Abstract: Device configurations for a machine-learned model. A device analytics system includes an electronic processor configured to determine, for a machine-learned model configured to detect a plurality of features, a prioritization ranking for a plurality of input parameters provided to the machine-learned model and receive, for each device, a confidence value for each feature included in the plurality of features. The electronic processor is configured to determine, for each device included in the plurality of devices, a performance value for each feature included in the plurality of features based on the prioritization ranking for the plurality of input parameters and the confidence value for each feature for the respective device. The electronic processor is configured to select, based on the performance value for each device included in the plurality of devices for a feature, a device configuration for the machine-learned model, and implement the selected device configuration.

Abstract: Device configurations for a machine-learned model. A device analytics system includes an electronic processor configured to determine, for a machine-learned model configured to detect a plurality of features, a prioritization ranking for a plurality of input parameters provided to the machine-learned model and receive, for each device, a confidence value for each feature included in the plurality of features. The electronic processor is configured to determine, for each device included in the plurality of devices, a performance value for each feature included in the plurality of features based on the prioritization ranking for the plurality of input parameters and the confidence value for each feature for the respective device. The electronic processor is configured to select, based on the performance value for each device included in the plurality of devices for a feature, a device configuration for the machine-learned model, and implement the selected device configuration.

Abstract: The disclosure provides methods for the direct optical visualization of graphene and its nanoscale defects on transparent substrates.

Abstract: Minimizing the latency of on-device detection of malicious executable files, without sacrificing accuracy, by applying a machine learning model to an executable file in quantized steps. Allowing a threshold confidence level to be set to different values enables controlling the tradeoff between accuracy and latency in generating a confidence level indicative of whether the executable file includes malware.

Abstract: The disclosure provides methods for the direct optical visualization of graphene and its nanoscale defects on transparent substrates.

Abstract: Minimizing the latency of on-device detection of malicious executable files, without sacrificing accuracy, by applying a machine learning model to an executable file in quantized steps. Allowing a threshold confidence level to be set to different values enables controlling the tradeoff between accuracy and latency in generating a confidence level indicative of whether the executable file includes malware.