Jump to content

Talk:Pyramid vector quantization: Difference between revisions

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Content deleted Content added
Created page with '{{WikiProject Computing |class=start |importance=low |software=y |software-importance=Low}}'
 
Jarek Duda (talk | contribs)
Line 1: Line 1:
{{WikiProject Computing |class=start |importance=low |software=y |software-importance=Low}}
{{WikiProject Computing |class=start |importance=low |software=y |software-importance=Low}}

== Improving uniformity by coordinate-wise power before projection ==

PVQ point density has characteristic nonuniformities - I have worked on improving it by adding some deformation, and just coordinate-wise power before applying projection works quite well, allows to reduce mean-squared quantization error by ~10% ( https://arxiv.org/pdf/1705.05285 ) - might be worth adding with image like attached:
[[File:PVQ power projection.png|thumb|600px|Improving uniformity of PVQ point distribution by coordinate-wise power <math>\vec{p} \to \vec{(p_i^w)_i}</math> of the vector before projection. Diagram presents constellations for <math>N=3</math> dimensions and written <math>K=11, 16, 23, 32</math> L1-norm.]]
--[[User:Jarek Duda|Jarek Duda]] ([[User talk:Jarek Duda|talk]]) 13:42, 15 May 2021 (UTC)

Revision as of 13:42, 15 May 2021

WikiProject iconComputing: Software Start‑class Low‑importance
WikiProject iconThis article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
StartThis article has been rated as Start-class on Wikipedia's content assessment scale.
LowThis article has been rated as Low-importance on the project's importance scale.
Taskforce icon
This article is supported by WikiProject Software (assessed as Low-importance).

Improving uniformity by coordinate-wise power before projection

PVQ point density has characteristic nonuniformities - I have worked on improving it by adding some deformation, and just coordinate-wise power before applying projection works quite well, allows to reduce mean-squared quantization error by ~10% ( https://arxiv.org/pdf/1705.05285 ) - might be worth adding with image like attached:

Improving uniformity of PVQ point distribution by coordinate-wise power of the vector before projection. Diagram presents constellations for dimensions and written L1-norm.

--Jarek Duda (talk) 13:42, 15 May 2021 (UTC)[reply]