Results 1 to 13 of 13

Thread: Neural network

  1. #1
    Join Date
    Feb 2013
    Posts
    33
    Mentioned
    0 Post(s)
    Quoted
    17 Post(s)

    Default Neural network

    Hey,
    Wondering if anyone ever made a bot using neurals? I've been wondering for some time now, is rid a true neurals bot, a mouse recorder with evolution algorithm or just a variable (gauss?) mouse recorder?
    Wouldn't a neural bot make botting 100% undetectable?..

  2. #2
    Join Date
    Dec 2007
    Posts
    2,112
    Mentioned
    71 Post(s)
    Quoted
    580 Post(s)

    Default

    It wouldn't matter. Whilst I currently don't believe in AI. For something to be considered AI, it would have to pass the Turing test. Jagex's anti-cheat isn't a Turing test. Their anti-cheat doesn't work on the basis of how human-like a bot is. That's the beauty of RID's Genesis, he has effectively scammed a bunch of people into believing that biometrics makes a difference. Unlucky.

  3. #3
    Join Date
    Feb 2013
    Posts
    33
    Mentioned
    0 Post(s)
    Quoted
    17 Post(s)

    Default

    It always comes down to biometrics. By using color bots, you eliminate any possibility of client sided detection, and all that's left is just clicks and movement for jagex to detect.

  4. #4
    Join Date
    Sep 2014
    Location
    C:\Simba\
    Posts
    565
    Mentioned
    9 Post(s)
    Quoted
    71 Post(s)

    Default

    No, what's left is it storing all your actions. Where you click, in what order you click, in what order you perform your actions. Perhaps construct some sort of loop of repetitive play and make a check list of what you do, which can later be applied to other accounts running that script.

    If I make a concrete example of predictable behavior:

    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep

    Now you can guess quite well what's coming next.

    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep

    Now ofcourse you can have it do something else like every 17th run (randomly). But that will also be detectable in the long run.
    If anything RiD fails even harder at this. It's been programmed to manipulate your behavior as closely as it can, and unless you feed it your entire brain information or a year worth of data, it's not going to behave human like at all.

    Also stuff like clicking whilst moving, or detecting e.g. an Astral altar by turning down the camera whilst runinng every now and then, will look more human like.

    so tl;dr, I am sure they track behaviour in some way, but not in the sense of "SUPER DUPER PRO MOUSE MOEVEMENTS BWÄÄÄÄÄÄHHHH".

    Edit: Some people might TakeAShit before EatCereal, change up on the order depending on time, or heck even doing both contemporarily. The pattern of occurence is usually unpredictable and thus some like RiD or If (Random(17) = 0) Then ... Else ... Will fail hard at being human-like.
    Last edited by Joopi; 03-28-2017 at 11:56 AM.
    Feel free to ask me any questions, I will do my best to answer them!

    Previously known as YouPee.

  5. #5
    Join Date
    Feb 2013
    Posts
    33
    Mentioned
    0 Post(s)
    Quoted
    17 Post(s)

    Default

    Quote Originally Posted by Joopi View Post
    No, what's left is it storing all your actions. Where you click, in what order you click, in what order you perform your actions. Perhaps construct some sort of loop of repetitive play and make a check list of what you do, which can later be applied to other accounts running that script.

    If I make a concrete example of predictable behavior:

    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep
    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep

    Now you can guess quite well what's coming next.

    WakeUp->EatCereal->TakeAShit->WashHands->GotoSchool->GoHome->Sleep

    Now ofcourse you can have it do something else like every 17th run (randomly). But that will also be detectable in the long run.
    If anything RiD fails even harder at this. It's been programmed to manipulate your behavior as closely as it can, and unless you feed it your entire brain information or a year worth of data, it's not going to behave human like at all.

    Also stuff like clicking whilst moving, or detecting e.g. an Astral altar by turning down the camera whilst runinng every now and then, will look more human like.

    so tl;dr, I am sure they track behaviour in some way, but not in the sense of "SUPER DUPER PRO MOUSE MOEVEMENTS BWÄÄÄÄÄÄHHHH".

    Edit: Some people might TakeAShit before EatCereal, change up on the order depending on time, or heck even doing both contemporarily. The pattern of occurence is usually unpredictable and thus some like RiD or If (Random(17) = 0) Then ... Else ... Will fail hard at being human-like.
    What you're saying is true, when it comes down to flagging account (assuming they do that). But they cant ban based solely on repetitiveness - as far as I know they manually confirm each and every ban (except for suiciding where they have it done fully auto). Thats where (imo) biometrics come in play.

  6. #6
    Join Date
    May 2012
    Location
    Glorious Nippon
    Posts
    1,011
    Mentioned
    50 Post(s)
    Quoted
    505 Post(s)

    Default

    Quote Originally Posted by Bago47 View Post
    But they cant ban based solely on repetitiveness
    ?? Says who? They can ban anyone for any reason. It's their game.
    Legit players have been automatically banned by their system.
    It would be insane for them to manually check every ban. They don't have that kind of manpower.

  7. #7
    Join Date
    Sep 2008
    Location
    Not here.
    Posts
    5,422
    Mentioned
    13 Post(s)
    Quoted
    242 Post(s)

    Default

    Why would you use a neural net for something that could be done with a much more simple ML model?

  8. #8
    Join Date
    Dec 2007
    Posts
    2,112
    Mentioned
    71 Post(s)
    Quoted
    580 Post(s)

    Default

    Quote Originally Posted by Bago47 View Post
    It always comes down to biometrics. By using color bots, you eliminate any possibility of client sided detection, and all that's left is just clicks and movement for jagex to detect.
    No, there's so many other ways to detect botting. Biometrics means fuck all. Detecting for human-like movement is a waste of time.

    TIL No-one has ever been banned from using color bots.

    That shit happens too, just less frequently. Jagex don't focus us, we don't make enough of an impact for them to care. Majority of bans here are probably picked up through smart / manual / ip / maybe even injected input detection / pattern detection between similar bot methods ie if you run nats, legit or otherwise, your account will probs be flagged at minimum.

  9. #9
    Join Date
    Mar 2017
    Posts
    1
    Mentioned
    0 Post(s)
    Quoted
    0 Post(s)

    Default

    me too

  10. #10
    Join Date
    Apr 2017
    Posts
    12
    Mentioned
    0 Post(s)
    Quoted
    2 Post(s)

    Default

    This would be an incredible leap in botting but it is far too complex to apply I think.

  11. #11
    Join Date
    Apr 2017
    Location
    The Great White North
    Posts
    15
    Mentioned
    0 Post(s)
    Quoted
    1 Post(s)

    Default

    This interests me as well, for multiple applications (Unity). What neural nets have you worked with?

    You more than likely are going to need to be on linux, but there are programs for windows as well. I feel like teaching a bot to organically learn to play rs would be downtown bantown. But then again, maybe you could let it spend a lot of time learning on a private server. I actually don't even know what RSPS is or if it needs an internet connection/jagex can see it.

  12. #12
    Join Date
    May 2007
    Posts
    526
    Mentioned
    12 Post(s)
    Quoted
    109 Post(s)

    Default

    Quote Originally Posted by mrpcp View Post
    This would be an incredible leap in botting but it is far too complex to apply I think.
    Not at all, it's only complex if you don't understand what it means or how it works.

  13. #13
    Join Date
    Apr 2017
    Location
    The Great White North
    Posts
    15
    Mentioned
    0 Post(s)
    Quoted
    1 Post(s)

    Default

    Is anyone here interested in helping me configure a recurrent neural network to run on my machine? I believe I have a GPU capable of doing calculations in a reasonable time frame.

    I am attempting to do something similar to what is featured in this article: http://karpathy.github.io/2015/05/21/rnn-effectiveness/

    This is the git link for the download: https://github.com/karpathy/char-rnn

    It is a very interesting read to be sure. I would be remiss if I didn't pass it on, but let me know if I am breaking a rule as far as posting links goes.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

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