summaryrefslogtreecommitdiff
path: root/ot/bregman.py
diff options
context:
space:
mode:
authorRémi Flamary <remi.flamary@gmail.com>2016-10-28 12:50:54 +0200
committerRémi Flamary <remi.flamary@gmail.com>2016-10-28 12:50:54 +0200
commit9523b1e95ed7c117554ff673532d150841092137 (patch)
tree1ed5b5060b3b59e79ca2e286058162cdd51c3b75 /ot/bregman.py
parentf33087d2b1790dac773782bb0d91bcfe7ce6a079 (diff)
doc datasets.py
Diffstat (limited to 'ot/bregman.py')
-rw-r--r--ot/bregman.py14
1 files changed, 4 insertions, 10 deletions
diff --git a/ot/bregman.py b/ot/bregman.py
index 9183bea..ad9a67a 100644
--- a/ot/bregman.py
+++ b/ot/bregman.py
@@ -276,14 +276,6 @@ def unmix(a,D,M,M0,h0,reg,reg0,alpha,numItermax = 1000, stopThr=1e-3,verbose=Fal
The optimization problem is solved suing the algorithm described in [4]
-
- distrib : distribution to unmix
- D : Dictionnary
- M : Metric matrix in the space of the distributions to unmix
- M0 : Metric matrix in the space of the 'abundance values' to solve for
- h0 : prior on solution (generally uniform distribution)
- reg,reg0 : transport regularizations
- alpha : how much should we trust the prior ? ([0,1])
Parameters
----------
@@ -300,7 +292,9 @@ def unmix(a,D,M,M0,h0,reg,reg0,alpha,numItermax = 1000, stopThr=1e-3,verbose=Fal
reg: float
Regularization term >0 (Wasserstein data fitting)
reg0: float
- Regularization term >0 (Wasserstein reg with h0)
+ Regularization term >0 (Wasserstein reg with h0)
+ alpha: float
+ How much should we trust the prior ([0,1])
numItermax: int, optional
Max number of iterations
stopThr: float, optional
@@ -318,7 +312,7 @@ def unmix(a,D,M,M0,h0,reg,reg0,alpha,numItermax = 1000, stopThr=1e-3,verbose=Fal
log: dict
log dictionary return only if log==True in parameters
- References
+ References
----------
.. [4] S. Nakhostin, N. Courty, R. Flamary, D. Tuia, T. Corpetti, Supervised planetary unmixing with optimal transport, Whorkshop on Hyperspectral Image and Signal Processing : Evolution in Remote Sensing (WHISPERS), 2016.