Private and federated learning
Techniques regarding privacy preservation in a federated learning environment are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory, and that can execute the computer executable components stored in the memory. The computer executable components can comprise a plurality of machine learning components that can execute a machine learning algorithm to generate a plurality of model parameters. The computer executable components can also comprise an aggregator component that can synthesize a machine learning model based on an aggregate of the plurality of model parameters. The aggregator component can communicate with the plurality of machine learning components via a data privacy scheme that comprises a privacy process and a homomorphic encryption process in a federated learning environment.
1. A system, comprising:
a processor; and
a memory coupled with the processor, wherein the memory is configured to provide the processor with instructions which when executed cause the processor to:
generate a plurality of queries regarding a machine learning algorithm, wherein a machine learning model is generated using the machine learning algorithm and is trained based on data held by one or more computer entities;
communicate the plurality of queries to the one or more computer entities, wherein the one or more computer entities implement a data privacy scheme that comprises a privacy process and a homomorphic encryption process in a federated learning environment;
receive an encrypted modified response to at least one of the plurality of queries, wherein the encrypted modified response has an amount of noise added to a generated response to the at least one of the plurality of queries; and
initiate a cryptographic process that processes the encrypted modified response.
2. The system of claim 1 , wherein the privacy process includes at least one member selected from a group consisting of: an anonymization process, a randomization process, a differential privacy process, a suppression process, and a generalization process.
3. The system of claim 1 , wherein the homomorphic encryption process is a threshold variant homomorphic encryption process.
4. The system of claim 1 , wherein at least one of the plurality of queries comprises a linear query requiring information from a dataset held or managed by at least one of the one or more computer entities.
5. The system of claim 1 , wherein the processor is further configured to:
aggregate other encrypted modified responses with the encrypted modified response to generate an encrypted response composition; and
query a plurality of the one or more computer entities to decrypt respective pieces of the encrypted response composition.
6. The system of claim 5 , wherein a threshold setting defines a number of the plurality of one or more computer entities queried.
7. The system of claim 1 , wherein the amount of noise depends on a privacy guarantee value provided by an entity associated with the system or a trust parameter associated with a number of non-colluding ones of the one or more computer entities in the federated learning environment.
8. A computer-implemented method, comprising:
generating, using a processor, a plurality of queries regarding a machine learning algorithm, wherein a machine learning model is generated using the machine learning algorithm and is trained based on data held by one or more computer entities;
communicating the plurality of queries to the one or more computer entities, wherein the one or more computer entities implement a data privacy scheme that comprises a privacy process and a homomorphic encryption process in a federated learning environment;
receiving an encrypted modified response to at least one of the plurality of queries, wherein the encrypted modified response has an amount of noise added to a generated response to the at least one of the plurality of queries; and
initiating a cryptographic process that processes the encrypted modified response.
9. The computer-implemented method of claim 8 , wherein the privacy process includes at least one member selected from a group consisting of: an anonymization process, a randomization process, a differential privacy process, a suppression process, and a generalization process.
10. The computer-implemented method of claim 8 , wherein the homomorphic encryption process is a threshold variant homomorphic encryption process.
11. The computer-implemented method of claim 8 , further comprising:
aggregating other encrypted modified responses with the encrypted modified response to generate an encrypted response composition; and
querying a plurality of the one or more computer entities to decrypt respective pieces of the encrypted response composition.
12. The computer-implemented method of claim 11 , wherein a threshold setting defines a number of the plurality of one or more computer entities queried.
13. A computer program product for performing private federated learning, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:
generate, by the processor, a plurality of queries regarding a machine learning algorithm, wherein a machine learning model is generated using the machine learning algorithm and is trained based on data held by one or more computer entities;
communicate the plurality of queries to the one or more computer entities, wherein the one or more computer entities implement a data privacy scheme that comprises a privacy process and a homomorphic encryption process in a federated learning environment;
receive an encrypted modified response to at least one of the plurality of queries, wherein the encrypted modified response has an amount of noise added to a generated response to the at least one of the plurality of queries; and
initiate a cryptographic process that processes the encrypted modified response.
14. The computer program product of claim 13 , wherein the privacy process includes at least one member selected from a group consisting of: an anonymization process, a randomization process, a differential privacy process, a suppression process, and a generalization process, and wherein the federated learning environment is facilitated by a cloud computing technology.