From: Jordan Dimov (jdimov@cis.clarion.edu)
Date: Mon May 07 2001 - 11:54:57 MDT
Here's my terrible attempt at translating the letter. The stuff in
[square brackets] is my comments. Keep in mind that neither Russian nor
English is my native language :-) So it's bound to be inaccurate.
Enjoy it.
-- Jordan
Dear Anton,
I am the principal developer of this model that has been so loudly
announced recently on the Internet-site that you mention.
The real situation is such:
1. There exists a program in C++, modeling functions (physical, inside
neurons, and logical, among neurons.) [Some short sentence, here, that I
can't make any sense of]. She [his program] can create nets of diverse
structural elements of the brain (various special cases of a
``supersmart'' neuron). The resulting temporary diagrams are akin to the
diagrams of a real neuron (obtained by experiments).
2. There is no ``normal'' neural-net model, and furthermore, there
isn't an ``artificial mind'' yet, unless, of course, doctor of biological
sciences Vitalii Borisovich Valcev has continued this research (as far as
it has become known to me recently, he has not). The problem is that not
everything is clear yet - for instance, I have implemented the learning
mechanisms purely from my own considerations (whatever seemed practical).
Additional research is needed, which, thanks to the limitations of the
results of our work, and some others, I can not (and wish not) conduct.
IP - hardly. [ Or is this ``IP is bad'' ?? ]
As it is, in my opinion, this work is promising enough, for instance,
with regard to speech recognition.
If there are any specific questions, don't be shy - write me.
You can also get in touch with V.B.Valcev himself, who lives in Moscow.
Best regards,
Cherkashin, Eugene Aleksandrovich
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:36 MDT