Protecting sensitive data from being exposed in graph embedding vectors
A method for protecting sensitive data from being exposed in graph embedding vectors. In some embodiments, a method may include generating first graph embedding vectors from an original graph and generating a proxy graph from the first graph embedding vectors. The proxy graph may include a plurality of proxy nodes and proxy edges connecting the proxy nodes. The proxy nodes may include one or more attributes of the original nodes that are included in the first graph embedding vectors. Second graph embedding vectors may then be generated by encoding the proxy graph and a reconstructed graph may be generated from the second graph embedding vectors. Finally, the reconstructed graph may be compared to the original graph and if a threshold level of similarity is met, a security action may be performed to protect sensitive data from being exposed.
1. A computer-implemented method for protecting sensitive data from being exposed in graph embedding vectors, at least a portion of the method being performed by a computing device comprising one or more processors, the computer-implemented method comprising:
identifying sensitive data;
creating an original graph from the sensitive data, the original graph including a plurality of original nodes and original edges connecting the original nodes;
generating first graph embedding vectors from the original graph, the first graph embedding vectors including one or more attributes of the original nodes;
generating from the first graph embedding vectors a proxy graph, the proxy graph including a plurality of proxy nodes and proxy edges connecting the proxy nodes, the proxy nodes including the one or more attributes of the original nodes included in the first graph embedding vectors;
generating second graph embedding vectors by encoding the proxy graph;
generating a reconstructed graph by decoding the second graph embedding vectors;
comparing the reconstructed graph to the original graph to determine whether a threshold level of similarity is met; and
in response to determining that the threshold level of similarity is met, performing a security action to protect the sensitive data from being exposed to a third-party through the first graph embedding vectors.
2. The computer-implemented method of claim 1 , wherein the first graph embedding vectors are generated by applying a graph embedding algorithm to the original graph.
3. The computer-implemented method of claim 1 , wherein the proxy graph is generated by using a mutual k-NN graph construction technique.
4. The computer-implemented method of claim 1 , wherein the generating of the second graph embedding vectors comprises using a graph autoencoder.
5. The computer-implemented method of claim 4 , wherein the graph autoencoder comprises an unsupervised machine learning algorithm.
6. The computer-implemented method of claim 1 , wherein the performing of the security action comprises injecting the first graph embedding vectors with random noise prior to sharing the first graph embedding vectors with the third-party.
7. One or more non-transitory computer-readable media comprising one or more computer-readable instructions that, when executed by one or more processors of a security server, cause the security server to perform a method for protecting sensitive data from being exposed in graph embedding vectors, the method comprising:
identifying sensitive data;
creating an original graph from the sensitive data, the original graph including a plurality of original nodes and original edges connecting the original nodes;
generating first graph embedding vectors from the original graph, the first graph embedding vectors including one or more attributes of the original nodes;
generating from the first graph embedding vectors a proxy graph, the proxy graph including a plurality of proxy nodes and proxy edges connecting the proxy nodes, the proxy nodes including the one or more attributes of the original nodes included in the first graph embedding vectors;
generating second graph embedding vectors by encoding the proxy graph;
generating a reconstructed graph by decoding the second graph embedding vectors;
comparing the reconstructed graph to the original graph to determine whether a threshold level of similarity is met; and
in response to determining that the threshold level of similarity is met, performing a security action to protect the sensitive data from being exposed to a third-party through the first graph embedding vectors.
8. The one or more non-transitory computer-readable media of claim 7 , wherein the first graph embedding vectors are generated by applying a graph embedding algorithm to the original graph.
9. The one or more non-transitory computer-readable media of claim 7 , wherein the proxy graph is generated by using a mutual k-NN graph construction technique.
10. The one or more non-transitory computer-readable media of claim 7 , wherein the generating of the second graph embedding vectors comprises using a graph autoencoder.
11. The one or more non-transitory computer-readable media of claim 10 , wherein the graph autoencoder comprises an unsupervised machine learning algorithm.
12. The one or more non-transitory computer-readable media of claim 7 , wherein the performing of the security action comprises injecting the first graph embedding vectors with random noise prior to sharing the first graph embedding vectors with the third-party.
13. A computing device comprising:
one or more processors; and
one or more non-transitory computer-readable media comprising one or more computer-readable instructions that, when executed by the one or more processors, cause the computing device to perform a method for protecting sensitive data from being exposed in graph embedding vectors, the method comprising:
identifying sensitive data;
creating an original graph from the sensitive data, the original graph including a plurality of original nodes and original edges connecting the original nodes;
generating first graph embedding vectors from the original graph, the first graph embedding vectors including one or more attributes of the original nodes;
generating from the first graph embedding vectors a proxy graph, the proxy graph including a plurality of proxy nodes and proxy edges connecting the proxy nodes, the proxy nodes including the one or more attributes of the original nodes included in the first graph embedding vectors;
generating second graph embedding vectors by encoding the proxy graph;
generating a reconstructed graph by decoding the second graph embedding vectors;
comparing the reconstructed graph to the original graph to determine whether a threshold level of similarity is met; and
in response to determining that the threshold level of similarity is met, performing a security action to protect the sensitive data from being exposed to a third-party through the first graph embedding vectors.
14. The computing device of claim 13 , wherein the proxy graph is generated by using a mutual k-NN graph construction technique.
15. The computing device of claim 13 , wherein the generating of the second graph embedding vectors comprises using a graph autoencoder.
16. The computing device of claim 15 , wherein the graph autoencoder comprises an unsupervised machine learning algorithm.
17. The computing device of claim 13 , wherein the performing of the security action comprises injecting the first graph embedding vectors with random noise.