[
https://issues.apache.org/jira/browse/MATH-1267?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14737858#comment-14737858
]
Gilles commented on MATH-1267:
------------------------------
bq. The Neuron2D is a metaphor
The question was whether that metaphor would bring anything more than the
current {{NeuronSquareMesh2D}} that is composed of a {{Network}} of {{Neuron}}
instances.
bq. Neuron2D coordinates need to be updated \[...\] IIUC this is the case in
general as well?
No. The layout is a {{Network}} property, and {{Neuron}} instances have only
one property: the weight vector.
Grid coordinate is an afterthought, for the specific kind of network that is a
2D-map.
I'm afraid that it looks overly important because that's what is currently
implemented (as the most often used SOFM).
We should really try and implement other kinds of ANN before attempting to draw
conclusions on what is central (for training and classification) or merely a
convenience (for visualization).
It is quite possible that I overlook important features of ANN, because I was
focused on a very simple one (SOFM).
> Grid coordinate of "Neuron" that belongs to a "NeuronSquareMesh2D"
> ------------------------------------------------------------------
>
> Key: MATH-1267
> URL: https://issues.apache.org/jira/browse/MATH-1267
> Project: Commons Math
> Issue Type: Wish
> Reporter: Gilles
> Assignee: Gilles
> Priority: Minor
> Fix For: 4.0, 3.6
>
> Attachments: LocationFinder.java
>
>
> When the network layout is a 2D grid, it useful to be able to retrieve the
> grid coordinate (row, column) of a a neuron.
> I propose to define a new class "LocationFinder" (in package
> "o.a.c.m.ml.neuralnet.twod.util") that will provide the functionality through
> a "getLocation(Neuron n)" method.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
