[
https://issues.apache.org/jira/browse/MATH-1268?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14732294#comment-14732294
]
Gilles commented on MATH-1268:
------------------------------
bq. Is there something that's going to implement these interfaces?
Yes.
For example, the "U-matrix" currently computed by a method in the {{MapUtils}}
class should be defined in its own class.
bq. The class NeuronSquareMesh2D sounds very 'visual'
Indeed, a 2D map is indeed much used due to its straight visualization capacity.
Thus, {{NeuronSquareMesh2D}} is useful in that it specializes the general
{{Network}} to provide methods for easier implementation of visualization tools.
Thanks to those accessors, algorithms that purport to offer some way of looking
at the map need not be defined within the class.
I think that if something can be independently defined, then it should be
(separation of concerns). It is more elegant (IMHO), and more maintainable (2
small classes are better than a big one).
Also, this makes it possible to use a "MapVisualization" in the same way,
whether the algo is defined within CM or not.
> SOFM: interfaces for 2D map visualization
> -----------------------------------------
>
> Key: MATH-1268
> URL: https://issues.apache.org/jira/browse/MATH-1268
> Project: Commons Math
> Issue Type: Wish
> Reporter: Gilles
> Assignee: Gilles
> Priority: Minor
> Labels: api
> Fix For: 4.0, 3.6
>
> Attachments: MapDataVisualization.java, MapVisualization.java
>
>
> To be implemented by various visualizations of a 2D SOFM, I propose to
> introduce the following interfaces (in a new package
> {{o.a.c.m.ml.neuralnet.twod.util}}).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
