×

Announcing: Slashdot Deals - Explore geek apps, games, gadgets and more. (what is this?)

Thank you!

We are sorry to see you leave - Beta is different and we value the time you took to try it out. Before you decide to go, please take a look at some value-adds for Beta and learn more about it. Thank you for reading Slashdot, and for making the site better!

Comments

top

Research Highlights How AI Sees and How It Knows What It's Looking At

Capt.Albatross Re:Actually a Great Step Forward (125 comments)

Computer learns to pick out salient features to identify images. Then we are shocked that when trained with no supervision the salient features aren’t what we would have chosen.

There is a huge difference: humans pick relevant features guided by a deep understanding of the world, while machine learning, unguided by any understanding, only does so by chance.

Now that we know what computers are picking out as salient features, we can modify the algorithms to add additional constraints on what additional salient features must or must not be in an object identified, such that it would correspond more closely to how humans would classify objects. Baseballs must have curvature for instance not just zig-zag red lines on white.

Hand-coded fixes are not AI - that would be as if we have we had a higher-level intelligent agent in our heads to correct our mistakes (see the homunculus fallacy).

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

Capt.Albatross Re:Not smart or stupid (125 comments)

These are computer programs, not artificial intelligences as some have come to think of them. They are simply some charges flipping around in some chips.

And minds are just charges flipping around in some brain (at one level of abstraction, it is chemical, but chemistry is explained by the movement of charges.)

As John Searle said, brains make minds.

Everything else is just speculating.

If you look at John Searle's arguments in detail, they ultimately end up as nothing more than "I can't believe that this is just physics." Searle's view is actually rather more speculative than the one he rejects, as it implies an unknown extension to atomic physics.

Nevertheless, none of what I write here should be construed as a claim that artificial intelligence has been achieved.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

Capt.Albatross Re:Clickbait (125 comments)

So it needs to learn that these exact images are tricks being played on it, so it can safely ignore it.

No. Learning that the "exact images" presented here are tricks would not be a solution to the problem revealed by this study. The goal in any form of machine learning is software that can effectively extrapolate beyond the training set.

What's the story?

Once you understand the problem, you will see what the story is.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

Capt.Albatross Re:Also... (125 comments)

If the network was trained to always return a "best match" then it's working correctly. To return "no image", it would need to be trained to be able to return that, just like humans are given feedback when there is no image.

It seems highly unlikely that such an elementary mistake was made: "Clune used one of the best DNNs, called AlexNet, created by researchers at the University of Toronto, Canada, in 2012 – its performance is so impressive that Google hired them last year."

The fact that the net returns a confidence level implies that it does have a way to return a result of 'not recognized'.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

Capt.Albatross Re:Training classifiers require "rejectable" sampl (125 comments)

The DNN examples were apparently trained to discriminate between a members of a labeled set. This only works when you have already cleaned up the input stream (a priori) and guarantee that the image must be an example of one of the classes.

These classifiers were not trained on samples from outside the target set.

This is not some network hastily trained by people who are ignorant of a very basic and long-known problem: "Clune used one of the best DNNs, called AlexNet, created by researchers at the University of Toronto, Canada, in 2012 – its performance is so impressive that Google hired them last year." From a paper by the developers of AlexNet: "To reduce overfitting in the globally connected layers we employed a new regularization method that proved to be very effective."

It does not seem plausible that this result can be explained away as an elementary mistake.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

Capt.Albatross Re:Also... (125 comments)

Nothing wrong with being wrong with confidence. Sounds like the majority of humanity the majority of the time.

Right, and it has created a great deal of misery throughout human history. Just because it is prevalent does not mean it is not a problem.

More specifically, the overconfidence displayed by the networks here should lead to a corresponding skepticism, in a rational observer, to the notion that they have cracked the image recognition problem.

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

Capt.Albatross Re:seems a lot like human vision to me (125 comments)

i even feel like i did sort of see a peacock in the one random image labeled peacock.

I know what you mean, but did you see a peacock before you read the label?

2 days ago
top

Research Highlights How AI Sees and How It Knows What It's Looking At

Capt.Albatross Re:seems a lot like human vision to me (125 comments)

I think it was fairly clear what was going on, the neural networks latch on to conditions that are necessary but not sufficient because they found common characteristics of real images but never got any negative feedback.

You seem to be suggesting that it is 'simply' a case of overfitting, but overfitting is a problem that has been recognized for some time. I don't doubt that the developers of these networks have thought long and hard about the issue, so this study suggests that it is a hard and as-yet unsolved problem in this domain.

One thing that humans use, but which these systems do not seem to have developed, is a fairly consistent theory of reality. Combine this with analytical reasoning, and we can take a difficult image and either work out what is being depicted, or realize that we are not succeeding in doing so.

2 days ago
top

Study of Massive Preprint Archive Hints At the Geography of Plagiarism

Capt.Albatross Re:who cares about plagiarism (53 comments)

So you are saying that the only reason that people do anything is for recognition or money?

Are you?

No, clearly not - that would be an unjustified extrapolation of an unwarranted generalization of a simplistic reading of the point under discussion.

about a week ago
top

Why Apple, Google, and FB Have Their Own Programming Languages

Capt.Albatross Missing the Point (161 comments)

While Mr Rosenberg claims that Go is distinguished by its approach to concurrency, his section 'The Essence of Go' is almost entirely devoted to the trivia of braces and semicolons. Yo won't learn anything about Go's approach to concurrency here.

about two weeks ago
top

Behind Apple's Sapphire Screen Debacle

Capt.Albatross Re:Then don't sign the contract (189 comments)

This isn't uncommon in industry (it's also not the normal way of things). If we want to to be certain that a supplier builds something the right way, we might specify every detail of the tooling, and sometimes buy it and install it ourselves.

I think the fact that Apple did not indicates that it did not think there was much chance of success, and was not, by then, expecting (or even much hoping) to ship with a sapphire screen.

about three weeks ago
top

Behind Apple's Sapphire Screen Debacle

Capt.Albatross Re:Then don't sign the contract (189 comments)

My guess is that at some point Apple decided the new manufacturing technology was unlikely to work in their timescale, and was not going to make its plans dependent on it (I imagine this was shortly before it backed out of acquiring the manufacturing equipment.) At this point, GT became the party desperately seeking a deal, and Apple effectively said 'show us, and we will consider it.'

about three weeks ago
top

A Toolbox That Helps Keep You From Losing Tools (Video)

Capt.Albatross Re:And this concept is standard in Aviation (82 comments)

But I watched the video, and then I thought, this for $35k? Intel? No, they must be more concerned with the work not getting done.... I hope.

I believe some of that chip-making equipment is pretty expensive - I am surprised that its protection was not given as the primary benefit. Perhaps they already have a solution for that environment, and this is for basic facilities maintenance.

about three weeks ago
top

Alva Noe: Don't Worry About the Singularity, We Can't Even Copy an Amoeba

Capt.Albatross Re:writer doesn't get jeopardy, or much of anythin (455 comments)

This is a level of manipulation of time (causality) which is completely out of reach of the most powerful computer.

Why exactly ? Computers can also run simulations.

The child not only manipulates time as an abstract concept, but also demonstrates a theory of mind. AFAIK, no simulation has yet achieved a theory of mind.

about three weeks ago
top

Google Maps Crunches Data, Tells You When To Drive On Thanksgiving

Capt.Albatross Re:So--don't travel during rush hour, then? (62 comments)

Traffic on Thanksgiving itself is also light, according to the data.

Another important revelation!

The data apparently hasn't heard of the Jersey Turnpike - probably because of its laser-like focus on cities.

about a month ago
top

Your Incompetent Boss Is Making You Unhappy

Capt.Albatross The worst place to have a bad boss (204 comments)

If you think it is unpleasant to have an incompetent boss at work, spare a thought for all the soldiers in WW1 whose bosses thought massive frontal assaults were the way to win.
 

about a month ago
top

Duke: No Mercy For CS 201 Cheaters Who Don't Turn Selves In By Wednesday

Capt.Albatross Re:Or just practicing for an actual job (320 comments)

Just out of curiosity are there any professional programmers out there who don't regularly copy functions from the Internet?

Part of being a contemporary coder is making use of available code. Libraries of functions are "other people's code". Languages are other people's code. Etc. it's all about other people's code.

This defense always comes up when cheating is the issue, and it is always wrong. The purpose of an examination is to determine if an individual understands the subject matter, and no argument how programming is done in practice alters the fact that a cheater has failed to demonstrate that he understands the subject matter (and has, in fact, provided good evidence that he does not.)

Furthermore, one should not be spending the time and money a degree from Duke costs, just to be a cut-and-paste coder - a semester at a community college should be enough, if you can't teach yourself. The purpose of a university education is to develop a thorough understanding of the subject matter in order to become the person who finds solutions to problems, rather than the people who copy them.

about a month ago
top

The Students Who Feel They Have the Right To Cheat

Capt.Albatross Just cheating themselves (438 comments)

Cheating only harms the cheater...

There is one other group harmed, and quite seriously, by widespread cheating: those who have worked hard and honestly for the best diploma they can both achieve and afford, but see it devalued to worthlessness because too many holders of the same diploma are cheaters, and incompetent.

about a month ago
top

French Health Watchdog: 3D Viewing May Damage Eyesight In Children

Capt.Albatross Re:I'm not a scientist... (99 comments)

According to Anses, the process of assimilating a three-dimensional effect requires the eyes to look at images in two different places at the same time before the brain translates it as one image.

Isn't that how normal vision works anyway?

In normal vision, we look at the same place from two slightly different directions. Furthermore, it is well-established that the neural 'wiring' for assimilating these two views into a single stereo image develops during childhood, in response to the stimuli. (I am not so sure about this, but I think this is also true for the wiring that controls the eye muscles and therefore the convergence of vision.) I am not a biologist, but I think there are grounds for concern here.

about a month and a half ago

Submissions

top

An Online Game of Skill, Played for Money

Capt.Albatross Capt.Albatross writes  |  about three weeks ago

Capt.Albatross (1301561) writes "Jason Rohrer, a game developer with an artistic flair (Passage, Sleep is Death) is developing a new game, Cordial Minuet (an anagram of 'demonic ritual'). It is a two-person game of skill, to be played online for money. Rohrer believes that, as a game of skill, it avoids falling foul of U.S. gambling legislation. Emanuel Maiberg's interview of Rohrer discusses the game play, Rohrer's steps to avoid legal problems while monetizing it, and whether games of skill avoid the ethical problems of gambling."
top

Yosemite Wi-Fi Problems

Capt.Albatross Capt.Albatross writes  |  about 2 months ago

Capt.Albatross (1301561) writes "Sophos' Naked Security blog is reporting that some users of Apple's OSX 10.10 (Yosemite) operating system are having problems with wi-fi: "Your network works fine for a while, typically between about 30 seconds and five minutes, and then fairly abruptly begins to suffer almost total traffic loss. The network shows up as active, and low-level packets such as PINGs can be sent and received as normal. But traffic such as UDP and TCP just doesn't get through."

Apple's own Support Community has much discussion, and some proposed workarounds, but no definite explanation or solution appears to have emerged yet."
top

Death Valley's Sailing Stones Caught in the Act

Capt.Albatross Capt.Albatross writes  |  about 4 months ago

Capt.Albatross (1301561) writes "The flat surface of the Racetrack Playa in Death Valley is littered with rocks, some weighing hundreds of kilograms, each at the end of a track indicating that it has somehow slid across the surface. The mechanism behind this has been the subject of much speculation but little evidence, until a trio of scientists caught them in action with cameras and GPS."
Link to Original Source
top

A Map of Publicly-Funded Creationism Teaching

Capt.Albatross Capt.Albatross writes  |  about a year ago

Capt.Albatross (1301561) writes "At Slate, Chris Kirk presents a map of schools in the USA that both receive public funding and teach creationism. It also shows public schools in those states where they are allowed to teach creationism (without necessarily implying that creationism is taught in all public schools of those states). There is a brief discussion of the regulations in those states where this occurs, but the amounts involved are not discussed.
 "
top

A Diagnosis for Healthcare.gov

Capt.Albatross Capt.Albatross writes  |  about a year ago

Capt.Albatross (1301561) writes "At Slate, David Auerbach reports on Thursday's hearing concerning the healthcare.gov debacle. It was "a spectacle of tech illiteracy and buck-passing", he says, which may not elicit much surprise around here. He is particularly scornful of the contractors' obsession with checking off milestones rather than with delivering something that works, their willingness to call something 'done' before having tested it, and their apparent obliviousness to how incompetent this shows them to be."
top

Twitter Buzz as an Election Predictor

Capt.Albatross Capt.Albatross writes  |  about a year ago

Capt.Albatross (1301561) writes "A study presented at the American Sociological Association's annual meeting suggests that simply comparing the frequency with which the candidates' names are mentioned in tweets can predict the result of elections almost as well as conventional polls, even without considering the sentiment (for or against the named candidate) of the messages. Furthermore, the correlation seems strongest in close elections.

Additional commentary can be found at the Wall Street Journal and from Indiana University."
top

Thorium Fuel has Proliferation Risk

Capt.Albatross Capt.Albatross writes  |  about 2 years ago

Capt.Albatross (1301561) writes "Thorium has attracted interest as a potentially safer fuel for nuclear power generation. In part, this has been because of the absence of a route to nuclear weapons, but a group of British scientists have identified a path that leads to uranium-233 via protactinium-233 from irradiated thorium. The protactinium separation could possibly be done with standard lab equipment, which would allow it to be done covertly, and deliver the minimum of U233 required for a weapon in less than a year.

The full article is in Nature, paywalled."

Link to Original Source
top

Promoting Arithmetic and Algebra by Example

Capt.Albatross Capt.Albatross writes  |  more than 2 years ago

Capt.Albatross (1301561) writes "A couple of months ago, the New York Times published political scientist Andrew Hacker's opinion that teaching algebra is harmful. Today, it has followed up with an article that is clearly intended to indicate the usefulness of basic mathematics by suggesting useful exercises in a variety of 'real-world' topics. While the starter questions in each topic involve formula evaluation rather than symbolic manipulation, the follow-up questions invite readers to delve more deeply.

The value of mathematics education has been a recurring issue in Slashdot."
top

Is Algebra Necessary? -- New York Times

Capt.Albatross Capt.Albatross writes  |  more than 2 years ago

Capt.Albatross (1301561) writes "Andrew Hacker, a professor of Political Science at the City University of New York and author of 'Higher Education? How Colleges Are Wasting Our Money and Failing Our Kids — and What We Can Do About It', attempts to answer this question in the negative in today's New York Times Sunday Review [registration may be required].

His primary claim is that mathematics requirements are prematurely and unreasonably limiting the level of education available to otherwise capable students ."
top

Righthaven Redux... With a Difference

Capt.Albatross Capt.Albatross writes  |  more than 2 years ago

Capt.Albatross (1301561) writes "At Boing Boing, Rob Beschizza reports that, in an act of delicious irony, Swiss ISP Ort Cloud [sic] has acquired Righthaven's domain name and has relaunched Righthaven.com as a web hosting service diametrically opposed to the practices of its original owner, a notorious but ultimately unsuccessful copyright troll. The new owners, in partnership with first amendment lawyer Marc Randazza (who was instrumental in the original Rigthhaven's demise), promise "infrajuridsictional infrastructure" — uptime that would require international cooperation to bring down. "Frivolous plaintiffs will find little comfort here" says Ort Cloud's Stefan Thalberg.

The domain name became available in a court-ordered auction of Righthaven LLC's assets, to pay its creditors."

Link to Original Source
top

To Learn, Test Yourself

Capt.Albatross Capt.Albatross writes  |  more than 3 years ago

Capt.Albatross (1301561) writes "The New York Times summarizes a paper published online by Science:

Taking a test is not just a passive mechanism for assessing how much people know, according to new research. It actually helps people learn, and it works better than a number of other studying techniques.

"

Link to Original Source

Journals

Capt.Albatross has no journal entries.

Slashdot Login

Need an Account?

Forgot your password?