Skip to content

[MRG] fix grad sign sr(F)GW with KL loss #611

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Mar 4, 2024
Merged

[MRG] fix grad sign sr(F)GW with KL loss #611

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8e67b40
fix grad sign
cedricvincentcuaz Mar 4, 2024
dc6c3d8
mrg
cedricvincentcuaz Mar 4, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions RELEASES.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,8 @@
- Fix doc and example for lowrank sinkhorn (PR #601)
- Fix issue with empty weights for `ot.emd2` (PR #606, Issue #534)
- Fix a sign error regarding the gradient of `ot.gromov._gw.fused_gromov_wasserstein2` and `ot.gromov._gw.gromov_wasserstein2` for the kl loss (PR #610)
- Fix same sign error for sr(F)GW conditional gradient solvers (PR #611)


## 0.9.2
*December 2023*
Expand Down
4 changes: 2 additions & 2 deletions ot/gromov/_semirelaxed.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -250,7 +250,7 @@ def semirelaxed_gromov_wasserstein2(C1, C2, p=None, loss_fun='square_loss', symm

elif loss_fun == 'kl_loss':
gC1 = nx.log(C1 + 1e-15) * nx.outer(p, p) - nx.dot(T, nx.dot(nx.log(C2 + 1e-15), T.T))
gC2 = nx.dot(T.T, nx.dot(C1, T)) / (C2 + 1e-15) + nx.outer(q, q)
gC2 = - nx.dot(T.T, nx.dot(C1, T)) / (C2 + 1e-15) + nx.outer(q, q)

srgw = nx.set_gradients(srgw, (C1, C2), (gC1, gC2))

Expand Down Expand Up @@ -509,7 +509,7 @@ def semirelaxed_fused_gromov_wasserstein2(M, C1, C2, p=None, loss_fun='square_lo

elif loss_fun == 'kl_loss':
gC1 = nx.log(C1 + 1e-15) * nx.outer(p, p) - nx.dot(T, nx.dot(nx.log(C2 + 1e-15), T.T))
gC2 = nx.dot(T.T, nx.dot(C1, T)) / (C2 + 1e-15) + nx.outer(q, q)
gC2 = - nx.dot(T.T, nx.dot(C1, T)) / (C2 + 1e-15) + nx.outer(q, q)

if isinstance(alpha, int) or isinstance(alpha, float):
srfgw_dist = nx.set_gradients(srfgw_dist, (C1, C2, M),
Expand Down