UVQ

11 views
Skip to first unread message

Bandurao Naik

unread,
Apr 5, 2022, 11:42:44 AM4/5/22
to OpendTect Users
Dear Friends...
Can anybody provide me the workflow for the SUPERVISED UVQ segmentation in OpendTect. Methodology for Unsupervised UVQ is available on YouTube but not much information about supervised UVQ.
 Also, earlier I posted my experience with MPSI in OpendTect and requested views and also provide ideas to improvise the results, but unfortunately I did not get any reply. Do ArkCLS/Earthworks will look into this...
With best regards
SB Rao Naik




--
Disclaimer:
The  information contained in this communication is intended solely for the use of the individual or entity to whom it is addressed and others  authorized to receive it. It may contain confidential or legally privileged information. If you are not the intended recipient you are hereby notified that any disclosure, copying,distribution or taking any action in reliance on the contents of this information is strictly prohibited and is unlawful. If you have received this communication in error, please notify me immediately by responding to this email and then delete it from your system. Please be aware that all email may be subject to recording and/or monitoring in accordance with relevant legislation.

Paul de Groot

unread,
Apr 6, 2022, 6:26:54 AM4/6/22
to us...@opendtect.org
Dear Mr. Naik,

UVQ is short for Unsupervised Vector Quantizer. The supervised version of UVQ is called LVQ (Linear Vector Quantizer). This model is not supported in OpendTect.

In the old Neural Networks plugin, which is nowadays launched from the ML control Center, you can do a nonlinear supervised waveform classification with a Multi-Layer-Perceptron. The workflow is as follows:
  • Create pointsets for all N classes
  • Create an attribute set that extracts wavelet samples (Open the Default Attribute Set: "Workflow - Unsupervised Waveform Segmentation")
  • Open the Neural Networks plugin from the ML control center
  • Select Pattern Recognition (Pointsets ...)
  • Select Supervised, Input the waveform attributes, output the N pointsets
  • Train and Apply
Alternative linear and nonlinear classifications are possible in the Machine Learning plugin. A linear classification can be done with a Support Vector Machine (SVM). This is a model in Scikit Learn in the new Machine Learning plugin. A SVM classifies the data by constructing hyperplanes in data space. If you select the linear kernel, the hyperplane will be a linear plane. The workflow for this is:
  • Create pointsets for all N classes
  • Select the "Seismic Classification (Supervised)" workflow from the Seismic tab in the ML control center
  • Select all pointsets as inputs
  • Select the waveform from the input seismic
  • Select Scikit Learn as Machine Learning Platform
  • Select Type SVM (Support Vector Model) and kernel is Linear
  • Train and Apply
Best regards,

Paul.

--
Paul de Groot
Geoscience Manager


-----------------------------------------------------------------------------------------------------------------------  
dGB Earth Sciences
Phone:+31 53 4315155
E-mail:paul.d...@dgbes.com
Internet:dgbes.com 
----------------------------------------------------------------------------------------------------------------------- 


--
You received this message because you are subscribed to the Google Groups "OpendTect Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to users+un...@opendtect.org.
To view this discussion on the web visit https://groups.google.com/a/opendtect.org/d/msgid/users/CABbBd9rfuoqMYRLkMM_uA0_NtNXRi3XUxrGSHJ5j%2Bkup3zMXxQ%40mail.gmail.com.
Reply all
Reply to author
Forward
0 new messages