Page 4 of 8 First 12345678 Last
  1. #31
    Christmas Spirit's Avatar
    Join Date
    Jun 2005
    Location
    Sinsinnatti Oh Hi Ho
    Posts
    13,829
    Style
    all things in Moderation
    --
    Hell yeah! Hell no!
    Quote Originally Posted by BKR View Post
    No no that's not it the artificial intelligence if it ever came today would not be deciding who gets to have weapons or not.
    So you are saying I should get the shotgun I want before I take anymore head trauma ...
    Quote Originally Posted by ghost55 View Post
    Violence is pretty uncommon in clubs in this area, and the dude didn't seem particularly hostile up until the moment he slapped me.
    I don't mean to sound bitter, cold, or cruel, but I am, so that's how it comes out.
    BILL HICKS,
    1961-1994

    Quote Originally Posted by WFMurphyPhD View Post
    Slamming the man in the bottom position from time to time keeps everybody on their toes and discourages butt scooting stupidity.

  2. #32
    BKR's Avatar
    Join Date
    Jun 2009
    Location
    Bonners Ferry, Idaho
    Posts
    16,782
    Style
    Kodokan Judo/BJJ
    --
    Hell yeah! Hell no!
    Quote Originally Posted by BackFistMonkey View Post
    So you are saying I should get the shotgun I want before I take anymore head trauma ...
    Judging from that post I'd say you've already had plenty of head trauma.

    I know I've had plenty of head trauma...
    Falling for Judo since 1980

    "You are wrong. Why? Because you move like a pregnant yak and talk like a spazzing 'I train UFC' noob." -DCS

    "The best part of getting you worked up is your backpack full of irony and lies." -It Is Fake

    "Banning BKR is like kicking a Quokka. It's foolishness of the first order." - Raycetpfl

  3. #33
    Nutcracker, sweet's Avatar
    Join Date
    Apr 2010
    Location
    KAUS
    Posts
    5,023
    Style
    BJJ
    6
    Hell yeah! Hell no!
    Reviewing, W. Rabbit and I aren't disagreeing on the viability of the tech, just on when it will be available to solve the problem of gun violence.

    When discussing "learning," with respect to neural nets, it's sort of a backwards thing. You tell the network "here's the result, and here are some of the ways you can get that result." Then, you run through test data to ensure that the desired result is output. If it isn't, then you go back, add or remove or change the ways the computer can get that result, and try again. It's an oversimplification, but that's the basic, high-level approach to date.

    AZG is significant, because instead presenting the end result and some ways to get the result, it was presented with a set of meta-instructions (the rules of the game), and the tech was able to derive the test data by running millions of simulations, which were then run in a feedback loop. It "taught," itself in a little under a seven weeks to be better than its predecessor. At one thing, with a finite set of very simple rules. Again, W. Rabbit was keen to point out that "It's best effort, YMMV."

    The problems remaining are
    a) teaching the computer what a mass murderer looks like
    b) feeding the computer with enough data

    There are some attributes which describe all mass murderers (most keenly that they've actually done the murders), and more attributes that describe most. But, what is not known is how many people in the world that will NEVER do a mass murder fit those same attributes, or at least enough attributes to recommend investigation. We also do not know if there are other attributes that are shared by all or most mass murderers. Therefore, we do not have a discrete set of "rules," which the computer can learn. In other words, we are giving the computer a deck of cards and telling it to make pairs, but we're not really sure if the game is Canasta or Gin or Bridge or Spades or Poker or Go-Fish, or something else entirely. The computer will never be able to win, because some "wins," will conflict with earlier "wins."

    The bigger problem is that once you get together the rules of the game, you need the cards! The cards, in this case, are going to be medical records, financial records, school records, military records, essays and other writings, drawings and other artwork, youtube videos watched or produced, social media interactions, website viewing history, magazine subscriptions, library records, personal and professional associations, locations, dates, religious leanings, political leanings, are you getting the point? We could go on forever listing just the inputs to this system. In short, everything that can be known about a single person. Then, do that for millions of people.

    All of that to get to the computer-generated supposition of "maybe," which then needs human due-diligence to occur.

    I don't think we're going to see such a thing anytime soon.

  4. #34
    Christmas Spirit's Avatar
    Join Date
    Jun 2005
    Location
    Sinsinnatti Oh Hi Ho
    Posts
    13,829
    Style
    all things in Moderation
    2
    Hell yeah! Hell no!
    Wow that is a lot of data for the guberment to collect huh?!?!

    Corporations already collect more than people think. Cambridge Analytica, for example, claims to have already has collected up to 5,000 data points from literally everything said on Facebook and Twitter to voting records, phone numbers, property, income, addresses, to known associations and family via social media on every U.S. voter. So ... yeah ... it isn't that far away.

    Cambridge Analytica is a... well hell, just go to their site they don't hide it.
    https://cambridgeanalytica.org/
    Quote Originally Posted by website
    Data drives all we do.
    Cambridge Analytica uses data to change audience behavior. Visit our Commercial or Political divisions to see how we can help you.
    Quote Originally Posted by ghost55 View Post
    Violence is pretty uncommon in clubs in this area, and the dude didn't seem particularly hostile up until the moment he slapped me.
    I don't mean to sound bitter, cold, or cruel, but I am, so that's how it comes out.
    BILL HICKS,
    1961-1994

    Quote Originally Posted by WFMurphyPhD View Post
    Slamming the man in the bottom position from time to time keeps everybody on their toes and discourages butt scooting stupidity.

  5. #35

    Join Date
    Sep 2015
    Location
    Italy
    Posts
    439
    Style
    Judo, Nippon Kempo
    --
    Hell yeah! Hell no!
    IMHO the problem is that, contrary to what W.Rabbit says, there is not a good way to predict who is going to go crazy, and this doesn't depend from a lack of calculating power, but rather from the fact that human behaviour is too random.
    We can already profile people who are more likely to commit crimes, such as e.g. drunkhards or people who have already a criminal history, but this profiling works only in a very approximative way, even if you add in computers the most you get is some beefed up version of "background checks".

    Of course if we could know in advantage who is the bad guy everything would be easier, but in most cases people turn out to be bad guys in hindsight but weren't before.

    The problem I have with this idea (and with Trump comment that the recent shooting is a problem of a crazy guy tot a problem of guns) is that it assumes an existentialist idea of the "bad guy": someone who is "evil inside", and the bad deed is just a consequence of being evil inside.

    But apart from a few madmen people aren't really evil inside, everyone in some extreme or weird situation can go postal and do something very stupid like killing wife and sons and then suicide etc..
    So the idea that you can solve everything by just screening the bad apples before anything happens is wrong in principle IMHO.

  6. #36
    Christmas Spirit's Avatar
    Join Date
    Jun 2005
    Location
    Sinsinnatti Oh Hi Ho
    Posts
    13,829
    Style
    all things in Moderation
    --
    Hell yeah! Hell no!
    Quote Originally Posted by BKR View Post
    Judging from that post I'd say you've already had plenty of head trauma.

    I know I've had plenty of head trauma...
    I missed the "not" in your post. I thought you said they would be used to decide gun ownership... sadly I was sober then.

    * shrugs * reading iz hard.
    Quote Originally Posted by ghost55 View Post
    Violence is pretty uncommon in clubs in this area, and the dude didn't seem particularly hostile up until the moment he slapped me.
    I don't mean to sound bitter, cold, or cruel, but I am, so that's how it comes out.
    BILL HICKS,
    1961-1994

    Quote Originally Posted by WFMurphyPhD View Post
    Slamming the man in the bottom position from time to time keeps everybody on their toes and discourages butt scooting stupidity.

  7. #37
    Christmas Spirit's Avatar
    Join Date
    Jun 2005
    Location
    Sinsinnatti Oh Hi Ho
    Posts
    13,829
    Style
    all things in Moderation
    1
    Hell yeah! Hell no!
    Quote Originally Posted by MisterMR View Post
    IMHO the problem is that, contrary to what W.Rabbit says, there is not a good way to predict who is going to go crazy,
    I disagree I was recently turned down from participating in a study that was using Machine Learning and video clips to diagnose mental illness specifically depression because they were looking for a control group and I had been diagnosed before. It really isn't scifi or too far off. They are working on it right now.
    Last edited by Christmas Spirit; 11/11/2017 1:14am at . Reason: clarity
    Quote Originally Posted by ghost55 View Post
    Violence is pretty uncommon in clubs in this area, and the dude didn't seem particularly hostile up until the moment he slapped me.
    I don't mean to sound bitter, cold, or cruel, but I am, so that's how it comes out.
    BILL HICKS,
    1961-1994

    Quote Originally Posted by WFMurphyPhD View Post
    Slamming the man in the bottom position from time to time keeps everybody on their toes and discourages butt scooting stupidity.

  8. #38

    Join Date
    Sep 2015
    Location
    Italy
    Posts
    439
    Style
    Judo, Nippon Kempo
    1
    Hell yeah! Hell no!
    Quote Originally Posted by BackFistMonkey View Post
    I disagree I was recently turned down from participating in a study that was using Machine Learning and video clips to diagnose mental illness specifically depression because they were looking for a control group and I had been diagnosed before. It really isn't scifi or too far off. They are working on it right now.
    Well, you have been diagnosed before - you don't really need computers to know that someone who already had depression has a tendency to depression.

    My point is that "computers" won't help much because, with the data we already have, we can already "profile" people (and in fact we already do); computers won't help much about this: at best they will do a lot of cluster analysis.

    Let's say that with this cluster analysis we know that BFM[1] has a 0.3% chance to go crazy berserker, MisterMr has 1% (really high), and Submessenger has 0.1%.
    This is the kind of result you get from cluster analysis.

    So at what percentage do we set the limit? MisterMr clearly shouldn't, but BFM?

    We already have some profiling stats, when you add more computers we change the situation into knowing that:

    BFM has a 0.273% chance of going berserk, MisterMr has 0.972%, and Submessenger 0.156%.
    How does this change the situation? The main problem is that we don't know were to put the limit.

    If we put the limit at 1% we have american-style "free for all" firearms, if we put it at 0.001% we have european-style "restrictive" gun laws.

    [1] you should totally make some use of this:
    http://images3.dacw.co/unglued-b-f-m...-1.1.0&w=1800&

  9. #39
    Christmas Spirit's Avatar
    Join Date
    Jun 2005
    Location
    Sinsinnatti Oh Hi Ho
    Posts
    13,829
    Style
    all things in Moderation
    1
    Hell yeah! Hell no!
    Being able to judge mental states of people using computers almost instantly via body language and very subtle clues like: involuntary movements (eye movements, twitches, etc), blowing coal, member ship in Alt Right groups, truck nuts, and open carry is on the way, and is my only point.

    How it is going to be used and the implications are unknown to me.

    Quote Originally Posted by MisterMR View Post
    Well, you have been diagnosed before - you don't really need computers to know that someone who already had depression has a tendency to depression.

    My point is that "computers" won't help much because, with the data we already have, we can already "profile" people (and in fact we already do); computers won't help much about this: at best they will do a lot of cluster analysis.

    Let's say that with this cluster analysis we know that BFM[1] has a 0.3% chance to go crazy berserker, MisterMr has 1% (really high), and Submessenger has 0.1%.
    This is the kind of result you get from cluster analysis.

    So at what percentage do we set the limit? MisterMr clearly shouldn't, but BFM?

    We already have some profiling stats, when you add more computers we change the situation into knowing that:

    BFM has a 0.273% chance of going berserk, MisterMr has 0.972%, and Submessenger 0.156%.
    How does this change the situation? The main problem is that we don't know were to put the limit.

    If we put the limit at 1% we have american-style "free for all" firearms, if we put it at 0.001% we have european-style "restrictive" gun laws.

    [1] you should totally make some use of this:
    http://images3.dacw.co/unglued-b-f-m...-1.1.0&w=1800&
    some of this post is humor.
    Quote Originally Posted by ghost55 View Post
    Violence is pretty uncommon in clubs in this area, and the dude didn't seem particularly hostile up until the moment he slapped me.
    I don't mean to sound bitter, cold, or cruel, but I am, so that's how it comes out.
    BILL HICKS,
    1961-1994

    Quote Originally Posted by WFMurphyPhD View Post
    Slamming the man in the bottom position from time to time keeps everybody on their toes and discourages butt scooting stupidity.

  10. #40
    Ice Hole's Avatar
    Join Date
    May 2010
    Location
    Suburbia
    Posts
    11,827
    Style
    Mixed
    --
    Hell yeah! Hell no!
    Quote Originally Posted by submessenger View Post
    At a conservative estimate, it took 4 years. And, the tech is still mostly not accessible to civilians, 60 years on.
    Stop trolling. Any kid can make an atomic weapon in their basement. Getting the radioactive elements might be challenging, but that's what Libyan terrorists are for.

    Quote Originally Posted by submessenger View Post
    You're parading AGZ as the killer app, but all its power means exactly dick to me,
    I guess you're not a big Go player. Some people consider it one of the most advanced games mankind has ever developed. It's more complex than chess, as far as combinatorics.

    Quote Originally Posted by submessenger View Post
    Here's the point where you tell me that the machine will figure out the model on its own.
    Here's an alternate point. We're talking about pattern recognition, a relatively simple signal analysis problem. Don't drift into Cameronesque Sky-net fantasies like some are doing.

    Quote Originally Posted by submessenger View Post
    And, here's the point where I tell you that instead of breeding a machine to catch killers, they are building a machine that can beat your ass at surrounding a chunk of beads with a differently colored chunk of beads. Humans don't even understand human nature, I think it's a stretch that humans can design a machine that will understand human nature. See also Godel's theorems of incompleteness.
    You're way beyond the problem at hand that resulted in this thread. Again, come back to science fact and pragmatic, available, applied science. Look at Kelley's document trail and tell me his signature can't be quantified in simple terms.

    Go is simple in rules but complex in permutations, which is exactly the nature of "Find Kelley Before He Wastes a Church" problem.
    Last edited by Ice Hole; 11/12/2017 2:51am at .

Page 4 of 8 First 12345678 Last

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  

Log in

Log in