Examples described herein provide a computer-implemented method that includes calculating, by a processing device, a motor acceleration error based at least in part on a motor torque and a motor speed. The method further includes calculating, by the processing device, a regression fit line based at least in part on the motor acceleration error. The method further includes identifying, by the processing device, a zero point using the regression fit line. The method further includes comparing, by the processing device, the zero point to a datum reference to determine a difference. The method further includes integrating, by the processing device, the difference to determine the lash angle. The method further includes controlling, by the processing device, the motor based at least in part on the lash angle.

Examples described herein provide a computer-implemented method that includes calculating, by a processing device, a motor acceleration error based at least in part on a motor torque and a motor speed. The method further includes calculating, by the processing device, a regression fit line based at least in part on the motor acceleration error. The method further includes identifying, by the processing device, a zero point using the regression fit line. The method further includes comparing, by the processing device, the zero point to a datum reference to determine a difference. The method further includes integrating, by the processing device, the difference to determine the lash angle. The method further includes controlling, by the processing device, the motor based at least in part on the lash angle.

A computer system trains a federated learning model. A federated learning model is distributed to a plurality of computing nodes, each having a set of local training data comprising labeled data samples. Statistical data is received from each computing node that indicates the node's count of data samples for each label, and is analyzed to identify one or more computing nodes having local training data in which a label category is underrepresented beyond a threshold value with respect to data samples. Additional data samples labeled with the underrepresented labels are provided, and the computing nodes perform training. Results of training are received and are processed to generate a trained global model. Embodiments of the present invention further include a method and program product for training a federated learning model in substantially the same manner described above.

A computer system trains a federated learning model. A federated learning model is distributed to a plurality of computing nodes, each having a set of local training data comprising labeled data samples. Statistical data is received from each computing node that indicates the node's count of data samples for each label, and is analyzed to identify one or more computing nodes having local training data in which a label category is underrepresented beyond a threshold value with respect to data samples. Additional data samples labeled with the underrepresented labels are provided, and the computing nodes perform training. Results of training are received and are processed to generate a trained global model. Embodiments of the present invention further include a method and program product for training a federated learning model in substantially the same manner described above.

Disclosed in the embodiments of the present application are a non-line of sight (NLOS) elimination method and device for a time of arrival (TOA) measurement value, and a terminal. The method includes: modeling the probability density of the TOA measurement value of each base station arriving at a terminal into a Gaussian mixture model, and performing selection and NLOS identification on the TOA measurement value subsequent to performing Gaussian mixture modeling, so as to obtain an identification tag, wherein the identification tag is used for indicating whether the selected TOA measurement values correspond to NLOS; and correcting the selected TOA measurement value according to the identification tag, so as to eliminate an error caused by NLOS in the selected TOA measurement value. The present invention improves the positioning accuracy of a user by performing Gaussian mixture modeling and selection on the probability density of each TOA measurement value, accurately finding the TOA measurement value corresponding to LOS is ensured that in the case that the LOS is aliased with the NLOS, and correcting the selected TOA measurement value to eliminate the error caused by the NLOS in the selected TOA measurement value.

A method of performing anisotropic texture filtering includes generating one or more parameters describing an elliptical footprint in texture space; performing isotropic filtering at each sampling point of a set of sampling points in an ellipse to be sampled to produce a plurality of isotropic filter results, the ellipse to be sampled based on the elliptical footprint; selecting, based on one or more parameters of the set of sampling points and one or more parameters of the ellipse to be sampled, weights of an anisotropic filter that minimize a cost function that penalises high frequencies in the filter response of the anisotropic filter under a constraint that the variance of the anisotropic filter is related to an anisotropic ratio squared, the anisotropic ratio being the ratio of a major radius of the ellipse to be sampled and a minor axis of the ellipse to be sampled; and combining the plurality of isotropic filter results using the selected weights of the anisotropic filter to generate at least a portion of a filter result.

A method of performing anisotropic texture filtering includes generating one or more parameters describing an elliptical footprint in texture space; performing isotropic filtering at each sampling point of a set of sampling points in an ellipse to be sampled to produce a plurality of isotropic filter results, the ellipse to be sampled based on the elliptical footprint; selecting, based on one or more parameters of the set of sampling points and one or more parameters of the ellipse to be sampled, weights of an anisotropic filter that minimize a cost function that penalises high frequencies in the filter response of the anisotropic filter under a constraint that the variance of the anisotropic filter is related to an anisotropic ratio squared, the anisotropic ratio being the ratio of a major radius of the ellipse to be sampled and a minor axis of the ellipse to be sampled; and combining the plurality of isotropic filter results using the selected weights of the anisotropic filter to generate at least a portion of a filter result.

Systems and methods for automatically building a machine learning model are disclosed. A plurality of variables is displayed via a graphical user interface (GUI). A target variable and a first independent variable are identified from the plurality of variables. A parameter associated with the machine learning model is identified. Collected data is received via the GUI. A first machine learning model is built using as inputs, the parameter and the collected data associated with the first independent variable and the target variable. A change is made to at least a portion of the inputs used to build the first machine learning model. A second machine learning model is built based on the change. A prediction accuracy of the first machine learning model is compared to the prediction accuracy of the second machine learning model. Either the first or second machine learning model is selected based on the prediction accuracy.

Systems and methods for automatically building a machine learning model are disclosed. A plurality of variables is displayed via a graphical user interface (GUI). A target variable and a first independent variable are identified from the plurality of variables. A parameter associated with the machine learning model is identified. Collected data is received via the GUI. A first machine learning model is built using as inputs, the parameter and the collected data associated with the first independent variable and the target variable. A change is made to at least a portion of the inputs used to build the first machine learning model. A second machine learning model is built based on the change. A prediction accuracy of the first machine learning model is compared to the prediction accuracy of the second machine learning model. Either the first or second machine learning model is selected based on the prediction accuracy.

The present disclosure relates to systems and methods for pharmaceutical injection supply management, and in particular, recommending an optimal amount of pharmaceutical injection supply for a user to load into a dispenser. Implementations of the systems and methods discussed herein can provide, via a mobile application, a recommendation to a pharmaceutical user of the amount of substance to load into their substance pump during a site change. The recommendation can be optimized to minimize substance costs and substance waste, or to accommodate a target replacement date and time.