OpenEuphoria

Re: Perceptron in Euphoria

• Posted by Hawke <mdeland at NWINFO.NET> Nov 29, 1998
• 523 views

Alfredo Brand wrote:
*extreme snippage to produce a paragraph of coherence to a particular
 facet of the topic to discuss*
> The perceptron is a program that learn concepts, i.e. it can learn to
> respond with True (1) or False (0) for inputs we present to it, by
> repeatedly "studying" examples presented to it.
> The perceptron is trained to respond to each input vector with a
> corresponding target output of either 0 or 1. The learning rule has been
> proven to converge on a solution in finite time if a solution exists.
> Perceptron networks have several limitations. First, the output values of a
> perceptron can take on only one of two values (True or False). Second,
> perceptrons can only classify linearly separable sets of vectors.
> If the vectors are not linearly separable learning will never reach
> a point where all vectors are classified properly.
I wrote (on main EU contributions page) a library that i believe would
go nicely with this, if you are interested.  It would extend your
perceptron to the point that the limitations you describe would vanish.
Its a library for fuzzy logic, and that would give it the ability
to provide weighted answers for your weighted inputs that are
not linearly seperable vectors.

In other words, we can combine the ability to provide *meaningful*
answers to approximate questions with the ability to learn from
the questions, the answers, and the bias of the mistakes.

The fuzzy logic library can (especially if i could find the time
to reapply some speed tweaking tricks I have picked up) resolve
the vectors and responses in swift abandon, perhaps even faster
than resolving to a boolean answer (floating point co-proc).

You're welcome to snag the library and apply it however you
wish, and I believe the examples and documentation within it
will guide you in your adapting the fuzzy logic library as
a front/back end to your learning machine.

The result of that endeavor, especially being coded in nearly
pure or fully pure euphoria could be quite a feather in EU's cap.

Also, another frame of reference, although I am sorry to say
not a fully working prototype, was an experiment in AI by
... I want to say michael sabal... is that right?

The code itself, however, the algorithms contained within, could be
useful to you as well.

--Hawke'