FAST IDENTIFICATION OF TRUSTWORTHY DEEP NEURAL NETWORKS

    公开(公告)号:US20200380123A1

    公开(公告)日:2020-12-03

    申请号:US16887623

    申请日:2020-05-29

    Abstract: A system and method including receiving a set of deep neural networks (DNN) including DNNs trained with an embedded trojan and DNNs trained without any embedded trojan, each of the trained DNNs being represented by a mathematical formulation learned by the DNNs and expressing a relationship between an input of the DNNs and an output of the DNNs; extracting at least one characteristic feature from the mathematical formulation of each of the trained DNNs; statistically analyzing the at least one characteristic feature to determine whether there is a difference between the DNNs trained with the embedded trojan and the DNNs trained without any embedded trojan; generating, in response to the determination indicating there is a difference, a detector model to execute the statistical analyzing on deep neural networks; and storing a file including the generated detector model in a memory device.

    ADDITIVE MANUFACTURING MACHINE CALIBRATION BASED ON A TEST-PAGE BASED OBJECT

    公开(公告)号:US20200081414A1

    公开(公告)日:2020-03-12

    申请号:US16127545

    申请日:2018-09-11

    Abstract: A method of calibrating an additive manufacturing machine includes obtaining a model for the additive manufacturing machine, obtaining a baseline sensor data set for a particular additive manufacturing machine, creating a machine-specific nominal fingerprint for the particular additive manufacturing machine with controllable variation for one or more process inputs, producing on the particular additive manufacturing machine a test-page based object, obtaining a current sensor data set of the test-page based object on the particular additive manufacturing machine, estimating a scaling factor or a bias for each of the one or more process inputs from the current data set, and updating a calibration file for the particular additive machine if the estimated scaling error or bias are greater than a respective predetermined tolerance. A system for implementing the method and a non-transitory computer-readable medium are also disclosed.

    METHODS AND SYSTEMS FOR GENERATING DEVICE-SPECIFIC MACHINE LEARNING MODEL

    公开(公告)号:US20200073850A1

    公开(公告)日:2020-03-05

    申请号:US16122184

    申请日:2018-09-05

    Abstract: A method of transferring operational parameter sets between different domains of additive manufacturing machines includes creating a parameter set for a first additive manufacturing machine domain, accessing a model of a second additive manufacturing machine domain, creating a second parameter set of operational settings used to operate the second additive manufacturing machine, obtaining a second sensor data suite during the operation of the second additive manufacturing machine, comparing the second sensor data suite to one or more predetermined performance thresholds to determine if a product sample is within quality assurance metrics, and if the product sample is not within the quality assurance metrics, adjusting the second parameter set. A system for implementing the method and a non-transitory computer-readable medium are also disclosed.

Patent Agency Ranking