Machine learning based on homomorphic encryption

    公开(公告)号:US11551035B2

    公开(公告)日:2023-01-10

    申请号:US16055705

    申请日:2018-08-06

    Applicant: ABB Schweiz AG

    Abstract: A method for evaluating data is based on a computational model, the computational model comprising model data, a training function and a prediction function. The method includes training the computational model by: receiving training data and training result data for training the computational model, and computing the model data from the training data and the training result data with the training function. The method includes predicting result data by: receiving field data for predicting result data; and computing the result data from the field data and the model data with the prediction function. The training data may be plaintext and the training result data may be encrypted with a homomorphic encryption algorithm, wherein the model data may be computed in encrypted form from the training data and the encrypted training result data with the training function. The field data may be plaintext, wherein the result data may be computed in encrypted form from the field data and the encrypted model data with the prediction function.

    Secure remote aggregation
    14.
    发明授权

    公开(公告)号:US10680799B2

    公开(公告)日:2020-06-09

    申请号:US16189126

    申请日:2018-11-13

    Applicant: ABB Schweiz AG

    Abstract: The application relates to a method for aggregation of a performance indicator of a device including: concatenating a respective first data item to a plurality of second data items in the device; encrypting the plurality of concatenated second data items relevant for computing the performance indicator using a first encryption key in the device, wherein the first encryption key is based on an additive homomorphic encryption scheme; sending the encrypted concatenated second data items to a computation cluster; computing the performance indicator on the computation cluster using the encrypted concatenated second data items and computing an aggregate value regarding the performance indicator by summing up the encrypted concatenated second data items; sending the aggregate value to a server of a service provider of the device; decrypting the aggregate value using a second encryption key on the server of the service provider; and verifying the decrypted result by checking whether the decrypted sum computed by summing up the encrypted concatenated second data items comprises a predetermined value. The present application also relates to a corresponding system and corresponding computer program product including one or more computer readable media having computer executable instructions for performing the steps of the method.

    METHODS FOR CONTROLLING A DEVICE AND CONTROL SYSTEM

    公开(公告)号:US20190116189A1

    公开(公告)日:2019-04-18

    申请号:US16161207

    申请日:2018-10-16

    Applicant: ABB Schweiz AG

    Abstract: A method for controlling a device includes: sending a command signed by an operator's signature to a server; verifying, in the server, that the operator is authenticated to transmit the command; assigning, in the server, a criticality level and an authorization level to the command; depending on the criticality level and the authorization level, sending an approval request relating to the command to at least one control user; approving or denying the approval request by at least a subset of the at least one control user; sending the denied or approved approval request back to the server; determining, in the server, whether the command was approved by sufficiently many control users based on the criticality level and the authorization level; and sending the command to the device for being carried out by the device in case the command was approved by sufficiently many control users, wherein at last one of the at least one control user and the operator is remote from each other.

    MACHINE LEARNING BASED ON HOMOMORPHIC ENCRYPTION

    公开(公告)号:US20180349740A1

    公开(公告)日:2018-12-06

    申请号:US16055705

    申请日:2018-08-06

    Applicant: ABB Schweiz AG

    CPC classification number: H04L9/008 H04L2209/46 H04L2209/76

    Abstract: A method for evaluating data is based on a computational model, the computational model comprising model data, a training function and a prediction function. The method includes training the computational model by: receiving training data and training result data for training the computational model, and computing the model data from the training data and the training result data with the training function. The method includes predicting result data by: receiving field data for predicting result data; and computing the result data from the field data and the model data with the prediction function. The training data may be plaintext and the training result data may be encrypted with a homomorphic encryption algorithm, wherein the model data may be computed in encrypted form from the training data and the encrypted training result data with the training function. The field data may be plaintext, wherein the result data may be computed in encrypted form from the field data and the encrypted model data with the prediction function.

    Methods for controlling a device and control system

    公开(公告)号:US11159535B2

    公开(公告)日:2021-10-26

    申请号:US16161207

    申请日:2018-10-16

    Applicant: ABB Schweiz AG

    Abstract: A method for controlling a device includes: sending a command signed by an operator's signature to a server; verifying, in the server, that the operator is authenticated to transmit the command; assigning, in the server, a criticality level and an authorization level to the command; depending on the criticality level and the authorization level, sending an approval request relating to the command to at least one control user; approving or denying the approval request by at least a subset of the at least one control user; sending the denied or approved approval request back to the server; determining, in the server, whether the command was approved by sufficiently many control users based on the criticality level and the authorization level; and sending the command to the device for being carried out by the device in case the command was approved by sufficiently many control users, wherein at last one of the at least one control user and the operator is remote from each other.

    Encryption scheme using multiple parties

    公开(公告)号:US11018857B2

    公开(公告)日:2021-05-25

    申请号:US15872201

    申请日:2018-01-16

    Applicant: ABB Schweiz AG

    Abstract: The present invention discloses a method for computing a secret value including a first secret using a function including an operation, comprising: computing, by a host, a first encrypted value of the first secret with a first key; sending, by the host, the first encrypted value to a value holder and the first key to a key holder, wherein the value holder and the key holder are independently trusted by the host; computing, by the value holder, a computed encrypted value from the first encrypted value using the function; and computing, by the key holder, a computed key from the first key using the function.

    Apparatus and method to monitor robot mechanical condition

    公开(公告)号:US10967514B2

    公开(公告)日:2021-04-06

    申请号:US16108992

    申请日:2018-08-22

    Applicant: ABB Schweiz AG

    Abstract: Mechanical condition monitoring of robots can be used to detect unexpected failure of robots. Data taken from a robot operation is processed and compared against a health baseline. Features extracted during the monitoring stage of robot operation are aligned with features extracted during the training stage in which the health baseline is established by projecting both onto a common subspace. A classifier which can include a distance assessment such as an L2-norm is used within the common subspace to assess the condition of the robot. Excursions of the distance assessment from a criteria indicate a failure or potential failure.

    Verifiable Analytics Platform with Smart Contracts

    公开(公告)号:US20200228324A1

    公开(公告)日:2020-07-16

    申请号:US16743521

    申请日:2020-01-15

    Applicant: ABB Schweiz AG

    Abstract: A method for offering data with objects to be performed on the basis of the data, providing solutions to the problems and verifying the provided solutions. The method including the steps of providing a digital platform; uploading the data with the object as challenge to the platform; uploading a smart contract relating to the challenge to the platform; transferring the challenge from the platform to at least one solution provider; uploading a signed hash of the solution provider's solution of the challenge to the platform; uploading the solution provider's solution of the challenge to the platform; and evaluating at the platform the uploaded solution based on predefined conditions.

Patent Agency Ranking