NND & AICI

  • 155 Replies
  • 51665 Views
*

Bragi

  • Trusty Member
  • ********
  • Replicant
  • *
  • 564
    • Neural network design blog
Re: NND & AICI
« Reply #90 on: March 16, 2011, 04:20:52 pm »
Yep, should be.

*

Bragi

  • Trusty Member
  • ********
  • Replicant
  • *
  • 564
    • Neural network design blog
Re: NND & AICI
« Reply #91 on: March 19, 2011, 04:00:26 pm »
I've been working on some screencasts to demo the designer. Here are the first 2:



*

DaveMorton

  • Trusty Member
  • ********
  • Replicant
  • *
  • 636
  • Safe, Reliable Insanity, Since 1961
    • Geek Cave Creations
Re: NND & AICI
« Reply #92 on: March 19, 2011, 04:20:39 pm »
(Privately, to Jan: Everyone else avert your eyes!) What software did you use to create those screencasts, Jan? (ok, everyone else can look now!)
Comforting the Disturbed, Disturbing the Comfortable
Chat with Morti!
LinkedIn Profile
CAPTCHA4us

*

Bragi

  • Trusty Member
  • ********
  • Replicant
  • *
  • 564
    • Neural network design blog
Re: NND & AICI
« Reply #93 on: March 19, 2011, 05:06:09 pm »
I used HyperCam2 (free) to create the video stream and Windows live movie maker to add the captions, remove unwanted parts and time stretch others.
For the thesaurus demo, I first made a little script with all the actions I wanted to show (of course, I forgot a bunch), which I then followed during recording (that are most of the pauses during which the captions are shown: me reading the script  ;D )

*

DaveMorton

  • Trusty Member
  • ********
  • Replicant
  • *
  • 636
  • Safe, Reliable Insanity, Since 1961
    • Geek Cave Creations
Re: NND & AICI
« Reply #94 on: March 19, 2011, 06:05:10 pm »
Thanks, Jan. I've tried Fraps in the past, but was unsatisfied with the results.WMM is ok, if a little clunky, but it's usable. :) Maybe I'll try making a screencast or two on setting up a chatbot "from scratch", and post them, when I get the time.
Comforting the Disturbed, Disturbing the Comfortable
Chat with Morti!
LinkedIn Profile
CAPTCHA4us

*

Bragi

  • Trusty Member
  • ********
  • Replicant
  • *
  • 564
    • Neural network design blog
Re: NND & AICI
« Reply #95 on: March 19, 2011, 06:13:06 pm »
Quote
WMM is ok, if a little clunky, but it's usable.
Yes, I was missing a little more control over timing, but for my purposes, and skill level, it's just fine.

*

Freddy

  • Administrator
  • **********************
  • Colossus
  • *
  • 6859
  • Mostly Harmless
Re: NND & AICI
« Reply #96 on: March 21, 2011, 01:27:01 pm »
Nice work Bragi, it's really impressive what you have done.  I just took a look at your blog and the bit about the 2008 CBC really caught my eye.  I really like the way you can teach it (can I say 'it' ?) and the way things are learned.  It seems like an engaging way to program something, ie with natural language.  In fact it seems more like teaching than programming.

Well done indeed, I only wish I understood half of it  O0

*

Bragi

  • Trusty Member
  • ********
  • Replicant
  • *
  • 564
    • Neural network design blog
Re: NND & AICI
« Reply #97 on: March 21, 2011, 02:26:03 pm »
To tell you the truth, I sort of worked towards the 'easier' questions. It turned out 2008 was the 'easiest' year to get working sort of, so that's what I focused on.  Some of the other years would have been a catastrophe.

Quote
I really like the way you can teach it (can I say 'it' ?) and the way things are learned.
Oh it's an it, definitely, more like a string of tricks, so I'd label it more of a recording device, for the moment.

I'm putting some other screen casts together, to show how you can expand the network from the designer. In the next one, I show how to add support for sentences like 'a human has 2 hands' from within the designer. The next ones will show how to make sentences work like 'I object to the object' (the thesaurus video already shows how to add the verb) and also  'top of the morning to you'.
I think, by showing how it's done, a lot will become more clearer. For me as well, it was a matter of sorting out my mind. By getting ready for the cbc, I've gotten a bit better defining the network, but I still can't explain it very well, so maybe it's better to simply show.

*

Bragi

  • Trusty Member
  • ********
  • Replicant
  • *
  • 564
    • Neural network design blog
Re: NND & AICI
« Reply #98 on: March 21, 2011, 02:27:17 pm »
By the way, the screen casts might take another couple of days, I'm taking it a bit slower the next week still (I'm also repainting the bathroom).

*

sunama

  • Trusty Member
  • ****
  • Electric Dreamer
  • *
  • 109
    • ai project
Re: NND & AICI
« Reply #99 on: April 26, 2011, 09:08:14 pm »
Hi,

I just tried to teach your ai something:

obama is the president of usa.
Who is the president of usa?

Your ai was not able to generate the correct answer.
Is this something that you are working on?

I was able to teach it that green is a colour and blue is a colour. It learnt this very quickly.
It could not understand about the president of USA.
www.aiproject.co.uk
Natural Language Processing

*

victorshulist

  • Trusty Member
  • ****
  • Electric Dreamer
  • *
  • 118
Re: NND & AICI
« Reply #100 on: April 26, 2011, 10:41:18 pm »
Probably because it involves a prepositional phrase  ( "of the USA").

prepositional phrases are complex.   The main reason being determining their antecedent.

 In this case it is easy though "of the USA" could apply to either:

subject "Obama"
verb "is"
direct compliment of verb ("president")

If the bot doesn't have knowledge of world, then it won't know that USA is a country and that countries can have presidents, and thus won't know to determine the correct antecent of "of the USA" (which of course is "president")

*

sunama

  • Trusty Member
  • ****
  • Electric Dreamer
  • *
  • 109
    • ai project
Re: NND & AICI
« Reply #101 on: April 26, 2011, 11:45:32 pm »
so, would the solution be:

usa is a country
usa has a president
the current president of usa is obama

would this work?
www.aiproject.co.uk
Natural Language Processing

*

Bragi

  • Trusty Member
  • ********
  • Replicant
  • *
  • 564
    • Neural network design blog
Re: NND & AICI
« Reply #102 on: April 27, 2011, 06:54:35 am »
Probably. It could be that USA was an unknown word.
I'm currently very buzzy on something. Can't say much yet. All will be clear in a couple of weeks I think. It will be interesting.   :D

*

Bragi

  • Trusty Member
  • ********
  • Replicant
  • *
  • 564
    • Neural network design blog
Re: NND & AICI
« Reply #103 on: April 27, 2011, 08:30:08 am »
so, would the solution be:

usa is a country
usa has a president
the current president of usa is obama

would this work?
Try something like:
a president is a property.
usa is a country.
the president of the usa is Obama.
I think something like that should work.

*

Freddy

  • Administrator
  • **********************
  • Colossus
  • *
  • 6859
  • Mostly Harmless
Re: NND & AICI
« Reply #104 on: May 23, 2012, 11:56:06 am »
Was thinking, is it time for another demonstration of how well AICI chats ? I mean a chat log to show off what it can do...

 


AI controlled F-16, for real!
by frankinstien (AI News )
Today at 05:40:28 am
Open AI GPT-4o - audio, vision, text combined reasoning
by MikeB (AI News )
May 14, 2024, 05:46:48 am
OpenAI Speech-to-Speech Reasoning Demo
by MikeB (AI News )
March 31, 2024, 01:00:53 pm
Say good-bye to GPUs...
by MikeB (AI News )
March 23, 2024, 09:23:52 am
Google Bard report
by ivan.moony (AI News )
February 14, 2024, 04:42:23 pm
Elon Musk's xAI Grok Chatbot
by MikeB (AI News )
December 11, 2023, 06:26:33 am
Nvidia Hype
by 8pla.net (AI News )
December 06, 2023, 10:04:52 pm
How will the OpenAI CEO being Fired affect ChatGPT?
by 8pla.net (AI News )
December 06, 2023, 09:54:25 pm

Users Online

261 Guests, 0 Users

Most Online Today: 269. Most Online Ever: 2369 (November 21, 2020, 04:08:13 pm)

Articles