update README with FWI example

vkazei · May 16, 2020 · a5b5fc7 · a5b5fc7
1 parent f9fac7c
commit a5b5fc7
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 0 deletions.
diff --git a/README.md b/README.md
@@ -32,6 +32,11 @@ We generate a set of pseudo-random models for training by cropping and skewing:
 Velocity model is then retrieved as an assembly of depth profiles. Deep learning models are naturally stochastic, so we train as set of five to provide initial uncertainty estimates:
 ![cmp_to_log](latex/Fig/inverted_models_EAGE2020.png)
 
+New training data is generated on-the-fly to save space and boost generalization power.
+Full-waveform inversion can help refine the results.
+![cmp_to_log](latex/Fig/dynamic_overthrust.png)
+
+
 
 
 

diff --git a/latex/Fig/dynamic_overthrust.png b/latex/Fig/dynamic_overthrust.png