User-level privacy preservation is implemented within federated machine learning. An aggregation server may distribute a machine learning model to multiple users each including respective private datasets. Individual users may train the model using the local, private dataset to generate one or more parameter updates. Prior to sending the generated parameter updates to the aggregation server for incorporation into the machine learning model, a user may modify the parameter updates by applying respective noise values to individual ones of the parameter updates to ensure differential privacy for the dataset private to the user. The aggregation server may then receive the respective modified parameter updates from the multiple users and aggregate the updates into a single set of parameter updates to update the machine learning model. The federated machine learning may further include iteratively performing said sending, training, modifying, receiving, aggregating and updating steps.

A computerized method for transforming a user interface according to machine learning includes selecting a persona and determining whether a first condition is true for an associated data structure. In response to determining the first condition is true, the method includes determining whether a second condition is true. In response to determining the second condition is not true, the method includes loading a first trained machine learning model, inputting a first set of explanatory variables to generate a first metric, and transforming the user interface according to the first metric. In response to determining the second condition is true, the method includes determining whether a third condition is true. In response to determining the third condition is true, loading a second trained machine learning model, inputting a second set of explanatory variables to generate a second metric, and transforming the user interface according to the second metric.

A computerized method for transforming a user interface according to machine learning includes selecting a persona and determining whether a first condition is true for an associated data structure. In response to determining the first condition is true, the method includes determining whether a second condition is true. In response to determining the second condition is not true, the method includes loading a first trained machine learning model, inputting a first set of explanatory variables to generate a first metric, and transforming the user interface according to the first metric. In response to determining the second condition is true, the method includes determining whether a third condition is true. In response to determining the third condition is true, loading a second trained machine learning model, inputting a second set of explanatory variables to generate a second metric, and transforming the user interface according to the second metric.

A management system operates in conjunction with entities to provide component recommendations for objects. The management system trains a machine learning model used to generate the component recommendations. The machine learning model is trained based on historical component entries describing components previously provided and identifiers of the components. The management system generates training data by classifying the historical component entries into predetermined component classifications. After the machine learning model is trained, the management system generates a customized recommendation of components for an object based on likelihoods of selection of the predetermined component classifications.

A management system operates in conjunction with entities to provide component recommendations for objects. The management system trains a machine learning model used to generate the component recommendations. The machine learning model is trained based on historical component entries describing components previously provided and identifiers of the components. The management system generates training data by classifying the historical component entries into predetermined component classifications. After the machine learning model is trained, the management system generates a customized recommendation of components for an object based on likelihoods of selection of the predetermined component classifications.

In some implementations, a credit decision platform may receive a credit request from an applicant and obtain domestic historical data associated with the applicant from a credit bureau device. The credit decision platform may obtain access to an email account associated with the applicant based on determining that the domestic historical data associated with the applicant is insufficient to process the credit request. The credit decision platform may identify, using one or more machine learning models, a set of email messages included in the email account that are relevant to the credit request and may analyze content included in the set of email messages to generate non-domestic historical data associated with the applicant. The credit decision platform may generate a decision on the credit request based on an estimated creditworthiness of the applicant, which may be determined based on the non-domestic historical data.

Methods and systems for training a model using machine learning for automatically distributing medical imaging studies to radiologists. One method includes receiving one or more medical images included in a medical study, each of the one or more medical images including image metadata defining characteristics of the corresponding medical image. The method further includes receiving radiologist metadata for each one of the plurality of radiologists, generating a state representation of the image metadata and the radiologist metadata, and providing the state representation to the model. The method further includes assigning, with the model, at least one of the one or more medical images to one of the plurality of radiologists, calculating feedback based on a change in the state representation after the at least one of the one or more medical images is assigned to one of the plurality of radiologists, and adjusting the model based on the feedback.

Methods and systems for training a model using machine learning for automatically distributing medical imaging studies to radiologists. One method includes receiving one or more medical images included in a medical study, each of the one or more medical images including image metadata defining characteristics of the corresponding medical image. The method further includes receiving radiologist metadata for each one of the plurality of radiologists, generating a state representation of the image metadata and the radiologist metadata, and providing the state representation to the model. The method further includes assigning, with the model, at least one of the one or more medical images to one of the plurality of radiologists, calculating feedback based on a change in the state representation after the at least one of the one or more medical images is assigned to one of the plurality of radiologists, and adjusting the model based on the feedback.

Methods and systems to provide a form of probabilistic labeling to associate an outage with a disturbance, which could itself be either known based on the available data or unknown. In the latter case, labeling is especially challenging, as it necessitates the discovery of the disturbance. One approach incorporates a statistical change-point analysis to time-series events that correspond to service tickets in the relevant geographic sub-regions. The method is calibrated to separate the regular periods from the environmental disturbance periods, under the assumption that disturbances significantly increase the rate of loss-causing events. To obtain the probability that a given loss-causing event is related to an environmental disturbance, the method leverages the difference between the rate of events expected in the absence of any disturbances (baseline) and the rate of actually observed events. In the analysis, the local disturbances are identified and estimators of their duration and magnitude are provided.

Methods and systems to provide a form of probabilistic labeling to associate an outage with a disturbance, which could itself be either known based on the available data or unknown. In the latter case, labeling is especially challenging, as it necessitates the discovery of the disturbance. One approach incorporates a statistical change-point analysis to time-series events that correspond to service tickets in the relevant geographic sub-regions. The method is calibrated to separate the regular periods from the environmental disturbance periods, under the assumption that disturbances significantly increase the rate of loss-causing events. To obtain the probability that a given loss-causing event is related to an environmental disturbance, the method leverages the difference between the rate of events expected in the absence of any disturbances (baseline) and the rate of actually observed events. In the analysis, the local disturbances are identified and estimators of their duration and magnitude are provided.