<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">Dear Andras,<br>
<br>
I see why it may seem on the first look that practopoiesis is
somehow related to reinforcement learning. However, any closer
look to the theory will reveal that this is not the case.
Practopoiesis is related to just a little bit to reinforcement
learning, and it is as much related to any other brain theory or
algorithm. At the end, in fact, it is quite a novel and exciting
way of looking at brain function.<br>
<br>
Maybe the contribution of practopoiesis can be appreciated best
if described in the language of learning. Imagine a system that
does not have one or two general learning rules that are applied
to all of the units of the system. Instead, imagine that the
number of different learning rules equals the number of units in
the system; a billion neurons means billion different learning
rules, whereby each learning rule maximizes a different function.
<br>
<br>
Moreover, imagine that, in this system, all its long-term memories
(explicit and implicit) are stored in those learning rules. Thus,
long-term memories are not stored primarily in the weights of
neuron connections (as widely presumed) but in the rules by which
the system changes its network in response to sensory inputs.
Then, when we retrieve a piece of information, or recognize a
stimulus, or make a decision, we use these learning mechanisms by
quickly applying them to the network (e.g., every second or even
faster) as a function of the incoming sensory inputs (or the
sensory-motor contingencies). As a result, the network
continuously changes its architecture with very high rate, and can
quickly come back to its previous architecture if it is presented
with a sequences of stimuli that has been presented already
previously. One of the key point is that this process of changing
the network is how we think, perceive, recall, categorize, etc.<br>
<br>
This process requires one more set of learning mechanisms that lay
behind those mentioned rules containing our long-term memory. This
latter set is responsible for acquiring our long-term memories
i.e., for determining for each unit which learning rules it needs
to use. Thus, there is a process of learning how to learn.<br>
<br>
Practopoietic theory explain how this is possible, why it works,
how such systems can be described in generalized cybernetics
terms, and why this approach is sufficiently adaptive to produce
intelligence on par with that of humans. In practopoietic theory,
the learning rules that store our long-term memories are referred
to, not as "learning", but as "reconstruction of knowledge" or in
Greek "anapoiesis". The paper also reviews behavioral evidence
indicating that our cognition in fact is anapoietic by its nature.<br>
<br>
I hope that this helps understand that practopoiesis is something
totally new and cannot simply be described with the existing
machine-leaning approaches and brain theories.<br>
<br>
With best regards,<br>
<br>
Danko<br>
<br>
<br>
On 4/14/14 3:05 PM, Andras Lorincz wrote:<br>
</div>
<blockquote
cite="mid:c90c18121a2a437a9a5201342cba0c5a@DB4PR06MB173.eurprd06.prod.outlook.com"
type="cite">
<meta http-equiv="Content-Type" content="text/html;
charset=ISO-8859-1">
<style type="text/css" style="display:none"><!--P{margin-top:0;margin-bottom:0;} .ms-cui-menu {background-color:#ffffff;border:1px rgb(171, 171, 171) solid;font-family:'Segoe UI WPC', 'Segoe UI', Tahoma, 'Microsoft Sans Serif', Verdana, sans-serif;font-size:11pt;color:rgb(51, 51, 51);} .ms-cui-menusection-title {display:none;} .ms-cui-ctl {vertical-align:text-top;text-decoration:none;color:rgb(51, 51, 51);} .ms-cui-ctl-on {background-color:rgb(223, 237, 250);opacity: 0.8;} .ms-cui-img-cont-float {display:inline-block;margin-top:2px} .ms-cui-smenu-inner {padding-top:0px;} .ms-owa-paste-option-icon {margin: 2px 4px 0px 4px;vertical-align:sub;padding-bottom: 2px;display:inline-block;} .ms-rtePasteFlyout-option:hover {background-color:rgb(223, 237, 250) !important;opacity:1 !important;} .ms-rtePasteFlyout-option {padding:8px 4px 8px 4px;outline:none;} .ms-cui-menusection {float:left; width:85px;height:24px;overflow:hidden}.wf {speak:none; font-weight:normal; font-variant:n!
ormal; tex
t-transform:none; -webkit-font-smoothing:antialiased; vertical-align:middle; display:inline-block;}.wf-family-owa {font-family:'o365Icons'}@font-face { font-family:'o365IconsIE8'; src:url('https://r4.res.outlook.com/owa/prem/15.0.918.10/resources/styles/office365icons.ie8.eot?#iefix') format('embedded-opentype'), url('https://r4.res.outlook.com/owa/prem/15.0.918.10/resources/styles/office365icons.ie8.woff') format('woff'), url('https://r4.res.outlook.com/owa/prem/15.0.918.10/resources/styles/office365icons.ie8.ttf') format('truetype'); font-weight:normal; font-style:normal;}@font-face { font-family:'o365IconsMouse'; src:url('https://r4.res.outlook.com/owa/prem/15.0.918.10/resources/styles/office365icons.mouse.eot?#iefix') format('embedded-opentype'), url('https://r4.res.outlook.com/owa/prem/15.0.918.10/resources/styles/office365icons.mouse.woff') format('woff'), url('https://r4.res.outlook.com/owa/prem/15.0.918.10/resources/styles/office!
365icons.m
ouse.ttf') format('truetype'); font-weight:normal; font-style:normal;}.wf-family-owa {font-family:'o365IconsMouse'}.ie8 .wf-family-owa {font-family:'o365IconsIE8'}.ie8 .wf-owa-play-large:before {content:'\e254';}.notIE8 .wf-owa-play-large:before {content:'\e054';}.ie8 .wf-owa-play-large {color:#FFFFFF/*$WFWhiteColor*/;}.notIE8 .wf-owa-play-large {border-color:#FFFFFF/*$WFWhiteColor*/; width:1.4em; height:1.4em; border-width:.1em; border-style:solid; border-radius:.8em; text-align:center; box-sizing:border-box; -moz-box-sizing:border-box; padding:0.1em; color:#FFFFFF/*$WFWhiteColor*/;}.ie8 .wf-size-play-large {width:40px; height:40px; font-size:30px}.notIE8 .wf-size-play-large {width:40px; height:40px; font-size:30px}--></style>
<div id="OWAFontStyleDivID"
style="font-size:12pt;color:#000000;background-color:#FFFFFF;font-family:Calibri,Arial,Helvetica,sans-serif;">
<p>You are heading reinforcement learning and the
so-called temporal difference learning in it. This is a good
direction, since many little details can be mapped to the
corico-basal ganglia-thalamocortical loops. Nonetheless, this
can't explain everything. The separation of outliers from the
generalizable subspace(s) is relevant, since the latter
enables one to fill in missing information, whereas the former
does not. This was the take-home message of the Netflix
competition and the subsequent developments on exact matrix
completion.<br>
</p>
<p>Andras<br>
</p>
<div>
<p><br>
</p>
<div name="divtagdefaultwrapper"
style="font-family:Calibri,Arial,Helvetica,sans-serif;
font-size:; margin:0">
<div>
<div>
<div>
<div><font>
<p class="MsoNormal" style="font-size:13px;
font-family:Calibri,Arial,Helvetica,sans-serif;
margin:0cm 0cm 0pt">
_________________________<br>
<font size="2">Andras Lorincz<br>
ECCAI Fellow</font></p>
<p class="MsoNormal" style="font-size:13px;
font-family:Calibri,Arial,Helvetica,sans-serif;
margin:0cm 0cm 0pt">
<font size="2">email: <a class="moz-txt-link-abbreviated" href="mailto:lorincz@inf.elte.hu">lorincz@inf.elte.hu</a> <br>
home: </font><a moz-do-not-send="true"
href="http://people.inf.elte.hu/lorincz"><font
size="2">http://people.inf.elte.hu/lorincz</font></a><font
size="2"> </font></p>
</font></div>
</div>
</div>
</div>
</div>
</div>
<div style="color: rgb(40, 40, 40);">
<hr tabindex="-1" style="display:inline-block; width:98%">
<div id="divRplyFwdMsg" dir="ltr"><font style="font-size:11pt"
face="Calibri, sans-serif" color="#000000"><b>From:</b>
Connectionists
<a class="moz-txt-link-rfc2396E" href="mailto:connectionists-bounces@mailman.srv.cs.cmu.edu"><connectionists-bounces@mailman.srv.cs.cmu.edu></a> on
behalf of Danko Nikolic
<a class="moz-txt-link-rfc2396E" href="mailto:danko.nikolic@googlemail.com"><danko.nikolic@googlemail.com></a><br>
<b>Sent:</b> Monday, April 14, 2014 12:51 PM<br>
<b>To:</b> <a class="moz-txt-link-abbreviated" href="mailto:connectionists@mailman.srv.cs.cmu.edu">connectionists@mailman.srv.cs.cmu.edu</a><br>
<b>Subject:</b> Re: Connectionists: how the brain works?
(UNCLASSIFIED)</font>
<div> </div>
</div>
<div>
<div class="moz-cite-prefix">Dear all,<br>
<br>
It has been very interesting to follow the discussion on
the functioning of ART, stability-plasticity dilemma and
the related issues. In that context, I would like to point
to an exciting property of the practopoietic theory, which
enables us to understand what is needed for a general
solution to the problems similar to the
stability-plasticity dilemma.
<br>
<br>
The issue of stability-plasticity dilemma can be described
as a problem of deciding when a new category of a stimulus
needs to be created and the system has to be adjusted as
opposed to deciding to treat the stimulus as old and
familiar and thus, not needing to adjust. Practopoietic
theory helps us understand how a general solution can be
implemented for deciding whether to use old types of
behavior or to come up with new ones. This is possible in
a so-called "T_3 system" in which a process called
"anapoiesis" takes place. When a system is organized in
such a T_3 way, every stimulus, old or new, is treated in
the same fashion, i.e., as new. The system always
adjusts--to everything(!)--even to stimuli that have been
seen thousands of times. There is never a simple direct
categorization (or pattern recognition) in which a
mathematical mapping would take place from input vectors
to output vectors, as traditionally implemented in
multi-layer neural networks.
<br>
<br>
Rather the system readjusts itself continuously to prepare
for interactions with the surrounding world. The only
simple input-output mappings that take place are the
sensory-motor loops that execute the actual behavior. The
internal processes corresponding to perception,
recognition, categorization etc. are implemented by the
mechanisms of internal system adjustments (based on
anapoiesis). These mechanisms create new sensory-motor
loops, which are then most similar to the traditional
mapping operations. The difference between old and new
stimuli (i.e., familiar and unfamiliar) is detectable in
the behavior of the system only because the system adjusts
quicker to the older that to the newer stimuli.
<br>
<br>
The claimed advantage of such a T_3 practopoietic system
is that only such a system can become generally
intelligent as a whole and behave adaptively and
consciously with understanding of what is going on around;
The system forms a general "adjustment machine" that can
become aware of its surroundings and can be capable of
interpreting the situation appropriately to decide on the
next action. Thus, the perceptual dilemma of stability vs.
plasticity is converted into a general understanding of
the current situation and the needs of the system. If the
current goals of the system requires treating a slightly
novel stimulus as new, it will be treated as "new".
However, if a slight change in the stimulus features does
not make a difference for the current goals and the
situation, than the stimulus will be treated as "old".<br>
<br>
Importantly, practopoietic theory is not formulated in
terms of neurons (inhibition, excitation, connections,
changes of synaptic weights, etc.). Instead, the theory is
formulated much more elegantly--in terms of interactions
between cybernetic control mechanisms organized into a
specific type of hierarchy (poietic hierarchy). This
abstract formulation is extremely helpful for two reasons.
First, it enables one to focus on the most important
functional aspects and thus, to understand much easier the
underlying principles of system operations. Second, it
tells us what is needed to create intelligent behavior
using any type of implementation, neuronal or
non-neuronal.<br>
<br>
I hope this will be motivating enough to give
practopoiesis a read.<br>
<br>
With best regards,<br>
<br>
Danko<br>
<br>
<br>
<br>
Link:<br>
<a moz-do-not-send="true" class="moz-txt-link-freetext"
href="http://www.danko-nikolic.com/practopoiesis/">http://www.danko-nikolic.com/practopoiesis/</a><br>
<br>
<br>
<br>
On 4/11/14 2:42 AM, Tsvi Achler wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">I can't comment on most of this, but I am
not sure if all models of sparsity and sparse coding
fall into the connectionist realm either because some
make statistical assumptions.
<div>-Tsvi</div>
</div>
<div class="gmail_extra"><br>
<br>
<div class="gmail_quote">On Tue, Apr 8, 2014 at 9:19 PM,
Juyang Weng <span dir="ltr">
<<a moz-do-not-send="true"
href="mailto:weng@cse.msu.edu" target="_blank">weng@cse.msu.edu</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex; border-left:1px #ccc solid; padding-left:1ex">
<div bgcolor="#FFFFFF">Tavi:<br>
<br>
Let me explain a little more detail:<br>
<br>
There are two large categories of biological
neurons, excitatory and inhibitory. Both are
developed through mainly signal statistics,
<br>
not specified primarily by the genomes. Not all
people agree with my this point, but please
tolerate my this view for now.
<br>
I gave a more detailed discussion on this view in
my NAI book. <br>
<br>
The main effect of inhibitory connections is to
reduce the number of firing neurons (David Field
called it sparse coding), with the help of
<br>
excitatory connections. This sparse coding is
important because those do not fire are long term
memory of the area at this point of time.<br>
My this view is different from David Field. He
wrote that sparse coding is for the current
representations. I think sparse coding is
<br>
necessary for long-term memory. Not all people
agree with my this point, but please tolerate my
this view for now.
<br>
<br>
However, this reduction requires very fast
parallel neuronal updates to avoid uncontrollable
large-magnitude oscillations.
<br>
Even with the fast biological parallel neuronal
updates, we still see slow but small-magnitude
oscillations such as the
<br>
well-known theta waves and alpha waves. My view
is that such slow but small-magnitude oscillations
are side effects of
<br>
excitatory and inhibitory connections that form
many loops, not something really desirable for the
brain operation (sorry,
<br>
Paul Werbos). Not all people agree with my this
point, but please tolerate my this view for now.
<br>
<br>
Therefore, as far as I understand, all computer
simulations for spiking neurons are not showing
major brain functions<br>
because they have to deal with the slow
oscillations that are very different from the
brain's, e.g., as Dr. Henry Markram reported<br>
(40Hz?). <br>
<br>
The above discussion again shows the power and
necessity of an overarching brain theory like that
in my NAI book.
<br>
Those who only simulate biological neurons using
superficial biological phenomena are not going to
demonstrate
<br>
any major brain functions. They can talk about
signal statistics from their simulations, but
signal statistics are far from brain functions.
<br>
<br>
-John
<div>
<div class="h5"><br>
<br>
<div>On 4/8/14 1:30 AM, Tsvi Achler wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">Hi John,
<div>ART evaluates distance between the
contending representation and the
current input through vigilance. If
they are too far apart, a poor vigilance
signal will be triggered.</div>
<div>The best resonance will be achieved
when they have the least amount of
distance.</div>
<div>If in your model, K-nearest neighbors
is used without a neural equivalent,
then your model is not quite in the
spirit of a connectionist model.</div>
<div>For example, Bayesian networks do a
great job emulating brain behavior,
modeling the integration of priors. and
has been invaluable to model cognitive
studies. However they assume a
statistical configuration of connections
and distributions which is not quite
known how to emulate with neurons. Thus
pure Bayesian models are also
questionable in terms of connectionist
modeling. But some connectionist models
can emulate some statistical models for
example see section 2.4 in Thomas &
McClelland's chapter in Sun's 2008 book
(<a moz-do-not-send="true"
href="http://www.psyc.bbk.ac.uk/people/academic/thomas_m/TM_Cambridge_sub.pdf"
target="_blank">http://www.psyc.bbk.ac.uk/people/academic/thomas_m/TM_Cambridge_sub.pdf</a>).</div>
<div>I am not suggesting <span
style="font-size:13.333333969116211px;
font-family:arial,sans-serif">Hodgkin-Huxley</span> level
detailed neuron models, however
connectionist models should have their
connections explicitly defined. </div>
<div>Sincerely,</div>
<div>-Tsvi</div>
<div><br>
</div>
</div>
<div class="gmail_extra"><br>
<br>
<div class="gmail_quote">On Mon, Apr 7,
2014 at 10:58 AM, Juyang Weng <span
dir="ltr">
<<a moz-do-not-send="true"
href="mailto:weng@cse.msu.edu"
target="_blank">weng@cse.msu.edu</a>></span>
wrote:<br>
<blockquote class="gmail_quote"
style="margin:0 0 0 .8ex;
border-left:1px #ccc solid;
padding-left:1ex">
<div bgcolor="#FFFFFF">Tsvi,<br>
<br>
Note that ART uses a vigilance value
to pick up the first "acceptable"
match in its sequential bottom-up
and top-down search.<br>
I believe that was Steve meant when
he mentioned vigilance. <br>
<br>
Why do you think "ART as a neural
way to implement a K-nearest
neighbor algorithm"?
<br>
If not all the neighbors have
sequentially participated,<br>
how can ART find the nearest
neighbor, let alone K-nearest
neighbor?<br>
<br>
Our DN uses an explicit k-nearest
mechanism to find the k-nearest
neighbors in every network update,
<br>
to avoid the problems of slow
resonance in existing models of
spiking neuronal networks.
<br>
The explicit k-nearest mechanism
itself is not meant to be
biologically plausible,
<br>
but it gives a computational
advantage for software simulation of
large networks <br>
at a speed slower than 1000 network
updates per second.<br>
<br>
I guess that more detailed molecular
simulations of individual neuronal
spikes (such as using the
Hodgkin-Huxley model of<br>
a neuron, using the <a
moz-do-not-send="true"
href="http://www.neuron.yale.edu/neuron/"
target="_blank">
NEURON software,</a> or like <a
moz-do-not-send="true"
href="http://bluebrain.epfl.ch/"
target="_blank">
the Blue Brain project</a>
directed by respected Dr. Henry
Markram) <br>
are very useful for showing some
detailed molecular, synaptic, and
neuronal properties.<br>
However, they miss necessary
brain-system-level mechanisms so
much that it is difficult for them
<br>
to show major brain-scale functions
<br>
(such as learning to recognize
objects and detection of natural
objects directly from natural
cluttered scenes).
<br>
<br>
According to my understanding, if
one uses a detailed neuronal model
for each of a variety of neuronal
types and<br>
connects those simulated neurons of
different types according to a
diagram of Brodmann areas,
<br>
his simulation is NOT going to lead
to any major brain function. <br>
He still needs brain-system-level
knowledge such as that taught in the
BMI 871 course.
<br>
<br>
-John <br>
<div>
<div><br>
<div>On 4/7/14 8:07 AM, Tsvi
Achler wrote:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div>Dear Steve, John</div>
I think such discussions are
great to spark interests in
feedback (output back to
input) such models which I
feel should be given much
more attention.
<div>In this vein it may be
better to discuss more of
the details here than to
suggest to read a
reference.</div>
<div><br>
</div>
<div>Basically I see ART as
a neural way to implement
a K-nearest neighbor
algorithm. Clearly the
way ART overcomes the
neural hurdles is immense
especially in figuring out
how to coordinate neurons.
However it is also
important to summarize
such methods in
algorithmic terms which I
attempt to do here (and
please comment/correct).</div>
<div>Instar learning is used
to find the best weights
for quick feedforward
recognition without too
much resonance (otherwise
more resonance will be
needed). Outstar learning
is used to find the
expectation of the
patterns. The resonance
mechanism evaluates
distances between the
"neighbors" evaluating how
close differing outputs
are to the input pattern
(using the expectation).
By choosing one winner
the network is equivalent
to a 1-nearest neighbor
model. If you open it up
to more winners (eg k
winners) as you suggest
then it becomes a
k-nearest neighbor
mechanism.</div>
<div><br>
</div>
<div>Clearly I focused here
on the main ART modules
and did not discuss other
additions. But I want to
just focus on the main
idea at this point.</div>
<div>Sincerely,</div>
<div>-Tsvi</div>
</div>
<div class="gmail_extra"><br>
<br>
<div class="gmail_quote">On
Sun, Apr 6, 2014 at 1:30
PM, Stephen Grossberg <span
dir="ltr">
<<a
moz-do-not-send="true"
href="mailto:steve@cns.bu.edu" target="_blank">steve@cns.bu.edu</a>></span>
wrote:<br>
<blockquote
class="gmail_quote"
style="margin:0 0 0
.8ex; border-left:1px
#ccc solid;
padding-left:1ex">
<div
style="word-wrap:break-word"><font
size="5"
face="Arial">Dear
John,</font>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">Thanks
for your
questions. I reply
below.</font></div>
<div><font size="5"
face="Arial"><br>
</font>
<div>
<div>
<div><font
size="5"
face="Arial">On
Apr 5, 2014,
at 10:51 AM,
Juyang Weng
wrote:</font></div>
<font size="5"
face="Arial"><br>
</font>
<blockquote
type="cite">
<div
bgcolor="#FFFFFF"><font
size="5"
face="Arial">Dear
Steve,<br>
<br>
This is one of
my long-time
questions that
I did not have
a chance to
ask you when I
met you many
times before.
<br>
But they may
be useful for
some people on
this list. <br>
Please accept
my apology of
my question
implies any
false
impression
that I did not
intend.<br>
<br>
(1) Your
statement
below seems to
have confirmed
my
understanding:
<br>
Your top-down
process in ART
in the late
1990's is
basically for
finding an
acceptable
match
<br>
between the
input feature
vector and the
stored feature
vectors
represented by
neurons (not
meant for the
nearest
match).
<br>
</font></div>
</blockquote>
<div><font
size="5"
face="Arial"><br>
</font></div>
</div>
<font size="5"
face="Arial">ART
has developed a
lot since the
1990s. A
non-technical
but fairly
comprehensive
review article
was published in
2012 in
<i>Neural
Networks</i>
and can be found
at <a
moz-do-not-send="true"
href="http://cns.bu.edu/%7Esteve/ART.pdf" target="_blank">http://cns.bu.edu/~steve/ART.pdf</a>.</font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">I
do not think
about the
top-down process
in ART in quite
the way that you
state above. My
reason for this
is summarized by
the acronym
CLEARS for the
processes of
Consciousness,
Learning,
Expectation,
Attention,
Resonance, and
Synchrony. </font><span
style="font-family:Arial;
font-size:x-large">All the CLEARS processes come into this story, and </span><span
style="font-family:Arial;
font-size:x-large">ART top-down mechanisms contribute to all of them.
For me, the most
fundamental
issues concern
how ART
dynamically
self-stabilizes
the memories
that are learned
within the
model's
bottom-up
adaptive filters
and top-down
expectations. </span></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">In
particular,
during learning,
a big enough
mismatch can
lead to
hypothesis
testing and
search for a
new, or
previously
learned,
category that
leads to an
acceptable
match. The
criterion for
what is "big
enough mismatch"
or "acceptable
match" is
regulated by a
vigilance
parameter that
can itself vary
in a
state-dependent
way.</font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">After
learning occurs,
a bottom-up
input pattern
typically
directly selects
the
best-matching
category,
without any
hypothesis
testing or
search. And even
if there is a
reset due to a
large initial
mismatch with a
previously
active category,
a single reset
event may lead
directly to a
matching
category that
can directly
resonate with
the data. </font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">I
should note that
all of the
foundational
predictions of
ART now have
substantial
bodies of
psychological
and
neurobiological
data to support
them. See the
review article
if you would
like to read
about them.</font></div>
<div>
<div><font
size="5"
face="Arial"><br>
</font>
<blockquote
type="cite">
<div
bgcolor="#FFFFFF"><font
size="5"
face="Arial">The
currently
active neuron
is the one
being examined
by the top
down process<br>
</font></div>
</blockquote>
<div><font
size="5"
face="Arial"><br>
</font></div>
</div>
<font size="5"
face="Arial">I'm
not sure what
you mean by
"being
examined", but
perhaps my
comment above
may deal with
it.</font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">I
should comment,
though, about
your use of the
word "currently
active neuron".
I assume that
you mean at the
category level. </font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">In
this
regard, there
are two ART's.
The first aspect
of ART is as a
cognitive and
neural theory
whose scope,
which includes
perceptual,
cognitive, and
adaptively timed
cognitive-emotional
dynamics, among
other processes,
is illustrated
by the above
referenced 2012
review article
in <i>Neural
Networks</i>.
In the
biological
theory, there is
no general
commitment to
just one
"currently
active neuron".
One always
considers the
neuronal
population, or
populations,
that represent a
learned
category.
Sometimes, but
not always, a
winner-take-all
category is
chosen. </font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">The
2012 review
article
illustrates some
of the large
data bases of
psychological
and
neurobiological
data that have
been explained
in a principled
way,
quantitatively
simulated, and
successfully
predicted by ART
over a period of
decades.
ART-like
processing is,
however,
certainly not
the only kind of
computation that
may be needed to
understand how
the brain works.
The paradigm
called
Complementary
Computing that I
introduced
awhile ago makes
precise the
sense in which
ART may be just
one kind of
dynamics
supported by
advanced brains.
This is also
summarized in
the review
article.<br>
</font>
<div><font
size="5"
face="Arial"><br>
</font></div>
<div><font
size="5"
face="Arial">The
second aspect
of ART is as a
series of
algorithms
that
mathematically
characterize
key ART design
principles and
mechanisms in
a focused
setting, and
provide
algorithms for
large-scale
applications
in engineering
and
technology.
ARTMAP, fuzzy
ARTMAP, and
distributed
ARTMAP are
among these,
all of them
developed with
Gail
Carpenter.
Some of these
algorithms use
winner-take-all
categories to
enable the
proof of
mathematical
theorems that
characterize
how underlying
design
principles
work. In
contrast, the
distributed
ARTMAP family
of algorithms,
developed by
Gail Carpenter
and her
colleagues,
allows for
distributed
category
representations
without losing
the benefits
of fast,
incremental,
self-stabilizing
learning and
prediction in
response to a
large
non-stationary
databases that
can include
many
unexpected
events. </font></div>
<div><font
size="5"
face="Arial"><br>
</font></div>
<div><font
size="5"
face="Arial">See,
e.g., <a
moz-do-not-send="true"
href="http://techlab.bu.edu/members/gail/articles/115_dART_NN_1997_.pdf"
target="_blank">http://techlab.bu.edu/members/gail/articles/115_dART_NN_1997_.pdf</a>
and
<a
moz-do-not-send="true"
href="http://techlab.bu.edu/members/gail/articles/155_Fusion2008_CarpenterRavindran.pdf"
target="_blank">
http://techlab.bu.edu/members/gail/articles/155_Fusion2008_CarpenterRavindran.pdf</a>.</font></div>
<div><font
size="5"
face="Arial"><br>
</font></div>
<div><font
size="5"
face="Arial">I
should note
that FAST
learning is a
technical
concept: it
means that
each adaptive
weight can
converge to
its new
equilibrium
value on EACH
learning
trial. That is
why ART
algorithms can
often
successfully
carry out
one-trial
incremental
learning of a
data base.
This is not
true of many
other
algorithms,
such as back
propagation,
simulated
annealing, and
the like,
which all
experience
catastrophic
forgetting if
they try to do
fast learning.
Almost all
other learning
algorithms
need to be run
using slow
learning, that
allows only a
small
increment in
the values of
adaptive
weights on
each learning
trial, to
avoid massive
memory
instabilities,
and work best
in response to
stationary
data. Such
algorithms
often fail to
detect
important rare
cases, among
other
limitations.
ART can
provably learn
in either the
fast or slow
mode in
response to
non-stationary
data.</font></div>
<div>
<div><font
size="5"
face="Arial"><br>
</font></div>
<blockquote
type="cite">
<div
bgcolor="#FFFFFF"><font
size="5"
face="Arial">in
a sequential
fashion: one
neuron after
another, until
an acceptable
neuron is
found.<br>
<br>
(2) The input
to the ART in
the late
1990's is for
a single
feature vector
as a
monolithic
input.
<br>
By monolithic,
I mean that
all neurons
take the
entire input
feature vector
as input.
<br>
I raise this
point here
because neuron
in ART in the
late 1990's
does not have
an explicit
local sensory
receptive
field (SRF),
<br>
i.e., are
fully
connected from
all components
of the input
vector. A
local SRF
means that
each neuron is
only connected
to a small
region
<br>
in an input
image. <br>
</font></div>
</blockquote>
<div><font
size="5"
face="Arial"><br>
</font></div>
</div>
<font size="5"
face="Arial">Various
ART algorithms
for technology
do use fully
connected
networks. They
represent a
single-channel
case, which is
often sufficient
in applications
and which
simplifies
mathematical
proofs. However,
the
single-channel
case is, as its
name suggests,
not a necessary
constraint on
ART design. </font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">In
addition, many
ART biological
models do not
restrict
themselves to
the
single-channel
case, and do
have receptive
fields. These
include the
LAMINART family
of models that
predict
functional roles
for many
identified cell
types in the
laminar circuits
of cerebral
cortex. These
models
illustrate how
variations of a
shared laminar
circuit design
can carry out
very different
intelligent
functions, such
as 3D vision
(e.g., 3D
LAMINART),
speech and
language (e.g.,
cARTWORD), and
cognitive
information
processing
(e.g., LIST
PARSE). They are
all summarized
in the 2012
review article,
with the
archival
articles
themselves on my
web page
<a
moz-do-not-send="true"
href="http://cns.bu.edu/%7Esteve" target="_blank">http://cns.bu.edu/~steve</a>. </font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">The
existence of
these laminar
variations-on-a-theme
provides an
existence proof
for the exciting
goal of
designing a
family of chips
whose
specializations
can realize all
aspects of
higher
intelligence,
and which can be
consistently
connected
because they all
share a similar
underlying
design. Work on
achieving this
goal can
productively
occupy lots of
creative
modelers and
technologists
for many years
to come.</font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">I
hope that the
above replies
provide some
relevant
information, as
well as pointers
for finding
more.</font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">Best,</font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial">Steve</font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div><font size="5"
face="Arial"><br>
</font></div>
<div>
<blockquote
type="cite">
<div
bgcolor="#FFFFFF">
<div><font
size="5"
face="Arial"><br>
My apology
again if my
understanding
above has
errors
although I
have examined
the above two
points
carefully
<br>
through
multiple your
papers.<br>
<br>
Best regards,<br>
<br>
-John<br>
<br>
</font></div>
<div>
<pre cols="72"><font face="Arial"><span style="font-size:18px">Juyang (John) Weng, Professor
Department of Computer Science and Engineering
MSU Cognitive Science Program and MSU Neuroscience Program
428 S Shaw Ln Rm 3115
Michigan State University
East Lansing, MI 48824 USA
Tel: <a moz-do-not-send="true" href="tel:517-353-4388" value="+15173534388" target="_blank">517-353-4388</a>
Fax: <a moz-do-not-send="true" href="tel:517-432-1061" value="+15174321061" target="_blank">517-432-1061</a>
Email: <a moz-do-not-send="true" href="mailto:weng@cse.msu.edu" target="_blank">weng@cse.msu.edu</a>
URL: <a moz-do-not-send="true" href="http://www.cse.msu.edu/%7Eweng/" target="_blank">http://www.cse.msu.edu/~weng/</a>
----------------------------------------------
</span></font></pre>
</div>
</div>
</blockquote>
</div>
<div><font size="5"
face="Arial"><br>
</font>
<div><font
size="5"
face="Arial"><span
style="line-height:normal;
text-indent:0px;
border-collapse:separate;
letter-spacing:normal;
font-variant:normal;
text-transform:none;
font-style:normal;
white-space:normal;
font-weight:normal;
word-spacing:0px"><span style="line-height:normal; text-indent:0px;
border-collapse:separate;
letter-spacing:normal;
font-variant:normal;
text-transform:none;
font-style:normal;
white-space:normal;
font-weight:normal;
word-spacing:0px">
<div
style="word-wrap:break-word"><span
style="line-height:normal;
text-indent:0px;
border-collapse:separate;
letter-spacing:normal;
font-variant:normal;
text-transform:none;
font-style:normal;
white-space:normal;
font-weight:normal;
word-spacing:0px">
<div
style="word-wrap:break-word">
<div>
<div>
<div>
<div>Stephen
Grossberg</div>
<div>Wang
Professor of
Cognitive and
Neural Systems</div>
<div>Professor
of
Mathematics,
Psychology,
and Biomedical
Engineering</div>
<div>
<div>Director,
Center for
Adaptive
Systems <a
moz-do-not-send="true"
href="http://www.cns.bu.edu/about/cas.html" target="_blank">http://www.cns.bu.edu/about/cas.html</a></div>
</div>
<div><a
moz-do-not-send="true"
href="http://cns.bu.edu/%7Esteve" target="_blank">http://cns.bu.edu/~steve</a></div>
<div><a
moz-do-not-send="true"
href="mailto:steve@bu.edu" target="_blank">steve@bu.edu</a></div>
</div>
</div>
</div>
<div><br>
</div>
</div>
</span></div>
</span><br>
</span><br>
</font></div>
<br>
</div>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</blockquote>
<br>
</div>
</div>
<pre cols="72"><span><font color="#888888">--
--
Juyang (John) Weng, Professor
Department of Computer Science and Engineering
MSU Cognitive Science Program and MSU Neuroscience Program
428 S Shaw Ln Rm 3115
Michigan State University
East Lansing, MI 48824 USA
Tel: <a moz-do-not-send="true" href="tel:517-353-4388" value="+15173534388" target="_blank">517-353-4388</a></font></span><div>
Fax: <a moz-do-not-send="true" href="tel:517-432-1061" value="+15174321061" target="_blank">517-432-1061</a>
Email: <a moz-do-not-send="true" href="mailto:weng@cse.msu.edu" target="_blank">weng@cse.msu.edu</a>
URL: <a moz-do-not-send="true" href="http://www.cse.msu.edu/%7Eweng/" target="_blank">http://www.cse.msu.edu/~weng/</a>
----------------------------------------------
</div></pre>
</div>
</blockquote>
</div>
<br>
</div>
</blockquote>
<br>
<pre cols="72">--
--
Juyang (John) Weng, Professor
Department of Computer Science and Engineering
MSU Cognitive Science Program and MSU Neuroscience Program
428 S Shaw Ln Rm 3115
Michigan State University
East Lansing, MI 48824 USA
Tel: <a moz-do-not-send="true" href="tel:517-353-4388" value="+15173534388" target="_blank">517-353-4388</a>
Fax: <a moz-do-not-send="true" href="tel:517-432-1061" value="+15174321061" target="_blank">517-432-1061</a>
Email: <a moz-do-not-send="true" href="mailto:weng@cse.msu.edu" target="_blank">weng@cse.msu.edu</a>
URL: <a moz-do-not-send="true" href="http://www.cse.msu.edu/%7Eweng/" target="_blank">http://www.cse.msu.edu/~weng/</a>
----------------------------------------------
</pre>
</div>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</blockquote>
<br>
<br>
<pre class="moz-signature" cols="72">--
Prof. Dr. Danko Nikolić
Web:
<a moz-do-not-send="true" class="moz-txt-link-freetext" href="http://www.danko-nikolic.com">http://www.danko-nikolic.com</a>
Mail address 1:
Department of Neurophysiology
Max Planck Institut for Brain Research
Deutschordenstr. 46
60528 Frankfurt am Main
GERMANY
Mail address 2:
Frankfurt Institute for Advanced Studies
Wolfgang Goethe University
Ruth-Moufang-Str. 1
60433 Frankfurt am Main
GERMANY
----------------------------
Office: (..49-69) 96769-736
Lab: (..49-69) 96769-209
Fax: (..49-69) 96769-327
<a moz-do-not-send="true" class="moz-txt-link-abbreviated" href="mailto:danko.nikolic@gmail.com">danko.nikolic@gmail.com</a>
----------------------------</pre>
</div>
</div>
</div>
</blockquote>
<br>
<br>
<pre class="moz-signature" cols="72">--
Prof. Dr. Danko Nikolić
Web:
<a class="moz-txt-link-freetext" href="http://www.danko-nikolic.com">http://www.danko-nikolic.com</a>
Mail address 1:
Department of Neurophysiology
Max Planck Institut for Brain Research
Deutschordenstr. 46
60528 Frankfurt am Main
GERMANY
Mail address 2:
Frankfurt Institute for Advanced Studies
Wolfgang Goethe University
Ruth-Moufang-Str. 1
60433 Frankfurt am Main
GERMANY
----------------------------
Office: (..49-69) 96769-736
Lab: (..49-69) 96769-209
Fax: (..49-69) 96769-327
<a class="moz-txt-link-abbreviated" href="mailto:danko.nikolic@gmail.com">danko.nikolic@gmail.com</a>
----------------------------</pre>
</body>
</html>