def _predict_scores_fixed(self, X, **kwargs):
"""Predict the scores for a given collection of sets of objects of same size.
Parameters
----------
X : array-like, shape (n_samples, n_objects, n_features)
Returns
-------
Y : array-like, shape (n_samples, n_objects)
Returns the scores of each of the objects for each of the samples.
"""
n_instances, n_objects, n_features = X.shape
assert n_features == self.n_object_features_fit_
outputs = [list() for _ in range(n_objects)]
for i, j in combinations(range(n_objects), 2):
x1 = X[:, i]
x2 = X[:, j]
x1x2 = np.concatenate((x1, x2), axis=1)
x2x1 = np.concatenate((x2, x1), axis=1)
n_g = np.dot(x1x2, self.weight1_) + self.bias1_
n_l = np.dot(x2x1, self.weight1_) + self.bias1_
outputs[i].append(n_g)
outputs[j].append(n_l)
outputs = np.array(outputs)
outputs = np.mean(outputs, axis=1).T
scores_zero = np.dot(X, self.weight2_) + self.bias2_
scores = sigmoid(self.W_last_[0] * scores_zero + self.W_last_[1] * outputs)
return scores
def _predict_scores_fixed(self, X, **kwargs):
n_instances, n_objects, n_features = X.shape
assert n_features == self.n_object_features_fit_
rep = np.mean(np.dot(X, self.weight1_), axis=1) + self.bias1_
rep = np.tile(rep[:, np.newaxis, :], (1, n_objects, 1))
X_n = np.concatenate((X, rep), axis=2)
scores = np.dot(X_n, self.weight2_) + self.bias2_
scores = sigmoid(scores)
return scores
def _predict_scores_fixed(self, X, **kwargs):
n_instances, n_objects, n_features = X.shape
assert n_features == self.n_object_features
outputs = [list() for _ in range(n_objects)]
for i, j in combinations(range(n_objects), 2):
x1 = X[:, i]
x2 = X[:, j]
x1x2 = np.concatenate((x1, x2), axis=1)
x2x1 = np.concatenate((x2, x1), axis=1)
n_g = np.dot(x1x2, self.weight1) + self.bias1
n_l = np.dot(x2x1, self.weight1) + self.bias1
outputs[i].append(n_g)
outputs[j].append(n_l)
outputs = np.array(outputs)
outputs = np.mean(outputs, axis=1).T
scores_zero = np.dot(X, self.weight2) + self.bias2
scores = sigmoid(self.W_last[0] * scores_zero +
self.W_last[1] * outputs)
return scores
def _predict_scores_using_pairs(self, X, **kwd):
n_instances, n_objects, n_features = X.shape
n2 = n_objects * (n_objects - 1)
pairs = np.empty((n2, 2, n_features))
scores = np.zeros((n_instances, n_objects))
for n in range(n_instances):
for k, (i, j) in enumerate(permutations(range(n_objects), 2)):
pairs[k] = (X[n, i], X[n, j])
result = self._predict_pair(pairs[:, 0],
pairs[:, 1],
only_pairwise=True,
**kwd)[:, 0]
scores[n] += result.reshape(n_objects, n_objects - 1).mean(axis=1)
del result
del pairs
if self._use_zeroth_model:
scores_zero = self.zero_order_model.predict(
X.reshape(-1, n_features))
scores_zero = scores_zero.reshape(n_instances, n_objects)
model = self._create_weighted_model(n_objects)
scores = model.predict([scores, scores_zero], **kwd)
else:
scores = sigmoid(scores)
return scores
def _predict_scores_using_pairs(self, X, **kwd):
scores = super()._predict_scores_using_pairs(X=X, **kwd)
scores = sigmoid(scores)
return scores