"""
Quantized (Fused) Gromov-Wasserstein solvers.
"""
# Author: Cédric Vincent-Cuaz <cedvincentcuaz@gmail.com>
#
# License: MIT License
import numpy as np
import warnings
try:
from networkx.algorithms.community import asyn_fluidc, louvain_communities
from networkx import from_numpy_array, pagerank
networkx_import = True
except ImportError:
networkx_import = False
try:
from sklearn.cluster import SpectralClustering, KMeans
sklearn_import = True
except ImportError:
sklearn_import = False
import random
from ..utils import list_to_array, unif, dist
from ..backend import get_backend
from ..lp import emd_1d
from ._gw import gromov_wasserstein, fused_gromov_wasserstein
from ._utils import init_matrix, gwloss
[docs]
def quantized_fused_gromov_wasserstein_partitioned(
CR1, CR2, list_R1, list_R2, list_p1, list_p2, MR=None,
alpha=1., build_OT=False, log=False, armijo=False, max_iter=1e4,
tol_rel=1e-9, tol_abs=1e-9, nx=None, **kwargs):
r"""
Returns the quantized Fused Gromov-Wasserstein transport between
:math:`(\mathbf{C_1}, \mathbf{F_1}, \mathbf{p})` and :math:`(\mathbf{C_2},
\mathbf{F_2}, \mathbf{q})`, whose samples are assigned to partitions and representants
:math:`\mathcal{P_1} = \{(\mathbf{P_{1, i}}, \mathbf{r_{1, i}})\}_{i \leq npart1}`
and :math:`\mathcal{P_2} = \{(\mathbf{P_{2, j}}, \mathbf{r_{2, j}})\}_{j \leq npart2}`.
The latter must be precomputed and encoded e.g for the source as: :math:`\mathbf{CR_1}`
structure matrix between representants; `list_R1` a list of relations between
representants and their associated samples; `list_p1` a list of nodes
distribution within each partition; :math:`\mathbf{FR_1}` feature matrix
of representants.
The function estimates the following optimization problem:
.. math::
\mathbf{T}^* \in \mathop{\arg \min}_\mathbf{T} \quad \alpha \sum_{i,j,k,l}
L(\mathbf{C_1}_{i,k}, \mathbf{C_2}_{j,l}) \mathbf{T}_{i,j} \mathbf{T}_{k,l}
+ (1-\alpha) \langle \mathbf{T}, M\rangle_F
s.t. \ \mathbf{T} \mathbf{1} &= \mathbf{p}
\mathbf{T}^T \mathbf{1} &= \mathbf{q}
\mathbf{T} &\geq 0
\mathbf{T}_{|\mathbf{P_{1, i}}, \mathbf{P_{2, j}}} &= T^{g}_{ij} \mathbf{T}^{(i,j)}
using a two-step strategy computing: i) a global alignment :math:`\mathbf{T}^{g}`
between representants joint structure and feature spaces; ii) local alignments
:math:`\mathbf{T}^{(i, j)}` between partitions :math:`\mathbf{P_{1, i}}`
and :math:`\mathbf{P_{2, j}}` seen as 1D measures.
Where :
- :math:`\mathbf{C_1}`: Metric cost matrix in the source space
- :math:`\mathbf{C_2}`: Metric cost matrix in the target space
- :math:`\mathbf{F_1}`: Feature matrix in the source space
- :math:`\mathbf{F_2}`: Feature matrix in the target space
- :math:`\mathbf{M}`: Pairwise similarity matrix between features
- :math:`\mathbf{p}`: distribution in the source space
- :math:`\mathbf{q}`: distribution in the target space
- :math:`L`: quadratic loss function to account for the misfit between the similarity matrices
.. note:: This function is backend-compatible and will work on arrays
from all compatible backends. But the algorithm uses the C++ CPU backend
which can lead to copy overhead on GPU arrays.
.. note:: All computations in the Gromov-Wasserstein conjugate gradient solver
are done with numpy to limit memory overhead.
Parameters
----------
CR1 : array-like, shape (npart1, npart1)
Structure matrix between partition representants in the source space.
CR2 : array-like, shape (npart2, npart2)
Structure matrix between partition representants in the target space.
list_R1 : list of npart1 arrays,
List of relations between representants and their associated samples in the source space.
list_R2 : list of npart2 arrays,
List of relations between representants and their associated samples in the target space.
list_p1 : list of npart1 arrays,
List of node distributions within each partition of the source space.
list_p : list of npart2 arrays,
List of node distributions within each partition of the target space.
MR : array-like, shape (npart1, npart2), optional. (Default is None)
Metric cost matrix between features of representants across spaces.
alpha: float, optional. Default is None.
FGW trade-off parameter in :math:`]0, 1]` between structure and features.
If `alpha = 1` features are ignored hence computing qGW.
build_OT: bool, optional. Default is False
Either to build or not the OT between non-partitioned structures.
log : bool, optional. Default is False
record log if True
armijo : bool, optional
If True the step of the line-search is found via an armijo research. Else closed form is used.
If there are convergence issues use False.
max_iter : int, optional
Max number of iterations
tol_rel : float, optional
Stop threshold on relative error (>0)
tol_abs : float, optional
Stop threshold on absolute error (>0)
nx : backend, optional
POT backend
**kwargs : dict
parameters can be directly passed to the ot.optim.cg solver
Returns
-------
T_global: array-like, shape (`npart1`, `npart2`)
Gromov-Wasserstein alignment :math:`\mathbf{T}^{g}` between representants.
Ts_local: dict of local OT matrices.
Dictionary with keys :math:`(i, j)` corresponding to 1D OT between
:math:`\mathbf{P_{1, i}}` and :math:`\mathbf{P_{2, j}}` if :math:`T^{g}_{ij} \neq 0`.
T: array-like, shape `(ns, nt)`
Coupling between the two spaces if `build_OT=True` else None.
log : dict, if `log=True`.
Convergence information and losses of inner OT problems.
References
----------
.. [68] Chowdhury, S., Miller, D., & Needham, T. (2021).
Quantized gromov-wasserstein. ECML PKDD 2021. Springer International Publishing.
"""
if nx is None:
arr = [CR1, CR2, *list_R1, *list_R2, *list_p1, *list_p2]
if MR is not None:
arr.append(MR)
nx = get_backend(*arr)
npart1 = len(list_R1)
npart2 = len(list_R2)
# compute marginals for global alignment
pR1 = nx.from_numpy(list_to_array([nx.sum(p) for p in list_p1]))
pR2 = nx.from_numpy(list_to_array([nx.sum(q) for q in list_p2]))
# compute global alignment
if alpha == 1.:
res_global = gromov_wasserstein(
CR1, CR2, pR1, pR2, loss_fun='square_loss', log=log,
armijo=armijo, max_iter=max_iter, tol_rel=tol_rel, tol_abs=tol_abs)
if log:
T_global, dist_global = res_global[0], res_global[1]['gw_dist']
else:
T_global = res_global
elif (alpha < 1.) and (alpha > 0.):
res_global = fused_gromov_wasserstein(
MR, CR1, CR2, pR1, pR2, 'square_loss', alpha=alpha, log=log,
armijo=armijo, max_iter=max_iter, tol_rel=tol_rel, tol_abs=tol_abs)
if log:
T_global, dist_global = res_global[0], res_global[1]['fgw_dist']
else:
T_global = res_global
else:
raise ValueError(
f"""
`alpha='{alpha}'` should be in ]0, 1].
""")
if log:
log_ = {}
log_['global dist'] = dist_global
# compute local alignments
Ts_local = {}
list_p1_norm = [p / nx.sum(p) for p in list_p1]
list_p2_norm = [q / nx.sum(q) for q in list_p2]
for i in range(npart1):
for j in range(npart2):
if T_global[i, j] != 0.:
res_1d = emd_1d(list_R1[i], list_R2[j], list_p1_norm[i], list_p2_norm[j],
metric='sqeuclidean', p=1., log=log)
if log:
T_local, log_local = res_1d
Ts_local[(i, j)] = T_local
log_[f'local dist ({i},{j})'] = log_local['cost']
else:
Ts_local[(i, j)] = res_1d
if build_OT:
T_rows = []
for i in range(npart1):
list_Ti = []
for j in range(npart2):
if T_global[i, j] == 0.:
T_local = nx.zeros((list_R1[i].shape[0], list_R2[j].shape[0]), type_as=T_global)
else:
T_local = T_global[i, j] * Ts_local[(i, j)]
list_Ti.append(T_local)
Ti = nx.concatenate(list_Ti, axis=1)
T_rows.append(Ti)
T = nx.concatenate(T_rows, axis=0)
else:
T = None
if log:
return T_global, Ts_local, T, log_
else:
return T_global, Ts_local, T
[docs]
def get_graph_partition(C, npart, part_method='random', F=None, alpha=1.,
random_state=0, nx=None):
"""
Partitioning a given graph with structure matrix :math:`\mathbf{C} \in R^{n \times n}`
into `npart` partitions either 'random', or using one of {'louvain', 'fluid'}
algorithms from networkx, or 'spectral' clustering from scikit-learn,
or (Fused) Gromov-Wasserstein projections from POT.
Parameters
----------
C : array-like, shape (n, n)
Structure matrix.
npart : int,
number of partitions/clusters smaller than the number of nodes in
:math:`\mathbf{C}`.
part_method : str, optional. Default is 'random'.
Partitioning algorithm to use among {'random', 'louvain', 'fluid', 'spectral', 'GW', 'FGW'}.
'random' for random sampling of points; 'louvain' and 'fluid' for graph
partitioning algorithm that works well on adjacency matrix, If the
louvain algorithm is used, `npart` is ignored; 'spectral' for spectral
clustering; '(F)GW' for (F)GW projection using sr(F)GW solvers.
F : array-like, shape (n, d), optional. (Default is None)
Optional feature matrix aligned with the graph structure. Only used if
`part_method="FGW"`.
alpha : float, optional. (Default is 1.)
Trade-off parameter between feature and structure matrices, taking
values in [0, 1] and only used if `F != None` and `part_method="FGW"`.
random_state: int, optional
Random seed for the partitioning algorithm.
nx : backend, optional
POT backend.
Returns
-------
part : array-like, shape (npart,)
Array of partition assignment for each node.
References
----------
.. [68] Chowdhury, S., Miller, D., & Needham, T. (2021).
Quantized gromov-wasserstein. ECML PKDD 2021. Springer International Publishing.
"""
if nx is None:
nx = get_backend(C)
n = C.shape[0]
C0 = C
if (alpha != 1.) and (F is None):
raise ValueError("`alpha != 1` but node features are not provided.")
if npart >= n:
warnings.warn(
"Requested number of partitions higher than the number of nodes"
"hence we enforce each node to be a partition.",
stacklevel=2
)
part = np.arange(n)
elif npart == 1:
part = np.zeros(n)
elif part_method == 'random':
# randomly partition the space
random.seed(random_state)
part = list_to_array(random.choices(np.arange(npart), k=C.shape[0]))
elif part_method == 'louvain':
C = nx.to_numpy(C0)
graph = from_numpy_array(C)
part_sets = louvain_communities(graph, seed=random_state)
part = np.zeros(n)
for iset_, set_ in enumerate(part_sets):
set_ = list(set_)
part[set_] = iset_
elif part_method == 'fluid':
C = nx.to_numpy(C0)
graph = from_numpy_array(C)
part_sets = asyn_fluidc(graph, npart, seed=random_state)
part = np.zeros(n)
for iset_, set_ in enumerate(part_sets):
set_ = list(set_)
part[set_] = iset_
elif part_method == 'spectral':
C = nx.to_numpy(C0)
sc = SpectralClustering(n_clusters=npart,
random_state=random_state,
affinity='precomputed').fit(C)
part = sc.labels_
elif part_method in ['GW', 'FGW']:
raise ValueError(f"`part_method == {part_method}` not implemented yet.")
else:
raise ValueError(
f"""
Unknown `part_method='{part_method}'`. Use one of:
{'random', 'louvain', 'fluid', 'spectral', 'GW', 'FGW'}.
""")
return nx.from_numpy(part, type_as=C0)
[docs]
def get_graph_representants(C, part, rep_method='pagerank', random_state=0, nx=None):
"""
Get representative node for each partition given by :math:`\mathbf{part} \in R^{n}`
of a graph with structure matrix :math:`\mathbf{C} \in R^{n \times n}`.
Selection is either done randomly or using 'pagerank' algorithm from networkx.
Parameters
----------
C : array-like, shape (n, n)
structure matrix.
part : array-like, shape (n,)
Array of partition assignment for each node.
rep_method : str, optional. Default is 'pagerank'.
Selection method for representant in each partition. Can be either 'random'
i.e random sampling within each partition, or 'pagerank' to select a
node with maximal pagerank.
random_state: int, optional
Random seed for the partitioning algorithm
nx : backend, optional
POT backend
Returns
-------
rep_indices : list, shape (npart,)
indices for representative node of each partition sorted
according to partition identifiers.
References
----------
.. [68] Chowdhury, S., Miller, D., & Needham, T. (2021).
Quantized gromov-wasserstein. ECML PKDD 2021. Springer International Publishing.
"""
if nx is None:
nx = get_backend(C, part)
rep_indices = []
part_ids = nx.unique(part)
n_part_ids = part_ids.shape[0]
if n_part_ids == C.shape[0]:
rep_indices = nx.arange(n_part_ids)
elif rep_method == 'random':
random.seed(random_state)
for id_, part_id in enumerate(part_ids):
indices = nx.where(part == part_id)[0]
rep_indices.append(random.choice(indices))
elif rep_method == 'pagerank':
C0, part0 = C, part
C = nx.to_numpy(C0)
part = nx.to_numpy(part0)
part_ids = np.unique(part)
for id_ in part_ids:
indices = np.where(part == id_)[0]
C_id = C[indices, :][:, indices]
graph = from_numpy_array(C_id)
pagerank_values = list(pagerank(graph).values())
rep_idx = np.argmax(pagerank_values)
rep_indices.append(indices[rep_idx])
else:
raise ValueError(
f"""
Unknown `rep_method='{rep_method}'`. Use one of:
{'random', 'pagerank'}.
""")
return rep_indices
[docs]
def quantized_fused_gromov_wasserstein(
C1, C2, npart1, npart2, p=None, q=None, C1_aux=None, C2_aux=None,
F1=None, F2=None, alpha=1., part_method='fluid',
rep_method='random', log=False, armijo=False, max_iter=1e4,
tol_rel=1e-9, tol_abs=1e-9, random_state=0, **kwargs):
r"""
Returns the quantized Fused Gromov-Wasserstein transport between
:math:`(\mathbf{C_1}, \mathbf{F_1}, \mathbf{p})` and :math:`(\mathbf{C_2},
\mathbf{F_2}, \mathbf{q})`, whose samples are assigned to partitions and
representants :math:`\mathcal{P_1} = \{(\mathbf{P_{1, i}}, \mathbf{r_{1, i}})\}_{i \leq npart1}`
and :math:`\mathcal{P_2} = \{(\mathbf{P_{2, j}}, \mathbf{r_{2, j}})\}_{j \leq npart2}`.
The function estimates the following optimization problem:
.. math::
\mathbf{T}^* \in \mathop{\arg \min}_\mathbf{T} \quad \alpha \sum_{i,j,k,l}
L(\mathbf{C_1}_{i,k}, \mathbf{C_2}_{j,l}) \mathbf{T}_{i,j} \mathbf{T}_{k,l}
+ (1-\alpha) \langle \mathbf{T}, \mathbf{D}(\mathbf{F_1}, \mathbf{F}_2) \rangle_F
s.t. \ \mathbf{T} \mathbf{1} &= \mathbf{p}
\mathbf{T}^T \mathbf{1} &= \mathbf{q}
\mathbf{T} &\geq 0
\mathbf{T}_{|\mathbf{P_{1, i}}, \mathbf{P_{2, j}}} &= T^{g}_{ij} \mathbf{T}^{(i,j)}
using a two-step strategy computing: i) a global alignment :math:`\mathbf{T}^{g}`
between representants across joint structure and feature spaces;
ii) local alignments :math:`\mathbf{T}^{(i, j)}` between partitions
:math:`\mathbf{P_{1, i}}` and :math:`\mathbf{P_{2, j}}` seen as 1D measures.
Where :
- :math:`\mathbf{C_1}`: Metric cost matrix in the source space
- :math:`\mathbf{C_2}`: Metric cost matrix in the target space
- :math:`\mathbf{F_1}`: Feature matrix in the source space
- :math:`\mathbf{F_2}`: Feature matrix in the target space
- :math:`\mathbf{D}(\mathbf{F_1}, \mathbf{F_2})`: Pairwise euclidean distance matrix between features
- :math:`\mathbf{p}`: distribution in the source space
- :math:`\mathbf{q}`: distribution in the target space
- :math:`L`: quadratic loss function to account for the misfit between the similarity matrices
.. note:: This function is backend-compatible and will work on arrays
from all compatible backends. But the algorithm uses the C++ CPU backend
which can lead to copy overhead on GPU arrays.
.. note:: All computations in the conjugate gradient solver are done with
numpy to limit memory overhead.
Parameters
----------
C1 : array-like, shape (ns, ns)
Structure matrix in the source space.
C2 : array-like, shape (nt, nt)
Structure matrix in the target space.
npart1 : int,
number of partition in the source space.
npart2 : int,
number of partition in the target space.
p : array-like, shape (ns,), optional
Distribution in the source space.
If let to its default value None, uniform distribution is taken.
q : array-like, shape (nt,), optional
Distribution in the target space.
If let to its default value None, uniform distribution is taken.
C1_aux : array-like, shape (ns, ns), optional. Default is None.
Auxiliary structure matrix in the source space to perform the partitioning
and representant selection.
C2_aux : array-like, shape (nt, nt), optional. Default is None.
Auxiliary structure matrix in the target space to perform the partitioning
and representant selection.
F1 : array-like, shape (ns, d), optional. Default is None.
Feature matrix in the source space.
F2 : array-like, shape (nt, d), optional. Default is None.
Feature matrix in the target space
alpha: float, optional. Default is 1.
FGW trade-off parameter in :math:`]0, 1]` between structure and features.
If `alpha = 1` features are ignored hence computing qGW, if `alpha=0`
structures are ignored and we compute the quantized Wasserstein transport.
part_method : str, optional. Default is 'spectral'.
Partitioning algorithm to use among {'random', 'louvain', 'fluid',
'spectral', 'louvain_fused', 'fluid_fused', 'spectral_fused', 'GW', 'FGW'}.
If part_method in {'louvain_fused', 'fluid_fused', 'spectral_fused'},
corresponding graph partitioning algorithm {'louvain', 'fluid', 'spectral'}
will be used on the modified structure matrix
:math:`\alpha \mathbf{C} + (1 - \alpha) \mathbf{D}(\mathbf{F})` where
:math:`\mathbf{D}(\mathbf{F})` is the pairwise euclidean matrix between features.
If part_method in {'GW', 'FGW'}, a (F)GW projection is used.
If the louvain algorithm is used, the requested number of partitions is
ignored.
rep_method : str, optional. Default is 'pagerank'.
Selection method for node representant in each partition.
Can be either 'random' i.e random sampling within each partition,
{'pagerank', 'pagerank_fused'} to select a node with maximal pagerank w.r.t
:math:`\mathbf{C}` or :math:`\alpha \mathbf{C} + (1 - \alpha) \mathbf{D}(\mathbf{F})`.
verbose : bool, optional
Print information along iterations
log : bool, optional
record log if True
armijo : bool, optional
If True the step of the line-search is found via an armijo research. Else closed form is used.
If there are convergence issues use False.
max_iter : int, optional
Max number of iterations
tol_rel : float, optional
Stop threshold on relative error (>0)
tol_abs : float, optional
Stop threshold on absolute error (>0)
random_state: int, optional
Random seed for the partitioning algorithm
**kwargs : dict
parameters can be directly passed to the ot.optim.cg solver
Returns
-------
T_global: array-like, shape (`npart1`, `npart2`)
Fused Gromov-Wasserstein alignment :math:`\mathbf{T}^{g}` between representants.
Ts_local: dict of local OT matrices.
Dictionary with keys :math:`(i, j)` corresponding to 1D OT between
:math:`\mathbf{P_{1, i}}` and :math:`\mathbf{P_{2, j}}` if :math:`T^{g}_{ij} \neq 0`.
T: array-like, shape `(ns, nt)`
Coupling between the two spaces.
log : dict
Convergence information for inner problems and qGW loss.
References
----------
.. [68] Chowdhury, S., Miller, D., & Needham, T. (2021).
Quantized gromov-wasserstein. ECML PKDD 2021. Springer International Publishing.
"""
if (part_method in ['fluid', 'louvain', 'fluid_fused', 'louvain_fused'] or (rep_method in ['pagerank', 'pagerank_fused'])):
if not networkx_import:
warnings.warn(
f"""
Networkx is not installed, so part_method={part_method} and/or
rep_method={rep_method} cannot be used and are set to `random`
default methods. Consider installing Networkx to fix this.
"""
)
part_method = 'random'
rep_method = 'random'
if (part_method in ['spectral', 'spectral_fused']) and (not sklearn_import):
warnings.warn(
f"""
Scikit-learn is not installed, so part_method={part_method} and/or
rep_method={rep_method} cannot be used and are set to `random`
default methods. Consider installing Scikit-learn to fix this.
"""
)
part_method = 'random'
rep_method = 'random'
if (('fused' in part_method) or ('fused' in rep_method) or (part_method == 'FGW')):
if (F1 is None) or (F2 is None):
raise ValueError(
f"""
`part_method='{part_method}'` and/or `rep_method='{rep_method}'`
require feature matrices which are not provided as inputs.
""")
arr = [C1, C2]
if C1_aux is not None:
arr.append(C1_aux)
else:
C1_aux = C1
if C2_aux is not None:
arr.append(C2_aux)
else:
C2_aux = C2
if p is not None:
arr.append(list_to_array(p))
else:
p = unif(C1.shape[0], type_as=C1)
if q is not None:
arr.append(list_to_array(q))
else:
q = unif(C2.shape[0], type_as=C1)
if F1 is not None:
arr.append(F1)
if F2 is not None:
arr.append(F1)
nx = get_backend(*arr)
DF1 = None
DF2 = None
# compute attributed graph partitions potentially using the auxiliary structure
if 'fused' in part_method:
DF1 = dist(F1, F1)
DF2 = dist(F2, F2)
C1_new = alpha * C1_aux + (1 - alpha) * DF1
C2_new = alpha * C2_aux + (1 - alpha) * DF2
part_method_ = part_method[:-6]
part1 = get_graph_partition(C1_new, npart1, part_method_, random_state=random_state, nx=nx)
part2 = get_graph_partition(C2_new, npart2, part_method_, random_state=random_state, nx=nx)
else:
part1 = get_graph_partition(C1_aux, npart1, part_method, F1, alpha, random_state, nx)
part2 = get_graph_partition(C2_aux, npart2, part_method, F2, alpha, random_state, nx)
if 'fused' in rep_method:
if DF1 is None:
DF1 = dist(F1, F1)
DF2 = dist(F2, F2)
C1_new = alpha * C1_aux + (1 - alpha) * DF1
C2_new = alpha * C2_aux + (1 - alpha) * DF2
rep_method_ = rep_method[:-6]
rep_indices1 = get_graph_representants(C1_new, part1, rep_method_, random_state, nx)
rep_indices2 = get_graph_representants(C2_new, part2, rep_method_, random_state, nx)
else:
rep_indices1 = get_graph_representants(C1_aux, part1, rep_method, random_state, nx)
rep_indices2 = get_graph_representants(C2_aux, part2, rep_method, random_state, nx)
# format partitions over (C1, F1) and (C2, F2)
if (F1 is None) and (F2 is None):
CR1, list_R1, list_p1 = format_partitioned_graph(C1, p, part1, rep_indices1, nx=nx)
CR2, list_R2, list_p2 = format_partitioned_graph(C2, q, part2, rep_indices2, nx=nx)
MR = None
else:
if DF1 is None:
DF1 = dist(F1, F1)
DF2 = dist(F2, F2)
CR1, list_R1, list_p1, FR1 = format_partitioned_graph(C1, p, part1, rep_indices1, F1, DF1, alpha, nx)
CR2, list_R2, list_p2, FR2 = format_partitioned_graph(C2, q, part2, rep_indices2, F2, DF2, alpha, nx)
MR = dist(FR1, FR2)
# call to partitioned quantized fused gromov-wasserstein solver
res = quantized_fused_gromov_wasserstein_partitioned(
CR1, CR2, list_R1, list_R2, list_p1, list_p2, MR, alpha, build_OT=True,
log=log, armijo=armijo, max_iter=max_iter, tol_rel=tol_rel,
tol_abs=tol_abs, nx=nx, **kwargs)
if log:
T_global, Ts_local, T, log_ = res
# compute the transport cost on structures
constC, hC1, hC2 = init_matrix(C1, C2, p, q, 'square_loss', nx)
structure_cost = gwloss(constC, hC1, hC2, T, nx)
if alpha != 1.:
M = dist(F1, F2)
feature_cost = nx.sum(M * T)
else:
feature_cost = 0.
log_['qFGW_dist'] = alpha * structure_cost + (1 - alpha) * feature_cost
return T_global, Ts_local, T, log_
else:
T_global, Ts_local, T = res
return T_global, Ts_local, T
[docs]
def get_partition_and_representants_samples(
X, npart, method='kmeans', random_state=0, nx=None):
"""
Compute `npart` partitions and representants over samples :math:`\mathbf{X} \in R^{n \times d}`
using either a random or a kmeans algorithm.
Parameters
----------
X : array-like, shape (n, d)
Samples endowed with an euclidean geometry.
npart : int,
number of partitions smaller than the number of samples in
:math:`\mathbf{X}`.
method : str, optional. Default is 'kmeans'.
Partitioning and representant selection algorithms to use among
{'random', 'kmeans'}. 'random' for random sampling of points; 'kmeans'
for k-means clustering using scikit-learn implementation where closest
points to centroids are considered as representants.
random_state: int, optional
Random seed for the partitioning algorithm.
nx : backend, optional
POT backend.
Returns
-------
part : array-like, shape (npart,)
Array of partition assignment for each node.
rep_indices : list, shape (npart,)
indices for representative node of each partition sorted
according to partition identifiers.
References
----------
.. [68] Chowdhury, S., Miller, D., & Needham, T. (2021).
Quantized gromov-wasserstein. ECML PKDD 2021. Springer International Publishing.
"""
if nx is None:
nx = get_backend(X)
n = X.shape[0]
X0 = X
if npart >= n:
warnings.warn(
"Requested number of partitions higher than the number of nodes"
"hence we enforce each node to be a partition.",
stacklevel=2
)
part = nx.arange(n)
rep_indices = nx.arange(n)
elif npart == 1:
random.seed(random_state)
part = nx.zeros(n)
rep_indices = [random.choice(nx.arange(n))]
elif method == 'random':
# randomly partition the space
random.seed(random_state)
part = list_to_array(random.choices(np.arange(npart), k=X.shape[0]))
part = nx.from_numpy(part, type_as=X0)
# randomly select representant in each partition
rep_indices = []
part_ids = nx.unique(part)
for id_, part_id in enumerate(part_ids):
indices = nx.where(part == part_id)[0]
rep_indices.append(random.choice(indices))
elif method == 'kmeans':
X = nx.to_numpy(X0)
km = KMeans(n_clusters=npart, random_state=random_state).fit(X)
part = nx.from_numpy(km.labels_, type_as=X0)
rep_indices = []
for part_id in range(npart):
indices = nx.where(part == part_id)[0]
dists = dist(X[indices], km.cluster_centers_[part_id][None, :])
best_idx = indices[dists.argmin()]
rep_indices.append(best_idx)
else:
raise ValueError(
f"""
Unknown `method='{method}'`. Use one of: {'random', 'kmeans'}
""")
return part, rep_indices
[docs]
def quantized_fused_gromov_wasserstein_samples(
X1, X2, npart1, npart2, p=None, q=None, F1=None, F2=None, alpha=1.,
method='kmeans', log=False, armijo=False, max_iter=1e4,
tol_rel=1e-9, tol_abs=1e-9, random_state=0, **kwargs):
r"""
Returns the quantized Fused Gromov-Wasserstein transport between samples
endowed with their respective euclidean geometry :math:`(\mathbf{D}(\mathbf{X_1}), \mathbf{F_1}, \mathbf{p})`
and :math:`(\mathbf{D}(\mathbf{X_1}), \mathbf{F_2}, \mathbf{q})`, whose samples are assigned to partitions and
representants :math:`\mathcal{P_1} = \{(\mathbf{P_{1, i}}, \mathbf{r_{1, i}})\}_{i \leq npart1}`
and :math:`\mathcal{P_2} = \{(\mathbf{P_{2, j}}, \mathbf{r_{2, j}})\}_{j \leq npart2}`.
The function estimates the following optimization problem:
.. math::
\mathbf{T}^* \in \mathop{\arg \min}_\mathbf{T} \quad \alpha \sum_{i,j,k,l}
L(\mathbf{D}(\mathbf{X_1})_{i,k}, \mathbf{D}(\mathbf{X_2})_{j,l}) \mathbf{T}_{i,j} \mathbf{T}_{k,l}
+ (1-\alpha) \langle \mathbf{T}, \mathbf{D}(\mathbf{F_1}, \mathbf{F}_2) \rangle_F
s.t. \ \mathbf{T} \mathbf{1} &= \mathbf{p}
\mathbf{T}^T \mathbf{1} &= \mathbf{q}
\mathbf{T} &\geq 0
\mathbf{T}_{|\mathbf{P_{1, i}}, \mathbf{P_{2, j}}} &= T^{g}_{ij} \mathbf{T}^{(i,j)}
using a two-step strategy computing: i) a global alignment :math:`\mathbf{T}^{g}`
between representants across joint structure and feature spaces;
ii) local alignments :math:`\mathbf{T}^{(i, j)}` between partitions
:math:`\mathbf{P_{1, i}}` and :math:`\mathbf{P_{2, j}}` seen as 1D measures.
Where :
- :math:`\mathbf{X_1}`: Samples in the source space
- :math:`\mathbf{X_2}`: Samples in the target space
- :math:`\mathbf{F_1}`: Feature matrix in the source space
- :math:`\mathbf{F_2}`: Feature matrix in the target space
- :math:`\mathbf{D}(\mathbf{F_1}, \mathbf{F_2})`: Pairwise euclidean distance matrix between features
- :math:`\mathbf{p}`: distribution in the source space
- :math:`\mathbf{q}`: distribution in the target space
- :math:`L`: quadratic loss function to account for the misfit between the similarity matrices
.. note:: This function is backend-compatible and will work on arrays
from all compatible backends. But the algorithm uses the C++ CPU backend
which can lead to copy overhead on GPU arrays.
.. note:: All computations in the conjugate gradient solver are done with
numpy to limit memory overhead.
Parameters
----------
X1 : array-like, shape (ns, ds)
Samples in the source space.
X2 : array-like, shape (nt, dt)
Samples in the target space.
npart1 : int,
number of partition in the source space.
npart2 : int,
number of partition in the target space.
p : array-like, shape (ns,), optional
Distribution in the source space.
If let to its default value None, uniform distribution is taken.
q : array-like, shape (nt,), optional
Distribution in the target space.
If let to its default value None, uniform distribution is taken.
F1 : array-like, shape (ns, d), optional. Default is None.
Feature matrix in the source space.
F2 : array-like, shape (nt, d), optional. Default is None.
Feature matrix in the target space
alpha: float, optional. Default is 1.
FGW trade-off parameter in :math:`]0, 1]` between structure and features.
If `alpha = 1` features are ignored hence computing qGW, if `alpha=0`
structures are ignored and we compute the quantized Wasserstein transport.
method : str, optional. Default is 'kmeans'.
Partitioning and representant selection algorithms to use among
{'random', 'kmeans', 'kmeans_fused'}.
If `part_method == 'kmeans_fused'`, kmeans is performed on augmented
samples :math:`[\alpha \mathbf{X}; (1 - \alpha) \mathbf{F}]`.
verbose : bool, optional
Print information along iterations
log : bool, optional
record log if True
armijo : bool, optional
If True the step of the line-search is found via an armijo research. Else closed form is used.
If there are convergence issues use False.
max_iter : int, optional
Max number of iterations
tol_rel : float, optional
Stop threshold on relative error (>0)
tol_abs : float, optional
Stop threshold on absolute error (>0)
random_state: int, optional
Random seed for the partitioning algorithm
**kwargs : dict
parameters can be directly passed to the ot.optim.cg solver
Returns
-------
T_global: array-like, shape (`npart1`, `npart2`)
Fused Gromov-Wasserstein alignment :math:`\mathbf{T}^{g}` between representants.
Ts_local: dict of local OT matrices.
Dictionary with keys :math:`(i, j)` corresponding to 1D OT between
:math:`\mathbf{P_{1, i}}` and :math:`\mathbf{P_{2, j}}` if :math:`T^{g}_{ij} \neq 0`.
T: array-like, shape `(ns, nt)`
Coupling between the two spaces.
log : dict
Convergence information for inner problems and qGW loss.
References
----------
.. [68] Chowdhury, S., Miller, D., & Needham, T. (2021).
Quantized gromov-wasserstein. ECML PKDD 2021. Springer International Publishing.
"""
if (method in ['kmeans', 'kmeans_fused']) and (not sklearn_import):
warnings.warn(
f"""
Scikit-learn is not installed, so method={method} cannot be used
and is set to `random` default methods. Consider installing
Scikit-learn to fix this.
"""
)
method = 'random'
if ('fused' in method) and ((F1 is None) or (F2 is None)):
raise ValueError(
f"""
`method='{method}'` requires feature matrices which are not provided as inputs.
""")
arr = [X1, X2]
if p is not None:
arr.append(list_to_array(p))
else:
p = unif(X1.shape[0], type_as=X1)
if q is not None:
arr.append(list_to_array(q))
else:
q = unif(X2.shape[0], type_as=X1)
if F1 is not None:
arr.append(F1)
if F2 is not None:
arr.append(F1)
nx = get_backend(*arr)
# compute attributed partitions and representants
if ('fused' in method) and (alpha != 1.):
X1_new = nx.concatenate([alpha * X1, (1 - alpha) * F1], axis=1)
X2_new = nx.concatenate([alpha * X2, (1 - alpha) * F2], axis=1)
method_ = method[:-6]
else:
X1_new, X2_new = X1, X2
method_ = method
part1, rep_indices1 = get_partition_and_representants_samples(
X1_new, npart1, method_, random_state, nx)
part2, rep_indices2 = get_partition_and_representants_samples(
X2_new, npart2, method_, random_state, nx)
# format partitions over (C1, F1) and (C2, F2)
if (F1 is None) and (F2 is None):
CR1, list_R1, list_p1 = format_partitioned_samples(
X1, p, part1, rep_indices1, nx=nx)
CR2, list_R2, list_p2 = format_partitioned_samples(
X2, q, part2, rep_indices2, nx=nx)
MR = None
else:
CR1, list_R1, list_p1, FR1 = format_partitioned_samples(
X1, p, part1, rep_indices1, F1, alpha, nx)
CR2, list_R2, list_p2, FR2 = format_partitioned_samples(
X2, q, part2, rep_indices2, F2, alpha, nx)
MR = dist(FR1, FR2)
# call to partitioned quantized fused gromov-wasserstein solver
res = quantized_fused_gromov_wasserstein_partitioned(
CR1, CR2, list_R1, list_R2, list_p1, list_p2, MR, alpha, build_OT=True,
log=log, armijo=armijo, max_iter=max_iter, tol_rel=tol_rel,
tol_abs=tol_abs, nx=nx, **kwargs)
if log:
T_global, Ts_local, T, log_ = res
C1 = dist(X1, X1)
C2 = dist(X2, X2)
# compute the transport cost on structures
constC, hC1, hC2 = init_matrix(C1, C2, p, q, 'square_loss', nx)
structure_cost = gwloss(constC, hC1, hC2, T, nx)
if alpha != 1.:
M = dist(F1, F2)
feature_cost = nx.sum(M * T)
else:
feature_cost = 0.
log_['qFGW_dist'] = alpha * structure_cost + (1 - alpha) * feature_cost
return T_global, Ts_local, T, log_
else:
T_global, Ts_local, T = res
return T_global, Ts_local, T