Abstract: Methods of machine learning for system deployments without performance regressions are performed by systems and devices. A performance safeguard system is used to design pre-production experiments for determining the production readiness of learned models based on a pre-production budget by leveraging big data processing infrastructure and deploying a large set of learned or optimized models for its query optimizer. A pipeline for learning and training differentiates the impact of query plans with and without the learned or optimized models, selects plan differences that are likely to lead to most dramatic performance difference, runs a constrained set of pre-production experiments to empirically observe the runtime performance, and finally picks the models that are expected to lead to consistently improved performance for deployment. The performance safeguard system enables safe deployment not just for learned or optimized models but also for additional of other ML-for-Systems features.

Abstract: Methods of machine learning for system deployments without performance regressions are performed by systems and devices. A performance safeguard system is used to design pre-production experiments for determining the production readiness of learned models based on a pre-production budget by leveraging big data processing infrastructure and deploying a large set of learned or optimized models for its query optimizer. A pipeline for learning and training differentiates the impact of query plans with and without the learned or optimized models, selects plan differences that are likely to lead to most dramatic performance difference, runs a constrained set of pre-production experiments to empirically observe the runtime performance, and finally picks the models that are expected to lead to consistently improved performance for deployment. The performance safeguard system enables safe deployment not just for learned or optimized models but also for additional of other ML-for-Systems features.

Abstract: Methods and systems of identifying a time reduction in a manufacturing time associated with a plurality of products. One system includes an electronic processor configured to receive training data. The electronic processor is also configured to determine a first set of testing parameters from the plurality of testing parameters to remove for the assembly line based on the training data and determine a second set of testing parameters to keep by removing the first set of testing parameters from the plurality of testing parameters. The electronic processor is also configured to determine a predictive model to replace the first set of testing parameters based on the training data associated with the second set of testing parameters, and automatically update a testing process for the assembly line to turn off the first set of testing parameters and use the predictive model in place of the first set of testing parameters.

Abstract: Methods of machine learning for system deployments without performance regressions are performed by systems and devices. A performance safeguard system is used to design pre-production experiments for determining the production readiness of learned models based on a pre-production budget by leveraging big data processing infrastructure and deploying a large set of learned or optimized models for its query optimizer. A pipeline for learning and training differentiates the impact of query plans with and without the learned or optimized models, selects plan differences that are likely to lead to most dramatic performance difference, runs a constrained set of pre-production experiments to empirically observe the runtime performance, and finally picks the models that are expected to lead to consistently improved performance for deployment. The performance safeguard system enables safe deployment not just for learned or optimized models but also for additional of other ML-for-Systems features.

Abstract: Methods and systems of identifying a time reduction in a manufacturing time associated with a plurality of products. One system includes an electronic processor configured to receive a dataset associated with an assembly line. The dataset includes a classification for each of the plurality of products produced by the assembly line, where the assembly line is associated with a plurality of tests. The electronic processor is also configured to determine a set of test combinations for the assembly line based on the plurality of tests and, for each test combination, determine a number of missing products based on the classification for each of the plurality of products. The electronic processor is also configured to determine at least one test to remove based on the number of missing products for each test combination and output a result including an indication of the at least one test to remove.

Abstract: Methods and systems of identifying a time reduction in a manufacturing time associated with a plurality of products. One system includes an electronic processor configured to receive training data. The electronic processor is also configured to determine a first set of testing parameters from the plurality of testing parameters to remove for the assembly line based on the training data and determine a second set of testing parameters to keep by removing the first set of testing parameters from the plurality of testing parameters. The electronic processor is also configured to determine a predictive model to replace the first set of testing parameters based on the training data associated with the second set of testing parameters, and automatically update a testing process for the assembly line to turn off the first set of testing parameters and use the predictive model in place of the first set of testing parameters.

Abstract: Methods and systems of identifying a time reduction in a manufacturing time associated with a plurality of products. One system includes an electronic processor configured to receive a dataset associated with an assembly line. The dataset includes a classification for each of the plurality of products produced by the assembly line, where the assembly line is associated with a plurality of tests. The electronic processor is also configured to determine a set of test combinations for the assembly line based on the plurality of tests and, for each test combination, determine a number of missing products based on the classification for each of the plurality of products. The electronic processor is also configured to determine at least one test to remove based on the number of missing products for each test combination and output a result including an indication of the at least one test to remove.

Abstract: Methods, systems, and apparatuses for performing target parameter analysis for an assembly line including a plurality of stations. One method includes receiving, with an electronic processor, training data associated with the assembly line. The training data including a plurality of attributes. The method also includes receiving, with the electronic processor, value addition data for each of the plurality of stations. The value addition data for each of the plurality of stations specifying a non-negative value added by each of the plurality of stations. The method also includes learning, with the electronic processor, a decision tree based on the training data and the value addition data. The method also includes performing the target parameter analysis based on the decision tree.

Abstract: A system to predict calibration values for a vehicle. The system is configured to receive a plurality of training data sets for a component of the vehicle. Each of the plurality of training data sets includes one or more training inputs and one or more corresponding training outputs. The system is further configured to automatically develop a prediction model based on the plurality of training data sets. The system is further configured to receive an input data set and determine, using the prediction model, a predicted calibration value based on the input data set. The system is further configured to transmit the predicted calibration value to an electronic control unit of the vehicle.

Abstract: A system to predict calibration values for a vehicle. The system is configured to receive a plurality of training data sets for a component of the vehicle. Each of the plurality of training data sets includes one or more training inputs and one or more corresponding training outputs. The system is further configured to automatically develop a prediction model based on the plurality of training data sets. The system is further configured to receive an input data set and determine, using the prediction model, a predicted calibration value based on the input data set. The system is further configured to transmit the predicted calibration value to an electronic control unit of the vehicle.

Abstract: Methods, systems, and apparatuses for performing target parameter analysis for an assembly line including a plurality of stations. One method includes receiving, with an electronic processor, training data associated with the assembly line. The training data including a plurality of attributes. The method also includes receiving, with the electronic processor, value addition data for each of the plurality of stations. The value addition data for each of the plurality of stations specifying a non-negative value added by each of the plurality of stations. The method also includes learning, with the electronic processor, a decision tree based on the training data and the value addition data. The method also includes performing the target parameter analysis based on the decision tree.

Abstract: A physical activity monitoring method and system in one embodiment includes a wearable sensor device configured to generate physiologic data associated with a sensed physiologic condition of a wearer, and to generate audio context data associated with a sensed audio context of the wearer, a memory including program instructions stored therein, a computer configured to receive the physiologic data and the audio context data and to execute the program instructions to distinguish between dynamic and static activities of the wearer based upon the physiologic data, and to generate activity data by differentiating between different classes of distinguished dynamic and static activities based upon the audio context data, and a user interface operably connected to the computer for rendering the activity data.

Abstract: A system for clustering a plurality of documents having input and output space data is disclosed that uses both input and output space criteria. The system can aggregate documents into clusters based on input and/or output space similarity measures, and then refine the clusters based on further input and/or output space similarity measures. Aggregation of documents into clusters can include forming a hierarchical tree based on the input and/or output space similarity measures where the hierarchical tree has a root node, branching into intermediate nodes, and branching into leaf nodes covering individual documents, where the hierarchical tree includes a leaf node for each document of the plurality of documents. The system can include forming a forest of sub-trees of the hierarchical tree based on cluster criteria. Textual and numeric similarity measures can be used depending on the type and distribution of data in the input and output spaces.

Abstract: A method of clustering a plurality of documents having input and output space data is disclosed that uses both input and output space criteria. The method can include aggregating documents into clusters based on input and/or output space similarity measures, and then refining the clusters based on further input and/or output space similarity measures. Aggregating the documents into clusters can include forming a hierarchical tree based on the input and/or output space similarity measures where the hierarchical tree has a root node, branching into intermediate nodes, and branching into leaf nodes covering individual documents, where the hierarchical tree includes a leaf node for each document of the plurality of documents. The method can then include forming a forest of sub-trees of the hierarchical tree based on cluster criteria. Textual and numeric similarity measures can be used depending on the type and distribution of data in the input and output spaces.

Abstract: A physical activity monitoring method and system in one embodiment includes a communications network, a wearable sensor device configured to generate physiologic data associated with a sensed physiologic condition of a wearer, and to generate context data associated with a sensed context of the wearer, and to transmit the physiologic data and the context data over the communications network, a memory for storing the physiologic data and the context data, a computer and a computer program executed by the computer, wherein the computer program comprises computer instructions for rendering first data associated with the physiologic data and second data associated with the context data, and a user interface operably connected to the computer for rendering the first data and the second data.

Abstract: A real-time and privacy-preserving method for delivering personalized information to a user within a specific geographical location is provided. The method comprises the steps of: storing information specific to a user in a database maintained by a centralized brokerage service; transmitting a user initiated request to the centralized brokerage service to provide a listing of products and services offered by retailers or third parties located in the approximate current geographical vicinity of the user; utilizing the stored information of the user to generate a personalized listing of products and services offered within the approximate current geographical vicinity of the user; and sending the generated personalized listing to the user.

Abstract: A system for clustering a plurality of documents having input and output space data is disclosed that uses both input and output space criteria. The system can aggregate documents into clusters based on input and/or output space similarity measures, and then refine the clusters based on further input and/or output space similarity measures. Aggregation of documents into clusters can include forming a hierarchical tree based on the input and/or output space similarity measures where the hierarchical tree has a root node, branching into intermediate nodes, and branching into leaf nodes covering individual documents, where the hierarchical tree includes a leaf node for each document of the plurality of documents. The system can include forming a forest of sub-trees of the hierarchical tree based on cluster criteria. Textual and numeric similarity measures can be used depending on the type and distribution of data in the input and output spaces.

Abstract: A method of clustering a plurality of documents having input and output space data is disclosed that uses both input and output space criteria. The method can include aggregating documents into clusters based on input and/or output space similarity measures, and then refining the clusters based on further input and/or output space similarity measures. Aggregating the documents into clusters can include forming a hierarchical tree based on the input and/or output space similarity measures where the hierarchical tree has a root node, branching into intermediate nodes, and branching into leaf nodes covering individual documents, where the hierarchical tree includes a leaf node for each document of the plurality of documents. The method can then include forming a forest of sub-trees of the hierarchical tree based on cluster criteria. Textual and numeric similarity measures can be used depending on the type and distribution of data in the input and output spaces.

Abstract: A method for estimating values of missing data in partial sets of medical data includes generating a Gaussian mixture with a time-varying mean and time and lag varying covariances. The method generates the estimate for a missing datum with the Gaussian distribution having a selected mean and covariance corresponding to the time of the missing datum. An estimate of the missing datum is generated with reference to the mean of the Gaussian distribution conditioned on other medical data that are observed at the time of the missing datum.

Abstract: In one embodiment, a method for identifying patients to receive a medical device has been developed. The method includes retrieving first medical data associated with one patient from a database, identifying a probability that a medical device provides a medical benefit to the one patient that exceeds a cost associated with providing the medical device to the one patient with reference to the first medical data and a probabilistic model, the probabilistic model having a plurality of model parameters, each model parameter corresponding to one type of datum in the first medical data, and providing the medical device to the one patient in response to the identified probability exceeding a first predetermined threshold.