google-research

1
# coding=utf-8
2
# Copyright 2024 The Google Research Authors.
3
#
4
# Licensed under the Apache License, Version 2.0 (the "License");
5
# you may not use this file except in compliance with the License.
6
# You may obtain a copy of the License at
7
#
8
#     http://www.apache.org/licenses/LICENSE-2.0
9
#
10
# Unless required by applicable law or agreed to in writing, software
11
# distributed under the License is distributed on an "AS IS" BASIS,
12
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
# See the License for the specific language governing permissions and
14
# limitations under the License.
15

16
"""A non-scalable InstantEmbedding implementation."""
17
import numpy as np
18
import scipy.sparse as sps
19

20

21
def instant_embedding(pprs, dimensionality):
22
  """Produces non-private InstantEmbeddings for given set of PPRs.
23

24
  Args:
25
    pprs: Input PPR matrix. Can be non-square (m x n).
26
    dimensionality: Dimensionality of the embedding vector.
27

28
  Returns:
29
    InstantEmbedding matrix of size (m x `dimensionality`).
30
  """
31
  n_input_nodes = pprs.shape[0]
32
  n_nodes = pprs.shape[1]
33
  ppr_trans = np.log(np.maximum(1, n_nodes * pprs))
34
  proj = sps.coo_matrix((
35
      np.random.choice([-1, 1], n_input_nodes),
36
      (np.arange(n_input_nodes), np.random.randint(0, dimensionality, n_nodes)),
37
  )).tocsr()
38
  return ppr_trans @ proj
39