A prediction model conversion method includes: converting a prediction model by converting at least one parameter which is included in the prediction model and is for performing homogenization processing into at least one parameter for performing processing including nonlinear processing, the prediction model being a neural network; and generating an encrypted prediction model that performs prediction processing with input in a secret state remaining secret by encrypting the prediction model that has been converted.

Multiple systems may determine neural-network output data and neural-network parameter data and may transmit the data therebetween to train and run the neural-network model to predict an event given input data. A data-provider system may perform a dot-product operation using encrypted data, and a secure-processing component may decrypt and process that data using an activation function to predict an event. Multiple secure-processing components may be used to perform a multiplication operation using homomorphic encrypted data.

The embodiments set forth a technique for securely identifying relevant computing devices that are nearby. The technique can be implemented at a first computing device, and include the steps of (1) receiving, from a second computing device, an advertisement packet that includes: (i) a network address that is associated with the second computing device, and (ii) a hash value that is calculated using the network address and an encryption key that is associated with the second computing device, and (2) for each known encryption key in a plurality of known encryption keys that are accessible to the first computing device: (i) calculating a temporary hash value using the network address and the known encryption key, and (ii) in response to identifying that the temporary hash value and the hash value match: carrying out an operation associated with the second computing device.

Methods and systems for managing cryptographic keys in on-premises and cloud computing environments and performing multi-party cryptography are disclosed. A cryptographic key can be retrieved from a hardware security module by a key management computer. The key management computer can generate key shares from the cryptographic key, and securely distribute the key shares to computer nodes or key share databases. The computer nodes can use the key shares in order to perform secure multi-party cryptography.

Disclosed is a smart contract and an automatic bargaining method for secure multi-party computation and a device embodies same. A first public key and a first private key are generated based on a homomorphic encryption algorithm in the smart contract, the first public key is sent to a demander and a supplier so that the demander and the supplier encrypt the purchase price and the sale price through the first public key, respectively; after receiving a second public key and an encrypted purchase price sent by the demander and a third public key and an encrypted sale price sent by the supplier, the encrypted sale price and the encrypted purchase price that meet the preset transaction conditions are decrypted.

A secure multi-party computation implements real number arithmetic using modular integer representation on the backend. As part of the implementation, a secret shared value jointly stored by multiple parties in a first modular representation is cast into a second modular representation having a larger most significant bit. The parties use a secret shared masking value in the first representation, the range of which is divided into two halves, to mask and reveal a sum of the secret shared value and the secret shared masking value. The parties use a secret shared bit that identifies the half of the range that contains the masking value, along with the sum to collaboratively construct a set of secret shares representing the secret shared value in the second modular format. In contrast with previous work, the disclosed solution eliminates a non-zero probability of error without sacrificing efficiency or security.

A method for performing secure computations on records, comprising: receiving a request to apply a computation on a record; assigning a respective partial record of a plurality of partial records of the record to each of a plurality of computational processes; instructing each of the plurality of computational processes to perform a computation scheme comprising: applying a semi honest multiparty computation on the partial record; iteratively repeating a predetermined number of times: using a secure multiparty arithmetic computation to generate random terms; using the secure multiparty arithmetic computation to assign the random terms and an outcome of the application to at least one predetermined equation; verifying an integrity of the semi honest multiparty computation by comparison of the assignments to the at least one predetermined equation to at least one constant; and when the integrity is valid, combining the applications of the semi honest multiparty computations on the partial records.

A method includes dividing a plurality of filters in a first layer of a neural network into a first set of filters and a second set of filters, applying each of the first set of filters to an input of the neural network, aggregating, at a second layer of the neural network, a respective one of a first set of outputs with a respective one of a second set of outputs, splitting respective weights of specific neurons activated in each remaining layer, at each specific neuron from each remaining layer, applying a respective filter associated with each specific neuron and a first corresponding weight, obtaining a second set of neuron outputs, for each specific neuron, aggregating one of the first set of neuron outputs with one of a second set of neuron outputs and generating an output of the neural network based on the aggregated neuron outputs.

An advanced encryption and key exchange (AEKE) algorithm for quantum safe cryptography is disclosed. The AEKE algorithm does not use hard mathematical problems that are easily solvable on a quantum computer with Shor's algorithm. Instead, new encryption algorithm uses simple linear algebra, rank deficient matrix and bilinear equation, which will be easy to understand, fast, efficient and practical but virtually impossible to crack.

An apparatus for set intersection operation according to an embodiment includes a ciphertext acquisition unit configured to acquire a ciphertext for a first vector corresponding to a first subset of a universal set including a plurality of elements from an encryption apparatus, a transform unit configured to generate a second vector corresponding to a second subset of the universal set, a computation unit configured to generate a ciphertext for a third vector corresponding to an intersection of the first subset and the second subset, based on the ciphertext for the first vector and the second vector, and a ciphertext providing unit configured to provides the ciphertext for the third vector to the encryption apparatus.