Abstract: Embodiments of this specification disclose secure multi-party computation for privacy protection. In an implementation, a method includes obtaining a fragment of first gradient data and a fragment of noise data, where the first gradient data is gradient data of a loss function. Based on the fragment of first gradient data by performing secure multi-party computation with another participant, obtaining a fragment of second gradient data, where the second gradient data is gradient data obtained after the first gradient data is clipped. Based on the fragment of second gradient data and the fragment of noise data, determining a fragment of third gradient data, where the third gradient data is the second gradient data with the noise data added. A fragment of a model parameter is determined based on the fragment of third gradient data.

Abstract: This specification provides example computer-implemented methods and apparatuses for jointly updating a service model based on privacy protection. In an example iteration process, a serving party provides, to each data party, global model parameters and a mapping relationship between the data party and N parameter groups obtained by dividing the global model parameters. Each data party updates a local service model by using the global model parameters, and further updates an updated local service model based on local service data, to upload model parameters in a new service model in a parameter group corresponding to the data party to the serving party. Then, the serving party successively fuses received parameter groups to update the global model parameters.

Abstract: Embodiments of this specification provide for generating dialogue annotation data. A method includes: obtaining grammar that has been pre-generated in a target scenario, the grammar including a first phrase and a second phrase; the first phrase includes a first intent operation and a first variable of a first intent, where the first variable invokes other phrases. The second phrase assigns a value to the first variable and includes a property acquisition operation and a second variable of a second intent. Generating a target meaning representation of nested phrases for simulating a certain human-machine dialogue by invoking the first phrase and the second phrase in the grammar. Performing the first intent operations based on a property value corresponding to the first variable of the second variable assigned to the first variable.

Abstract: Disclosed herein are methods, systems, and apparatus, including computer programs encoded on computer storage media, for payment based on wireless communication and biological features. One of the methods includes determining at least one wireless communications device is within a wireless communication range of a payment collection system. Data specifying at least one account associated with the at least one wireless communications device and data specifying a biometric feature associated with the at least one account is obtained from a payment platform. The payment collection system collects a target biometric feature of a target user. In response to determining that the target biometric feature matches the biometric feature associated with the at least one account, a target account of the target user is determined based on the at least one account, and a deduction operation is performed on the target account.

Abstract: Some embodiments of this specification disclose data encryption and data preprocessing. In an implementation, a method includes for online data encryption: generating, based on plaintext data, a second component of ciphertext data. From a data set and as a selected first component of ciphertext data, a first component of ciphertext data is selected. Based on the first component of ciphertext data and the second component of ciphertext data, ciphertext data of plaintext data is calculated. For offline data preprocessing, the first component of ciphertext data is generated and plaintext encryption uses the first component of ciphertext data and the second component of ciphertext data. Alternatively, a data acquisition request is received, the first component of ciphertext data is generated, and fed back, and stored for use in plaintext encryption.

Abstract: A consensus method for a blockchain based on a Byzantine fault tolerance algorithm, includes: initiating, by a first consensus node of a consortium blockchain, a consensus proposal comprising a first proposal digest generated based on a first last block hash maintained by the first consensus node; in response to the consensus proposal, generating, by a second consensus node of the consortium blockchain, a second proposal digest based on a second last block hash maintained by the second consensus node; and not participating in, by the second consensus node, a consensus on the consensus proposal if it is detected that the first proposal digest does not match the second proposal digest.

Abstract: In response to a graphics memory allocation request generated during the running of a target task and for graphics memory needed during running of the target task, target data generated during running of each sub-task of multiple sub-tasks is classified, where a type of the target data comprises at least first data, and where the first data is not used by a subsequent sub-task. Multiple target graphics memory pools are allocated to the multiple sub-tasks. Each target graphics memory pool of the multiple target graphics memory pools is divided into at least one graphics memory block based on a type of the target data, where the at least one graphics memory block includes at least a first graphics memory block corresponding to the first data, and where multiple first graphics memory blocks corresponding to the multiple sub-tasks are mapped to a same target physical memory address.

Abstract: The present specification provides a blockchain-based offline resource transfer method and apparatus. A transferor terminal and a receiver terminal of offline resource transfer include multiple execution environments. Multiple computing nodes constructed based on the execution environments further constitute an offline blockchain network as blockchain nodes.

Abstract: This disclosure relates to a method and a system for recognizing a disguised device. A client of an application is installed on the device for interacting with the application, and the method includes: in response to an interaction with an application installed on a device, receiving environment information and a device ID associated with the device, determining whether the device ID exists in a pre-stored association table of a unique environment identifier and a device ID, and in response to determining that the device ID does not exist in the association table, calculating a unique environment identifier associated with the device based on the environment information, searching the association table for a device ID associated with the unique environment identifier, and determining, based on a quantity of different identified device IDs associated with the unique environment identifier, whether the device is disguised.

Abstract: Some embodiments of this specification provide data processing methods, apparatuses, and devices. One method includes: receiving encrypted debugging information and an encrypted first key from a client device, determining the device key corresponding to the client device, obtaining the debugging information based on the device key, the encrypted debugging information, and the encrypted first key, and performing abnormality detection on the trusted execution environment of the client device based on the debugging information to determine an abnormality detection result for the trusted execution environment.

Abstract: A target sample including a sample feature, a first label, and a second label is obtained, where a user corresponds to the target sample, the first label indicates whether a target object is clicked on by the user, and the second label indicates whether the user implements a target behavior related to the target object. Model processing is performed on the sample feature using a prediction model with a first branch and first probability and a second branch and second probability. A first loss is determined based on a first label value and the first probability. When a predetermined condition is satisfied, the second loss is determined based on a second label value of the second label and the second probability, and a predicted loss of the target sample is determined based on the first loss and the second loss. Training the prediction model based on the predicted loss.

Abstract: Implementations of the present specification disclose methods, apparatuses, and devices for privacy-preserving multi-party joint data clustering. In one aspect, the method includes: performing, based on a first feature matrix associated with a first party and a first centroid partition of a current centroid matrix, a first multi-party secure calculation with a second party associated with a second feature matrix to obtain a first distance partition of a distance matrix; performing, based on the first distance partition, a secure comparison calculation with the second party with respect to a second distance partition of the distance matrix associated with the second party to obtain a first index partition of a cluster index matrix; and performing, based on the first feature matrix and the first index partition, a second multi-party secure calculation with the second party to obtain an updated first centroid partition in the current iteration.

Abstract: Data processing includes acquiring service data of a target service by using a trusted application in a trusted execution environment (TEE), generating a service session key corresponding to the target service, and transferring the service data and the service session key to a secure element and encrypted based on the service session key to obtain service ciphertext information, and encrypting the service session key based on a target service key to obtain a service session key ciphertext. A key identifier of the target service key, the service ciphertext information, and the service session key ciphertext are sent to a server device to determine the target service key, which is used to decrypt the service session key ciphertext to obtain the service session key, which is used to decrypt the service ciphertext information to obtain the service data of the target service to perform target service processing.

Abstract: Implementations of the present specification disclose methods, apparatuses, and devices for training an object detection system by using a gradient fine-tuning technique. In one aspect, the method includes: providing a training image as input to the object detection system; processing the training image by the object detection system; determining a gradient norm of each neural network layer based on object annotation data and the detection result corresponding to the training image; and updating, for each neural network layer, parameter values of the neural network layer based on an average of gradient norms and the gradient norm of the neural network layer.

Abstract: Methods, apparatuses, and systems for protecting privacy are described. In an example, a first party encrypts a first data set, and sends a single-encrypted full data set to a second party. The second party re-encrypts the single-encrypted full data set to obtain a double-encrypted full data set, obtains a single-encrypted sampling data set, and sends the single-encrypted sampling data set to the first party. The first party encrypts the single-encrypted sampling data set to obtain a double-encrypted sampling data set, determines an intersection of the double-encrypted full data set and the double-encrypted sampling data set, performs upsampling on a complement other than the intersection in the double-encrypted sampling data set to obtain a noise-added intersection, and sends an element index of the noise-added intersection to the second party. The second party determines related information of common data based on the element index.

Abstract: Computer-implemented methods, apparatuses, and computer-readable media for model training and text processing are disclosed. In an example, an original text, a positive sample, and a negative sample are obtained. The positive sample and the negative sample are rewritten texts of the original text. A first, second, and third distribution parameter corresponding to the original text, the positive sample, and the negative sample, respectively, are determined by using an encoder. The first, second, and third distribution parameter are used to describe a first, second, and third semantic distribution, respectively. A model parameter of the encoder is optimized by using a loss function based on the first, second, and third distribution parameter. The loss function includes a first term that is used to constrain a degree of similarity between the first and second semantic distribution to be higher than a degree of similarity between the first and the third semantic distribution.

Abstract: Implementations of the present specification disclose methods, apparatuses, and devices for performing security knowledge level tests. In one aspect, the method includes: receiving a security knowledge test request; determining a target security regulatory authority to perform a security knowledge level test for the target user; obtaining a target security knowledge test requirement information set that corresponds to the target security regulatory authority, wherein the target security knowledge test requirement information set comprises a mapping relationship between user features and security knowledge test questions; determining corresponding test questions based on the one or more target features of the target user; providing the corresponding test questions for presentation on a client device to the target user; and determining a security knowledge level test result from inputs to the client device by the target user in response to viewing the corresponding test questions.

Abstract: Implementations of the present specification provide a vehicle identification method and apparatus, a primary device, and a secondary device. In the vehicle identification method, after sending a low-frequency signal, the primary device receives a response signal sent by the secondary device for the low-frequency signal, and communicates with the secondary device to obtain identity information of a vehicle on which the secondary device is located, so that the vehicle can be identified. The response signal is sent from the secondary device after the secondary device receives the low-frequency signal and detects that the signal strength of the low-frequency signal is greater than or equal to a predetermined threshold, and the signal strength of the low-frequency signal rapidly attenuates as the distance increases, so that the primary device can determine the location of the vehicle on which the secondary device is located solely based on the received response signal.

Abstract: Embodiments of this specification provide a transaction processing method, apparatus, and electronic device for a blockchain. The method includes: executing, by a blockchain node in the blockchain, a consensus protocol for an N-th round of consensus, wherein the consensus protocol comprises reaching a consensus on a blockchain transaction of the N-th round of consensus and upon the consensus being reached, executing the blockchain transaction of the N-th round of consensus; writing, by the blockchain node after the execution of the blockchain transaction, a block comprising the blockchain transaction of the N-th round of consensus into the blockchain; and concurrently with the writing of the block, executing, by the blockchain node, the consensus protocol for an (N+1)-th round of consensus in the blockchain.

Abstract: Intent recognition is described. Obtained data to be recognized is preprocessed to obtain target data. Feature extraction processing is performed on the target data based on an intent recognition model to obtain a linear feature and a non-linear feature of the target data, where the intent recognition model is obtained by optimizing and training a bidirectional encoder representations from transformers (BERT) model. Intent recognition processing is performed based on the linear feature, the non-linear feature, and the intent recognition model to obtain an intent recognition result of the obtained data to be recognized.