🚀 go-pugleaf

Steven Packham wrote:
> 
> Erik Vick wrote:
> >
> > Louis Savain wrote:
> > >
> > > In article <343AAFF1.6D35@dial.pipex.com>, Howard Snart
> > > <aac88@dial.pipex.com> wrote:
> > >
> > > >[...]
> > > >This approach stems from several misgivings I have over the existing
> > > >methods of teaching neural nets... most significantly
> > > >backtracking.(which I see as un-scalable, and fundementally wrong... you
> > > >force the outputs to conform, whereas the problem may lie far from the
> > > >outputs)
> > >
> > >   I agree.  Traditional neural nets are not biologically plausible,
> > > nor do they do the kind of learning that is needed to gain common
> > > sense.
> > >
> 
>         I disagree. I feel that the formation of a topographical map
> of the environment is obvious starting in Striata cortex area 17 (field
> of vision division,
> edge detection), 18 and 19 (starting three dimensional mappings with
> stereoscopic
> details. Where the corpus colosum begins to wire both hemisphere
> together).
> 
>         All of this structure and organization within the brain can be
> explained
>  with mechanism which the neuron does posses. The mechanism seem to
> contain some
> very simple functionality which parallels the learning process that
> back-prop
> engages. This is only one very insignificant aspect of computing ability
> which neurons do posses. In fact Back-Prop can be broken down
> into a very basic model that neuron could employ quite easily.

But backprop is NP-Complete! Read the book by Stephen Judd.

Ask a mathemetician about NP-Completeness. It isn't get-roundable.
 
> We are not ignoring this are we?
> 
>         The drift of gaze through the center mass of a pattern,
> this with the method area 17 and 18 divide the visual pattern can
> minimize the non linearity of the sensed 2-D space information, by the
> way.

This sentence makes no sense.

> Which can also introduce size invariance and the sense of rotational
> invariance of patterns to the degree which the brain does hold. Just
> modeling
> concepts. Many of these concepts which arise in the brain can be
> attribute to
> these simple computing ideas as the numerous neural networking papers
> have
> suggested.

Are you talking about Zipser now? His results are just human-eye
comparisons of the shape of some 3d plots ... not exactly scientifically
significant. In fact, ALL the papers I've seen trying to model visual
processing with backprop have had to fix, or at least fudge the results
in some way to make it look realistic.

>         Actually the Back Prop model is a very simplified model of a neuron.
> Yet it can mirror a portion of the computing ability the neuron
> contains. One
> that does not require the back prop algorithm to exist as a subroutine.
> It has the possibility to be mirrored because of the nature of patterns
> themselves.

Wrong. The neuron is FAR more complicated than the simple computational
model offered by an artificial neuron, which is of similar computational
significance to a synapse.

Nature doesn't tend to generously overendow.
 
>           When adding up all the other variables which the various real neuron
> structures present throughout the brain they can really add up. I am
> actually
> very surprised how obvious the computational mechanism of the brain are.

This, again, makes little sense.

> The
> problem I see is there vast variations. The many different structures
> are
> enough to take quite a while to familiarize ones self with.

This is true ... different areas of the brain do appear to perform
computation very differently. However, no-one is saying here that we are
close to being able to reproduce a brain. If we could reproduce EVEN A
SMALL PART of the brains functionality (for instance a plausible
learning algorithm for any part of it), then that would knock me
backwards off my seat in surprise.
 
> Opinion, "obviously common sense lies in
> the brains ability to model and communicate internally in
> some modeling mode.

Pardon?
 
> Failure in most systems to communicate with us as we do with each other
> seem to lie in their inability to track time, and model space.

Yes ... and temporal functionality is PRECISELY what is missing from
backprop, and what Louis is trying to suggest as useful, although it
appears to me that he is encountering the classic resistance of a
threatened paradigm.

Backprop is dead already ... it died six years ago, but has forgotten to
stop breathing.
 
walrus.

--

Crap to get around Nutscrape error:

lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`
lr;iwquhium]98=9 pooiq rp9ii0]09`1 49 i#ir#oi	#008fg]`248r`