Results 1 to 11 of 11

Thread: RC6, about the CAILib class

  1. #1

    Thread Starter
    Junior Member
    Join Date
    Aug 2019
    Posts
    28

    RC6, about the CAILib class

    Hello Olaf,

    I was expecting the new inclusion of the cAILib class, but I didn't find anything in the new version.

    Will you include it in the next version this year?

    I hope for your clarification.

  2. #2
    PowerPoster
    Join Date
    Jun 2013
    Posts
    7,219

    Re: RC6, about the CAILib class

    Quote Originally Posted by Napstore View Post
    I was expecting the new inclusion of the cAILib class, but I didn't find anything in the new version.

    Will you include it in the next version this year?

    I hope for your clarification.
    cRegExp and cAILib are "yet unfinished" (but planned to include in one of the next releases).

    As for cAILib-wrapping of "existing OpenSource-libs":
    - the lib in question should have a "compatible license" (BSD, Apache, MIT, LGPL)
    - and should (preferrably) be "C-Code" (or have a C-interface, when it is C++-based)

    My current candidate is libFANN:
    - GitHub: https://github.com/libfann/fann
    - and normal Homepage: http://leenissen.dk/fann/wp/

    If FANN does not cover (functionality-wise) what you have in mind with your own AI-implementations,
    then please let me know.

    Currently it seems like the best compromise between "size, stability, features and speed" to me...

    Olaf

  3. #3

    Thread Starter
    Junior Member
    Join Date
    Aug 2019
    Posts
    28

    Re: RC6, about the CAILib class

    Thanks for the quick answer Olaf

    Well, libFANN has good implementations in the upper layer.

    I have already experimented with it in Python.

    but we hope to see the new CAILib inclusion released, it will speed up my character recognition project.

  4. #4
    PowerPoster wqweto's Avatar
    Join Date
    May 2011
    Location
    Sofia, Bulgaria
    Posts
    5,120

    Re: RC6, about the CAILib class

    Quote Originally Posted by Napstore View Post
    Thanks for the quick answer Olaf

    Well, libFANN has good implementations in the upper layer.

    I have already experimented with it in Python.

    but we hope to see the new CAILib inclusion released, it will speed up my character recognition project.
    Best would be to post some links to other AI projects which you want for Olaf to consider for inclusion.

    Search for "new CAILib" in google leads to nothing. What are you talking about?

    cheers,
    </wqw>

  5. #5
    PowerPoster
    Join Date
    Jun 2013
    Posts
    7,219

    Re: RC6, about the CAILib class

    Quote Originally Posted by wqweto View Post
    Search for "new CAILib" in google leads to nothing. What are you talking about?
    I've mentioned this new ClassName, when I talked about "planned, future inclusions" in the RC6.

    Ah..well, since you're here already (regarding new inclusions) ... -
    what about including your (MS-dependency-free) cHttpRequest-Project into one of the next RC6-releases?

    Are there any objections from your end?

    Olaf

  6. #6
    PowerPoster wqweto's Avatar
    Join Date
    May 2011
    Location
    Sofia, Bulgaria
    Posts
    5,120

    Re: RC6, about the CAILib class

    I wouldn't mind including any of my classes shared under MIT license but I wouldn't exactly consider adding 400KB additional size to the binary just for this class with it's custom TLS implementation a wise proposition.

    There is way too much code for PKI/certificates which cannot be shed automatically (like being in a .bas file) but is never actually used by cHttpRequest.

    I'm (frankly speaking) quite disappointed by the bloat accumulating from recent commits to this project.

    cheers,
    </wqw>

  7. #7
    Fanatic Member
    Join Date
    Sep 2010
    Location
    Italy
    Posts
    678

    Re: RC6, about the CAILib class

    Quote Originally Posted by Schmidt View Post

    If FANN does not cover (functionality-wise) what you have in mind with your own AI-implementations,
    then please let me know.
    At first glance FANN looks very good.
    One thing that for me is indispensable in AI is the ability to perform unsupervised learning.

    FANN does not seem to have this option.
    Enumeration of Training Algotirhm:
    http://libfann.github.io/fann/docs/f...ann_train_enum

    The technique I use to perform unsupervised learning is based on combining NN with Genetic Algorithm: [Neuroevolution]
    ( Small/simple example by me here NNandGA )
    How neuroevolution works | Risto Miikkulainen and Lex Fridman : https://www.youtube.com/watch?v=R6zStewfuZs

    This technique consists in a few words of the following steps:
    -Selection
    -Crossover
    -Mutation
    applied to a population of NNs


    Looking at the Reference Manual of FANN I see functions like:
    Code:
    fann_get_num_input Get the number of input neurons.
    fann_get_num_output Get the number of output neurons.
    fann_get_total_neurons Get the total number of neurons in the entire network.
    fann_get_total_connections Get the total number of connections in the entire network.
    fann_get_bias_array Get the number of biases in each layer in the network.
    fann_get_connection_array Get the connections in the network.
    fann_set_weight_array Set connections in the network.
    fann_set_weight Set a connection in the network.
    fann_get_weights Get all the network weights.
    fann_set_weights Set network weights.


    If these functions ( and perhaps others) will be sufficient to have a population of NNs and modify it following the evolution process using genetic algorithm, then I will deem FANN functionality-wise to me.
    Otherwise it will be functional only for "something narrower."

    That is why I suggest that we also look at NEAT (NeuroEvolution of Augmenting Topologies )

    It can be found in various versions and languages (also on Github)
    I do not know the license type and other useful information at this time.


    https://github.com/wagenaartje/neataptic


    Find the Right Version of NEAT for Your Needs:
    http://eplex.cs.ucf.edu/neat_software/#NEAT


    https://www.vbforums.com/showthread....=1#post5559446

  8. #8
    PowerPoster
    Join Date
    Jan 2020
    Posts
    3,746

    Re: RC6, about the CAILib class

    Perhaps it is a good method to develop a DLL alone.Rc6.dll should be kept a little smaller.

  9. #9

    Thread Starter
    Junior Member
    Join Date
    Aug 2019
    Posts
    28

    Re: RC6, about the CAILib class

    Quote Originally Posted by Schmidt View Post
    cRegExp and cAILib are "yet unfinished" (but planned to include in one of the next releases).

    Olaf

    Hello dear Olaf

    I come back to you to know if the updates of the new inclusions are in progress?

  10. #10
    PowerPoster wqweto's Avatar
    Join Date
    May 2011
    Location
    Sofia, Bulgaria
    Posts
    5,120

    Re: RC6, about the CAILib class

    Quote Originally Posted by Napstore View Post
    I come back to you to know if the updates of the new inclusions are in progress?
    Keep in mind that some plans get abandoned i.e. it's always an option for something planned to *not* happen.

    cheers,
    </wqw>

  11. #11
    Hyperactive Member
    Join Date
    Jan 2015
    Posts
    323

    Re: RC6, about the CAILib class

    Quote Originally Posted by Schmidt View Post
    cRegExp and cAILib are "yet unfinished" (but planned to include in one of the next releases).
    Olaf
    will cRegExp be as great as PCRE2?

Posting Permissions

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



Click Here to Expand Forum to Full Width