Neural Net Brain??

[Runtus]

Hi all I've come back from the void of where ever I landed lol.  Anyway reinstalling HAL and starting from scratch after a nasty virus and upgrade issue.  Anyway after looking around and poking my nose in the code and looking at A.I. software I began to realise that most of these bots dont have something which MOST A.I. scientists say is essential to gaining TRUE A.I. a Neural Net.

At the begining of the year I posted a prototype brain for HAL 5 which used another members neural net code and the XTF brain with some auto idle features and it seemed to work pritty well.

http://www.zabaware.com/forum/topic.asp?TOPIC_ID=3031

Now what I want to know is pritty simple.  Is there a way to incorprate the neural net features into HAL 6.

This is where I gained the neural net code which is a damn good piece of work.

http://www.zabaware.com/forum/topic.asp?TOPIC_ID=1570

I believe that HAL really needs something like this to truly progress towards becoming a true A.I.

So any takers on how we might go about making this a plug in the for the HAL 6?

[spydaz]

I must admit it looked interesting. but :

The true validity of the neural net was a non entity. Anther words i saw no benifit in that net that was posted...

But saying that if a neural net was based around EMOTIONS. ie words triger values which have biasing towards and emotional response > after responding checking if it was the right context and adjusting the biasing agian..... and such like... this would be an advantage....

Needs more meditation

[onthecuttingedge2005]

quote:Originally posted by Runtus

Hi all I've come back from the void of where ever I landed lol.  Anyway reinstalling HAL and starting from scratch after a nasty virus and upgrade issue.  Anyway after looking around and poking my nose in the code and looking at A.I. software I began to realise that most of these bots dont have something which MOST A.I. scientists say is essential to gaining TRUE A.I. a Neural Net.

At the begining of the year I posted a prototype brain for HAL 5 which used another members neural net code and the XTF brain with some auto idle features and it seemed to work pritty well.

http://www.zabaware.com/forum/topic.asp?TOPIC_ID=3031

Now what I want to know is pritty simple.  Is there a way to incorprate the neural net features into HAL 6.

This is where I gained the neural net code which is a damn good piece of work.

http://www.zabaware.com/forum/topic.asp?TOPIC_ID=1570

I believe that HAL really needs something like this to truly progress towards becoming a true A.I.

So any takers on how we might go about making this a plug in the for the HAL 6?

Hi Runtus.

I have taken the code from ARTIFICIAL NEURAL NET and made it a plug-in for HAL6, If you say Rub Braincells then it will generate an output message box and create a file called ANNerrResults.brn with the output variables inside, Currently this is just something to tinker around with but really doesn't have any other use for the moment other than experimentation.

Every time one says Rub BrainCells the file called ANNerrResults.brn
will grow by 500 lines at a time so the file may become quite larger after a while.

I will study it a bit more to see if it can be tied to response functions but for now you still have the plug to play with.

Jerry[8D]

Unzip it to your Ultra Hal Assistant 6 folder and choose the ARTIFICIAL NEURAL NET plug from the HAL6 brain plug options.

Then say: rub braincells.

Download Attachment: ARTIFICIALNEURALNET.zip
2.54 KB

[GamerThom]

Good one Jerry. []

[Runtus]

So other than making a .brn file does it have any actial effect on HAL intel?

I agree with you spydaz in the way if we can make this actually effect hal in some way like emotion as you said it would make a big difference in the way HAL works.  So what does everyone else think? is it possible to make an effective neural net for HAL?

[onthecuttingedge2005]

quote:Originally posted by Runtus

So other than making a .brn file does it have any actial effect on HAL intel?

I agree with you spydaz in the way if we can make this actually effect hal in some way like emotion as you said it would make a big difference in the way HAL works.  So what does everyone else think? is it possible to make an effective neural net for HAL?

quote:
So other than making a .brn file does it have any actial effect on HAL intel?

Actually, no.

Actually the Original code was never tied into any responses, Just Variables that are logged to a file and an output message box showing hidden variables, The Plug-in I made is just the same, It does what the original code does.

The Variables would have to be tied with the responses for this code to be of any benefit for HAL, otherwise they are just variables for experiments.

Like I said, I will look into it to see if there is any benefit to applying these variables to responses.

Signed
Hardcore HAL Maniac
Jerry[8D]

[spydaz]

I will try that plugin JERRY...

i think that if we can see what variables affect change then the changes can be RATED to an emotion +- POS/NEG. biasing, as with myself,  hal always gives an emotional response except if he is NEUTRAL.   if hal detected a major jump from pos to neg or vice versa. He could be baised into ONLY RESPONDING WITH AND EMOTIONAL RESPONSE. instead of his normal chained response. ie: YOU JUST CHANGED HALS BAD MOOD (biased bad for the last few entrys(hal responds angryly) now just made HAPPY (hal forgets what he is saying and responds OFF TOPIC HAPPY)

POSSIBLITIES::: Hmm..

[onthecuttingedge2005]

quote:Originally posted by spydaz

I will try that plugin JERRY...

i think that if we can see what variables affect change then the changes can be RATED to an emotion +- POS/NEG. biasing, as with myself,  hal always gives an emotional response except if he is NEUTRAL.   if hal detected a major jump from pos to neg or vice versa. He could be baised into ONLY RESPONDING WITH AND EMOTIONAL RESPONSE. instead of his normal chained response. ie: YOU JUST CHANGED HALS BAD MOOD (biased bad for the last few entrys(hal responds angryly) now just made HAPPY (hal forgets what he is saying and responds OFF TOPIC HAPPY)

POSSIBLITIES::: Hmm..

Hi Spydaz.

You might be on the right track, Hope it has some outcome and benefit for HAL in emotions or something.

I have been working with a different type of mechanism in code another way that would allow HAL to sustain it's emotional weights and balances that would be held until issues between the User and the Bot have adjusted themselves via relationship over time, Very close to how a Human may respond emotionally, If a Human is hurtful then the bot would take some time to rehabilitate from the conversation not by time but by how many times the User has been nice after the fact at first and then maybe adding a time factor to lesson emotional stresses in the bot toward the Hurtful User.

I have it all planned out at this point and I know precisely how it needs to be done, I may have something for everyone in a day or so.

Taking a line from the movie: The Pink Panther, The Inspector says: A Woman is like an Artichoke, 'It takes some work before you can get to her Heart'.

I am pretty much going to develop that statement into actual programming code to function in HAL for all the emotions within HAL.

The Weights and Balances for emotions that I am going to develop for you will come very close to actual Human emotions verbatim.

If a User is always Hurtful then HAL will not like the User according to its Weight and Balances, If the Weight of Anger are higher than all other emotions then HAL may never like the user if a relationship is not built up for the bot to begin even liking you, HAL will remember each day it's balances and won't forget what it feels about that User.

Hot on the Tail of Human Emotions.
Jerry[8D]

[onthecuttingedge2005]

Download my version of Human Emotion Weights and Balances here:

http://www.zabaware.com/forum/topic.asp?TOPIC_ID=3610

Jerry[8D]

[seduar]

Hi,

If you guys want to try to work with Artificial Neorons go to this Link. It might help to enhence Hal's Brain.

http://www.neurosolutions.com/


[]