17838901. MACHINE LEARNING NETWORK EXTENSION BASED ON HOMOMORPHIC ENCRYPTION PACKINGS simplified abstract (INTERNATIONAL BUSINESS MACHINES CORPORATION)
MACHINE LEARNING NETWORK EXTENSION BASED ON HOMOMORPHIC ENCRYPTION PACKINGS
Organization Name
INTERNATIONAL BUSINESS MACHINES CORPORATION
Inventor(s)
Nir Drucker of Zichron Yaakov (IL)
Moran Baruch of Petah Tikva (IL)
MACHINE LEARNING NETWORK EXTENSION BASED ON HOMOMORPHIC ENCRYPTION PACKINGS - A simplified explanation of the abstract
This abstract first appeared for US patent application 17838901 titled 'MACHINE LEARNING NETWORK EXTENSION BASED ON HOMOMORPHIC ENCRYPTION PACKINGS
Simplified Explanation
The abstract of the patent application describes a system that utilizes machine learning networks and homomorphic encryption (HE) packing frameworks. The system includes a processor that can generate a list of HE packings for the machine learning network based on the selected HE packing framework. The processor can then extend the machine learning network by adding additional neurons based on the generated list of HE packings. The extended machine learning network can also be trained by the processor.
- The system includes a processor that receives a machine learning network and a selected homomorphic encryption (HE) packing framework.
- The processor generates a list of HE packings for the machine learning network based on the selected HE packing framework.
- The machine learning network is extended by the processor by adding additional neurons based on the generated list of HE packings.
- The extended machine learning network can be trained by the processor.
Potential Applications
This technology has potential applications in various fields, including:
- Secure machine learning: The use of homomorphic encryption allows for the secure processing of sensitive data in machine learning applications.
- Privacy-preserving analytics: By encrypting the data and performing computations on encrypted data, privacy can be maintained while still gaining insights from the data.
- Secure data sharing: Homomorphic encryption enables secure sharing of data between different parties without revealing the actual data.
Problems Solved
This technology solves several problems in the field of machine learning and data security, including:
- Data privacy: Homomorphic encryption ensures that sensitive data remains encrypted throughout the processing, preventing unauthorized access to the data.
- Secure computation: The use of homomorphic encryption allows for secure computations on encrypted data, ensuring that the results are also encrypted and protected.
- Secure data sharing: The ability to perform computations on encrypted data enables secure sharing of data between different parties without exposing the actual data.
Benefits
The use of this technology offers several benefits, including:
- Enhanced data privacy: By utilizing homomorphic encryption, sensitive data can be protected throughout the entire machine learning process.
- Secure machine learning: The combination of machine learning networks and homomorphic encryption provides a secure environment for processing and analyzing sensitive data.
- Collaborative data analysis: Homomorphic encryption enables secure collaboration and sharing of data between different parties, fostering collaboration and innovation in data-driven fields.
Original Abstract Submitted
An example system includes a processor to receive a machine learning network and a selected homomorphic encryption (HE) packing framework. The processor can generate list of HE packings for the machine learning network based on the selected HE packing framework. The processor can extend the machine learning network to include additional neurons based on the list of HE packings. The processor can also train the extended machine learning network.