1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258
|
"""
Sliced OT Distances
"""
# Author: Adrien Corenflos <adrien.corenflos@aalto.fi>
# Nicolas Courty <ncourty@irisa.fr>
# Rémi Flamary <remi.flamary@polytechnique.edu>
#
# License: MIT License
import numpy as np
from .backend import get_backend, NumpyBackend
from .utils import list_to_array
def get_random_projections(d, n_projections, seed=None, backend=None, type_as=None):
r"""
Generates n_projections samples from the uniform on the unit sphere of dimension :math:`d-1`: :math:`\mathcal{U}(\mathcal{S}^{d-1})`
Parameters
----------
d : int
dimension of the space
n_projections : int
number of samples requested
seed: int or RandomState, optional
Seed used for numpy random number generator
backend:
Backend to ue for random generation
Returns
-------
out: ndarray, shape (d, n_projections)
The uniform unit vectors on the sphere
Examples
--------
>>> n_projections = 100
>>> d = 5
>>> projs = get_random_projections(d, n_projections)
>>> np.allclose(np.sum(np.square(projs), 0), 1.) # doctest: +NORMALIZE_WHITESPACE
True
"""
if backend is None:
nx = NumpyBackend()
else:
nx = backend
if isinstance(seed, np.random.RandomState) and str(nx) == 'numpy':
projections = seed.randn(d, n_projections)
else:
if seed is not None:
nx.seed(seed)
projections = nx.randn(d, n_projections, type_as=type_as)
projections = projections / nx.sqrt(nx.sum(projections**2, 0, keepdims=True))
return projections
def sliced_wasserstein_distance(X_s, X_t, a=None, b=None, n_projections=50, p=2,
projections=None, seed=None, log=False):
r"""
Computes a Monte-Carlo approximation of the p-Sliced Wasserstein distance
.. math::
\mathcal{SWD}_p(\mu, \nu) = \underset{\theta \sim \mathcal{U}(\mathbb{S}^{d-1})}{\mathbb{E}}\left(\mathcal{W}_p^p(\theta_\# \mu, \theta_\# \nu)\right)^{\frac{1}{p}}
where :
- :math:`\theta_\# \mu` stands for the pushforwards of the projection :math:`X \in \mathbb{R}^d \mapsto \langle \theta, X \rangle`
Parameters
----------
X_s : ndarray, shape (n_samples_a, dim)
samples in the source domain
X_t : ndarray, shape (n_samples_b, dim)
samples in the target domain
a : ndarray, shape (n_samples_a,), optional
samples weights in the source domain
b : ndarray, shape (n_samples_b,), optional
samples weights in the target domain
n_projections : int, optional
Number of projections used for the Monte-Carlo approximation
p: float, optional =
Power p used for computing the sliced Wasserstein
projections: shape (dim, n_projections), optional
Projection matrix (n_projections and seed are not used in this case)
seed: int or RandomState or None, optional
Seed used for random number generator
log: bool, optional
if True, sliced_wasserstein_distance returns the projections used and their associated EMD.
Returns
-------
cost: float
Sliced Wasserstein Cost
log : dict, optional
log dictionary return only if log==True in parameters
Examples
--------
>>> n_samples_a = 20
>>> reg = 0.1
>>> X = np.random.normal(0., 1., (n_samples_a, 5))
>>> sliced_wasserstein_distance(X, X, seed=0) # doctest: +NORMALIZE_WHITESPACE
0.0
References
----------
.. [31] Bonneel, Nicolas, et al. "Sliced and radon wasserstein barycenters of measures." Journal of Mathematical Imaging and Vision 51.1 (2015): 22-45
"""
from .lp import wasserstein_1d
X_s, X_t = list_to_array(X_s, X_t)
if a is not None and b is not None and projections is None:
nx = get_backend(X_s, X_t, a, b)
elif a is not None and b is not None and projections is not None:
nx = get_backend(X_s, X_t, a, b, projections)
elif a is None and b is None and projections is not None:
nx = get_backend(X_s, X_t, projections)
else:
nx = get_backend(X_s, X_t)
n = X_s.shape[0]
m = X_t.shape[0]
if X_s.shape[1] != X_t.shape[1]:
raise ValueError(
"X_s and X_t must have the same number of dimensions {} and {} respectively given".format(X_s.shape[1],
X_t.shape[1]))
if a is None:
a = nx.full(n, 1 / n, type_as=X_s)
if b is None:
b = nx.full(m, 1 / m, type_as=X_s)
d = X_s.shape[1]
if projections is None:
projections = get_random_projections(d, n_projections, seed, backend=nx, type_as=X_s)
X_s_projections = nx.dot(X_s, projections)
X_t_projections = nx.dot(X_t, projections)
projected_emd = wasserstein_1d(X_s_projections, X_t_projections, a, b, p=p)
res = (nx.sum(projected_emd) / n_projections) ** (1.0 / p)
if log:
return res, {"projections": projections, "projected_emds": projected_emd}
return res
def max_sliced_wasserstein_distance(X_s, X_t, a=None, b=None, n_projections=50, p=2,
projections=None, seed=None, log=False):
r"""
Computes a Monte-Carlo approximation of the max p-Sliced Wasserstein distance
.. math::
\mathcal{Max-SWD}_p(\mu, \nu) = \underset{\theta _in
\mathcal{U}(\mathbb{S}^{d-1})}{\max} [\mathcal{W}_p^p(\theta_\#
\mu, \theta_\# \nu)]^{\frac{1}{p}}
where :
- :math:`\theta_\# \mu` stands for the pushforwars of the projection :math:`\mathbb{R}^d \ni X \mapsto \langle \theta, X \rangle`
Parameters
----------
X_s : ndarray, shape (n_samples_a, dim)
samples in the source domain
X_t : ndarray, shape (n_samples_b, dim)
samples in the target domain
a : ndarray, shape (n_samples_a,), optional
samples weights in the source domain
b : ndarray, shape (n_samples_b,), optional
samples weights in the target domain
n_projections : int, optional
Number of projections used for the Monte-Carlo approximation
p: float, optional =
Power p used for computing the sliced Wasserstein
projections: shape (dim, n_projections), optional
Projection matrix (n_projections and seed are not used in this case)
seed: int or RandomState or None, optional
Seed used for random number generator
log: bool, optional
if True, sliced_wasserstein_distance returns the projections used and their associated EMD.
Returns
-------
cost: float
Sliced Wasserstein Cost
log : dict, optional
log dictionary return only if log==True in parameters
Examples
--------
>>> n_samples_a = 20
>>> reg = 0.1
>>> X = np.random.normal(0., 1., (n_samples_a, 5))
>>> sliced_wasserstein_distance(X, X, seed=0) # doctest: +NORMALIZE_WHITESPACE
0.0
References
----------
.. [35] Deshpande, I., Hu, Y. T., Sun, R., Pyrros, A., Siddiqui, N., Koyejo, S., ... & Schwing, A. G. (2019). Max-sliced wasserstein distance and its use for gans. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 10648-10656).
"""
from .lp import wasserstein_1d
X_s, X_t = list_to_array(X_s, X_t)
if a is not None and b is not None and projections is None:
nx = get_backend(X_s, X_t, a, b)
elif a is not None and b is not None and projections is not None:
nx = get_backend(X_s, X_t, a, b, projections)
elif a is None and b is None and projections is not None:
nx = get_backend(X_s, X_t, projections)
else:
nx = get_backend(X_s, X_t)
n = X_s.shape[0]
m = X_t.shape[0]
if X_s.shape[1] != X_t.shape[1]:
raise ValueError(
"X_s and X_t must have the same number of dimensions {} and {} respectively given".format(X_s.shape[1],
X_t.shape[1]))
if a is None:
a = nx.full(n, 1 / n, type_as=X_s)
if b is None:
b = nx.full(m, 1 / m, type_as=X_s)
d = X_s.shape[1]
if projections is None:
projections = get_random_projections(d, n_projections, seed, backend=nx, type_as=X_s)
X_s_projections = nx.dot(X_s, projections)
X_t_projections = nx.dot(X_t, projections)
projected_emd = wasserstein_1d(X_s_projections, X_t_projections, a, b, p=p)
res = nx.max(projected_emd) ** (1.0 / p)
if log:
return res, {"projections": projections, "projected_emds": projected_emd}
return res
|