DataSet synthetic generator
for each tx (3 x 5 x 1):
sample ps-transrot
-1 0 1 2
sample ps-sagrot
-1 0 1 2
sample ps-corrot 0
90% 0,0,0
sample mac
xlate table dir
+1 90%
sample mac
rotate table dir
+1 90%
sample protocol: period0,…
90% 3 90%
sample init table shift
random(-10..10)
rt plan encoding this
- can be imported
in musiq
for each session (3 x 5 x 9):
init from protocol
sample 6 DoF xlate rot
OR Id if no image
gaussian
calc table shift from
init
6DoF+pp+table dir
+ noise
sro+tx record
to encode this
Each sample is 3x5x10
MLP 3 layer to calculate energy
3x50
> 6x24
> 12x12
> 1
Train!
Train MLP model with langevin dynamic using https://colab.research.google.com/drive/1QtqPGTUUx7OWvpod0Lm6F78utK8WZnCE#scrollTo=Zy44zat-GlIC
DataSet synthetic generator
for each tx (3 x 5 x 1):
for each session (3 x 5 x 9):
Each sample is 3x5x10
MLP 3 layer to calculate energy
3x50
> 6x24
> 12x12
> 1
Train!
Train MLP model with langevin dynamic using https://colab.research.google.com/drive/1QtqPGTUUx7OWvpod0Lm6F78utK8WZnCE#scrollTo=Zy44zat-GlIC