diff --git a/bob/learn/em/gmm.py b/bob/learn/em/gmm.py
index ce742f8225228258461e7b0b3fef5a285855fac1..c2ebe217fc5884a47e5be9c4b837b4f2ef12d411 100644
--- a/bob/learn/em/gmm.py
+++ b/bob/learn/em/gmm.py
@@ -212,7 +212,9 @@ class GMMStats:
return (self.n_gaussians, self.n_features)
def compute(self):
- for name in ("log_likelihood", "n", "sum_px", "sum_pxx"):
+ for name in ("log_likelihood", "t"):
+ setattr(self, name, float(getattr(self, name)))
+ for name in ("n", "sum_px", "sum_pxx"):
setattr(self, name, np.asarray(getattr(self, name)))
@@ -272,9 +274,8 @@ class GMMMachine(BaseEstimator):
update_variances: bool = False,
update_weights: bool = False,
mean_var_update_threshold: float = EPSILON,
- alpha: float = 0.5,
- relevance_factor: Union[None, float] = 4,
- variance_thresholds: float = EPSILON,
+ map_alpha: float = 0.5,
+ map_relevance_factor: Union[None, float] = 4,
):
"""
Parameters
@@ -304,17 +305,14 @@ class GMMMachine(BaseEstimator):
Update the Gaussians variances at every m step.
update_weights
Update the GMM weights at every m step.
- mean_var_update_threshold:
+ mean_var_update_threshold
Threshold value used when updating the means and variances.
- alpha:
+ map_alpha
Ratio for MAP adaptation. Used when `trainer == "map"` and
`relevance_factor is None`)
- relevance_factor:
+ map_relevance_factor
Factor for the computation of alpha with Reynolds adaptation. (Used when
`trainer == "map"`)
- variance_thresholds:
- The variance flooring thresholds, i.e. the minimum allowed value of variance in each dimension.
- The variance will be set to this value if an attempt is made to set it to a smaller value.
"""
self.n_gaussians = n_gaussians
@@ -361,8 +359,8 @@ class GMMMachine(BaseEstimator):
)
if weights is not None:
self.weights = weights
- self.alpha = alpha
- self.relevance_factor = relevance_factor
+ self.map_alpha = map_alpha
+ self.map_relevance_factor = map_relevance_factor
@property
def weights(self):
@@ -729,9 +727,9 @@ class GMMMachine(BaseEstimator):
update_variances=self.update_variances,
update_weights=self.update_weights,
mean_var_update_threshold=self.mean_var_update_threshold,
- reynolds_adaptation=self.relevance_factor is not None,
- alpha=self.alpha,
- relevance_factor=self.relevance_factor,
+ reynolds_adaptation=self.map_relevance_factor is not None,
+ alpha=self.map_alpha,
+ relevance_factor=self.map_relevance_factor,
**kwargs,
)
@@ -839,14 +837,14 @@ def ml_gmm_m_step(
"""Updates a gmm machine parameter according to the e-step statistics."""
logger.debug("ML GMM Trainer m-step")
+ # Threshold the low n to prevent divide by zero
+ thresholded_n = np.clip(statistics.n, mean_var_update_threshold, None)
+
# Update weights if requested
# (Equation 9.26 of Bishop, "Pattern recognition and machine learning", 2006)
if update_weights:
logger.debug("Update weights.")
- machine.weights = statistics.n / statistics.t
-
- # Threshold the low n to prevent divide by zero
- thresholded_n = np.clip(statistics.n, mean_var_update_threshold, None)
+ machine.weights = thresholded_n / statistics.t
# Update GMM parameters using the sufficient statistics (m_ss):
diff --git a/bob/learn/em/k_means.py b/bob/learn/em/k_means.py
index 830b4e4da6cb613a7b8b846281e6558b72402100..3b73a610fd1beee0fbbd2f1a6844818d4965caa7 100644
--- a/bob/learn/em/k_means.py
+++ b/bob/learn/em/k_means.py
@@ -42,7 +42,8 @@ class KMeansMachine(BaseEstimator):
convergence_threshold: float = 1e-5,
max_iter: int = 20,
random_state: Union[int, np.random.RandomState] = 0,
- init_max_iter: Union[int, None] = None,
+ init_max_iter: Union[int, None] = 5,
+ oversampling_factor: float = 2,
) -> None:
"""
Parameters
@@ -68,6 +69,7 @@ class KMeansMachine(BaseEstimator):
self.max_iter = max_iter
self.random_state = random_state
self.init_max_iter = init_max_iter
+ self.oversampling_factor = oversampling_factor
self.average_min_distance = np.inf
self.zeroeth_order_statistics = None
self.first_order_statistics = None
@@ -193,6 +195,7 @@ class KMeansMachine(BaseEstimator):
init=self.init_method,
random_state=self.random_state,
max_iter=self.init_max_iter,
+ oversampling_factor=self.oversampling_factor,
)
def e_step(self, data: np.ndarray):
@@ -226,7 +229,7 @@ class KMeansMachine(BaseEstimator):
step += 1
logger.info(
f"Iteration {step:3d}"
- + (f"/{self.max_iter}" if self.max_iter else "")
+ + (f"/{self.max_iter:3d}" if self.max_iter else "")
)
distance_previous = distance
self.e_step(data=X)
@@ -241,7 +244,7 @@ class KMeansMachine(BaseEstimator):
distance = float(self.average_min_distance)
- logger.info(
+ logger.debug(
f"Average minimal squared Euclidean distance = {distance}"
)
@@ -249,7 +252,7 @@ class KMeansMachine(BaseEstimator):
convergence_value = abs(
(distance_previous - distance) / distance_previous
)
- logger.info(f"Convergence value = {convergence_value}")
+ logger.debug(f"Convergence value = {convergence_value}")
# Terminates if converged (and threshold is set)
if (
diff --git a/bob/learn/em/test/test_gmm.py b/bob/learn/em/test/test_gmm.py
index 51a8bed00cc29adb3278be5f9dbdb2ff26fdd8a6..7eda1c63df3a7827ce7631eb6a0c78ea0971f5f7 100644
--- a/bob/learn/em/test/test_gmm.py
+++ b/bob/learn/em/test/test_gmm.py
@@ -202,6 +202,7 @@ def test_GMMMachine():
with tempfile.NamedTemporaryFile(suffix=".hdf5") as f:
filename = f.name
gmm.save(HDF5File(filename, "w"))
+ # Using from_hdf5
gmm1 = GMMMachine.from_hdf5(HDF5File(filename, "r"))
assert type(gmm1.n_gaussians) is np.int64
assert type(gmm1.update_means) is np.bool_
@@ -210,6 +211,17 @@ def test_GMMMachine():
assert type(gmm1.trainer) is str
assert gmm1.ubm is None
assert gmm == gmm1
+ # Using load
+ gmm1 = GMMMachine(n_gaussians=gmm.n_gaussians)
+ gmm1.load(HDF5File(filename, "r"))
+ assert type(gmm1.n_gaussians) is np.int64
+ assert type(gmm1.update_means) is np.bool_
+ assert type(gmm1.update_variances) is np.bool_
+ assert type(gmm1.update_weights) is np.bool_
+ assert type(gmm1.trainer) is str
+ assert gmm1.ubm is None
+ assert gmm == gmm1
+
with tempfile.NamedTemporaryFile(suffix=".hdf5") as f:
filename = f.name
gmm.save(filename)
@@ -923,7 +935,7 @@ def test_gmm_MAP_3():
update_variances=False,
update_weights=False,
mean_var_update_threshold=accuracy,
- relevance_factor=None,
+ map_relevance_factor=None,
)
gmm.variance_thresholds = threshold
@@ -1071,7 +1083,7 @@ def test_gmm_MAP_dask():
update_variances=False,
update_weights=False,
mean_var_update_threshold=accuracy,
- relevance_factor=None,
+ map_relevance_factor=None,
)
gmm.variance_thresholds = threshold