Messages

1

General Discussion / Re: THIS AND THAT CONCEPTION IN HAL

« on: May 08, 2014, 06:01:46 pm »

ex: This is a pencil.
The subject pronoun 'this' is coupled to a physical object; therefore, assume that the speaker
is presenting or pointing to the (local) object.  In my experiments, a batch file calls an image capture,
and the image and the 'complement' are stored together as: 'seen_xxxxxxxx.jpg be @pencil'

ex: (It is raining.) 
      This/that/it is true.
The subject pronoun is coupled to an abstract complement, so the subject is not a thing but an action.
That action is the last mentioned verb.  (In this case stored the same as a 'yes'; ie, increase the trust
of the verb in: 'now rain progressive'

ex: That is the moon.
The subject pronoun 'that' is coupled to a physical object; so, assume that the speaker is
pointing to a distal object.  (My experiment can't image process to extract the pointer.)

2

General Discussion / Re: Self Awareness

« on: May 08, 2014, 05:39:59 pm »

OTC sounds a bit 'off the handle', but ART sounds sane.  I usually sound inane:
To reach self awareness:
1. remember what you did.
2. be able to observe parameters of yourself (battery level, volume level, directories, networks ...)
You are now self aware!  If that doesn't impress OTC:
3. have assigned goals.
4. allocate a portion of your memory for hypothetical events.
5. generate random tasks in that 'virtual world'.
6. predict the outcomes of those tasks (based on experience in the real world).
7. assess the effect on your goals.
8. If the assessment is positive, make the tasks into real commands; but don't be short sighted!

3

General Discussion / Re: Interesting A.I. article in The New Yorker...

« on: October 30, 2013, 01:08:08 pm »

Thank you, Zaphod,  (and say hi to Slarty..)
  The Winograd Schema test is more scientific than the Turing test.
ex: Sally scolded Mary because she was angry/clumsy.  She is Sally/Mary.
The antecedant of 'she' can't be determined until an association is made:
1. anger may cause one to scold.
2. being clumsy may cause one to be scolded.
Making the association is hard, not because the info isn't findable, but
because recursion (parallelism) is needed.  The brain finds both possible
antecedants and processes the rest of the sentence, and then ranks
the plausibility of the outcomes.  Then it must discard the losing scenarios.
That is hard (messy)!
In many cases a statistical association might find the answer (IBM Watson),
but only processing of all the possibilities will guarantee a correct (human) understanding.
   

4

Programming using the Ultra Hal Brain Editor / Re: One of these is not like the other.

« on: May 08, 2013, 02:41:09 pm »

'All' and 'every' are universal quantifiers, but there is a tendency to
pluralize the complements with 'all'.  For example: 'All cats have tails.'
This gives less info than 'Every cat has a tail.'  You could say: 'All cats have a tail.',
but that gets confused with all cats sharing one tail.  I prefer storing universals
in the singular form (every).  When you mention 'some cats' or 'black cats', I still
refer to the universal 'cat', but store the data as rules.  For example:
'Black cats are scary' can be stored as:
every cat
may[be] black
if cat [be] black
[then] [be] scary

  The biggest challenge is knowing whether a statement is universal (defining) or an instance,
since we so often omit the qualifiers:
A cat is a mammal. (defining)
Cats eat mice. (defining)
A cat is eating. (an instance because the verb is not infinite)
Cats are on the bed. (instances because locations are not universal)

5

General Discussion / Re: a plug in idea : what day is this ?

« on: May 08, 2013, 02:05:00 pm »

  I'll repeat myself since it pertains to this thread.  How my brain handles Time:
My internal dictionary flags 'time' words with a code; for example '.t'
Thus: time.t, day.t, month.t, hour.t ...
When I get a question such as 'what month is it?', the time code on 'month' tells me
that 'it' refers to time, and not the last mentioned genderless thing; that it, the pronoun
is not in the 3rd person - it is in the 4th person (think 4th dimension).
  Pronouns that are not in the 3rd person (1,2,4) have special antecedants that point to
current values.  I call the antecedant of time 'now'.  {3rd person antecedants can be
retrieved by finding the latest mentioned thing with the proper gender}
  I also know that:
'today is [at] now'
'tomorrow is [at] the day after today'
'Tuesday is the day after Monday'
...
 
 

6

Programming using the Ultra Hal Brain Editor / Re: learn from text

« on: January 31, 2013, 02:17:50 pm »

Art, thanks for the response.
  So, what syntax should I use to get HAL to:
- read a text file   (not just open a file, but read it as though someone was speaking)

- read a text file containing questions

- write (talk) into a text file     (with control to start and stop doing so)

If this is already covered in the documentation, please point me.
(I'm not so much lazy as old, esp my eyes)

7

Ultra Hal 7.0 / Re: I WILL PAY FOR PROGRAM PURPOSE.

« on: January 07, 2013, 11:56:01 am »

  It should be a goal of the programmers not to put any 'English' in EXE or DLL files.
My own experiments use secondary dictionaries for 'articulations'. 
The Lexicon dictionary is mostly for handling input, so it can contain foreign words.
Output requires non-lexical articulation which is language specific, so different files
are referenced based on the majority language of lexical words in the input.
The exe code should use internal code words which get translated thru the artic dict.
It is a pain putting gender on genderless things in the lexicon...
  Since Graphics Engines were mentioned, I'd like to plug the Distance Estimation Method
for rapid (real time!) ray tracing.  The concept is to distort the space so that the shapes
become simple enough that the ray can progress by leaps and bounds without fear of
missing a part of the shape.  Some video examples at YTALINFLUSA.
   

8

General Discussion / Re: Hal to recognize colors plug in lightspeed idea 11-2-2012

« on: November 16, 2012, 04:57:48 pm »

Sorry, I was speaking pseudo-code as I am not a HALbrain programmer.
Can you change: if InStr(1,OriginalSentence,"tea green")>0 then ...
to something like: //C code
string color; //global
bool ColorIn(string InS) {
  bool fnd=false;
  color="";
  if(!fnd) {color="red"; fnd=(InStr(1,InS,color));}
  ...
  if(!fnd) {color="blue"; fnd=(InStr(1,InS,color));}
  return fnd
}

and then:
if(ColorIn(OriginalSentence)) then {//process using the var color

  //the response string could be broken into smaller routines concatenated:
  GetResponse = OptS("Yes, ")+OptS("I see that ")+"it is a "+color+OptS(" color");
 
OptS would return its string 50% of the time.
Other routines could be, for instance:
string Agree() {
  case ..brain.Random(2) {
  1: return("yes");
  2: return("definitely");
 ...
}

I know this isn't a full answer, but I wanted you to think about deeper methods.  Best Wishes..

9

Ultra Hal Assistant File Sharing Area / Re: HalVisionX Beta

« on: November 15, 2012, 05:22:46 pm »

I answer my own question:
  I downloaded WebCamImageSave
command line usage: webcamimagesave /capture /Filename "this.jpg"

10

General Discussion / Re: Hal to recognize colors plug in lightspeed idea 11-2-2012

« on: November 15, 2012, 03:48:50 pm »

Lightspeed (and OTC2005),
  In the example where the input string includes "tea green", the output comes from a list:
 - "Yes, it is tea green"
 - "I see that it is tea green"
 - etc
You would have a separate output list for each color! (also refer to your temperature idea)
Why doesn't "tea green" become a variable, with just one output list.
This list could include decisions based on inferred variables (such as "warm color").
Thus you increase the dimensions of the matrix (the n-trix) to reduce the storage (and
possibly increase the intelligence.)

 

11

Ultra Hal Assistant File Sharing Area / Re: HalVisionX Beta

« on: November 15, 2012, 03:37:01 pm »

Sorry to be a bother, but I live in the wilds of MN and don't have regular internet service.
Can YawCam be used in offline mode?  I can't figure out how.
  For years, computer interfaces have moved towards GUI and away from text.
Now we are giving programs the ability of using English and we've lost apps that can
do anything without a mouse and eyeballs! 
  For example, the Logitech camera site has bloated software with circuitous downloading
and comments about crapware (not from me).  How I long for a simple camera that I can
call from the command line to just take a picture!  Can anyone help me?
Again, I'm sorry I can't afford more net time to do my own searching..

12

Ultra Hal Assistant File Sharing Area / Re: HalVisionX Beta

« on: November 13, 2012, 03:41:47 pm »

I wrote a minimal C# (.net) program Show.exe (16K) for showing a picture.
I wrote a Pascal program FindFile.exe for making a list of files (FndFile.TXT) that match a search_spec
  ex: FindFile cat*.jpg
and a program Choose for picking a random list entry (to Choice.TXT)
'Show me a cat' causes (example):
   FindFile cat*.jpg cat*.gif
   Choose FndFile.TXT
   read Choice.TXT
   Show cat_in_tree.jpg

This is crude but it works.  The weakest part is 'Choose' since I can't infer on external data
(can't use 'the', 'other', etc.)

Webcams use proprietary interfaces.  I haven't found one that will just take a picture from
the command line, so I can say 'See cat_eating_mouse.jpg'

13

General Discussion / Re: Hal to recognize colors plug in lightspeed idea 11-2-2012

« on: November 13, 2012, 03:20:34 pm »

LightSpeed,  you are going to kill HAL with so many canned phrases.
  Try to think in more dimensions: You have a matrix of lists - try making those lists into matrices..

14

General Discussion / Re: Hal to recognize colors plug in lightspeed idea 11-2-2012

« on: November 06, 2012, 01:46:24 pm »

How about vision and colors combined?
ex:  Hal, what color is this?
pronoun 'this', verb 'be'  triggers camera {execute 'SEE'}
SEE COLOR: do a variant of:
 - isolate fovea (central part of image)
 - remove noise (pixels not similar to surrounding pixels)
 - if multiple clusters then repeat isolate fovea (zoom in)
 - average the RGB values
 - find that RGB value in a list of color names and ranges
 - say 'that is ' _color

OK, maybe you should get the camera working first... (before HAL gets a job at the paint store)

15

Ultra Hal Assistant File Sharing Area / Re: HalVisionX Beta

« on: October 26, 2012, 05:48:50 pm »

May I suggest a vision use:
 Use the phrase 'this is' to trigger camera capture.
ex:  This is my cat.  {Save image 'my_cat.jpg'}
 Use the phrase 'show me|us' to trigger slide show.
ex: Show me my cat.   {View image 'my_cat.jpg'}

With some more work, it could handle:
ex: Show me a cat.  {View image '*cat.jpg'}