1975-1982 pre-slide era and from 2000-2009 with slides. Believe me, having slides is ALWAYS better than not. A fast prof is far less of a problem than an illegible or incoherent one.
As to profs with slides who move too fast... The solution is simple. Print the slides before class and then write your notes on them. Any reputable prof makes their slides available (which is a damned sight better than your having to transcribe their illegible chalkboard scribblings, ESP. in math classes). Just ask that they be posted before class so you can print 'em.
FYI for the younger crowd... Back before the 'net, the better departments/profs actually would designate a student to transcribe and rewrite each class's lecture notes for redistribution to the rest of the class, just so note takers didn't have to struggle to keep pace with a fast lecturer. Today's presentation slides do this for us.
Final advice... Get used to it. You're going to see slides in every single presentation you attend as a pro. Bad presenters are inevitable, but don't blame the medium for a message that sucks.