Abstract: A system for development of an Artificial Intelligence (AI) model while protecting sensitive user information includes: a confidential computing environment in which original prompts to the AI model written by users are collected; a trained synthetic prompt generator to generate synthetic prompts based on the original prompts, wherein the synthetic prompt generator generates anonymized synthetic prompts without sensitive user information identifiable from the original prompts; and a developer computing environment in which the synthetic prompts are submitted to the AI model under development to generate a dataset that includes the synthetic prompts and corresponding AI model output for analysis to determine updates for the AI model while protecting the sensitive user information of actual users.

Abstract: A data processing system implements receiving a plurality of natural language prompts for a language model; storing the prompts in a privacy protecting datastore in which the prompts are inaccessible from outside of the privacy protecting datastore; analyzing the prompts stored to generate embedding vectors representing the plurality of natural language prompts; storing the embedding vectors in the privacy protecting datastore; analyzing the embedding vectors using a clustering algorithm to generate clusters of related embedding vectors in the privacy protecting datastore; generating, for each cluster, theme information summarizing the embedding vectors of the cluster; and storing the theme information associated with each cluster of the plurality of clusters in a second datastore outside of the privacy protecting datastore, the second datastore being accessible for querying the data stored therein.

Abstract: A dataset comprising tables is received. Embeddings are generated for column titles of a table. Based on the embeddings, similar tables are clustered. The tables are organized into smaller clusters based on statistical similarities. Similarity scores are calculated for tables within the same cluster. A relatedness graph is created based on the similarity scores; similar tables are represented by nodes connected by edges. If the similarity score for a pair of tables exceeds a threshold, a table is deleted.

Abstract: Methods, systems, and computer programs are presented for generating a usage model for predicting user commands in an app. One method includes receiving model information from client devices. The model is obtained at each client device by training a machine-learning program with app usage data. The server generates synthetic data using the models from the client devices. A machine-learning program is trained using the synthetic data to obtain a global model, which receives as input information about recent commands entered on the app and generates an output with a prediction for the next command expected to be received by the app. The information of the global model is transmitted to a first client device, and the app provides at least one command option in the app user interface based on a prediction, generated by the global model, of the next command expected.

Abstract: Technology disclosed herein describe systems, methods, and devices for sending telemetry data comprising a representation of characteristics of an electronic document. In an implementation, a service operating on computing device generates a multi-dimensional representation of an electronic document which includes values that are indicative of characteristics of the document. From the multi-dimensional representation, the service generates a reduced dimension representation by converting the values to a single value. In some implementations, generating the reduced dimension representation of the document includes generating reduced resolution values of the data. The service then generates a noisy representation of the reduced dimension representation and sends telemetry data based at least on the noisy representation to a telemetry service.

Abstract: A method may include detecting an error during execution of a function of an application executing on a system; inputting an identifier of the error into a first client-side telemetry data filter model; based on the inputting, receiving an output value from the first client-side telemetry data filter model; determining that the output value is above an importance threshold; and in response to the determining, transmitting telemetry data associated with the error to a server computing device.

Abstract: Methods, systems, and computer programs are presented for generating a usage model for predicting user commands in an app. One method includes receiving model information from client devices. The model is obtained at each client device by training a machine-learning program with app usage data. The server generates synthetic data using the models from the client devices. A machine-learning program is trained using the synthetic data to obtain a global model, which receives as input information about recent commands entered on the app and generates an output with a prediction for the next command expected to be received by the app. The information of the global model is transmitted to a first client device, and the app provides at least one command option in the app user interface based on a prediction, generated by the global model, of the next command expected.

Abstract: A system and method for sampling telemetry events are provided. The method includes monitoring, using at least one listening module, execution of instructions by a processor, identifying, using the at least one listening module, one or more unique tags, determining, based on one or more trigger conditions, if the one or more unique tags is associated with at least one of the one or more trigger conditions, wherein the one or more trigger conditions are dynamically modifiable via a remote device, obtaining, based on the determination, a portion of a stack data structure associated with the execution of the instructions by the processor; and performing, based on the determination, at least one action.

Abstract: A system and method for sampling telemetry events are provided. The method includes receiving, by a cloud-based server, a plurality of telemetry events, related to an application, from a plurality of client devices; generating, by the cloud-based server, a sampling model for collecting a telemetry event based on the plurality of telemetry events, where the sampling model defines under what conditions the telemetry event is to be reported by a client device; generating, by the cloud-based server, an instruction for determining whether or not to report an incoming same telemetry event by the client device based on the sampling model; and providing, by the cloud-based server, the instruction to the client device, to allow the client device to determine whether or not to report the incoming same telemetry event based on the instruction.

Abstract: A method may include detecting an error during execution of a function of an application executing on a system; inputting an identifier of the error into a first client-side telemetry data filter model; based on the inputting, receiving an output value from the first client-side telemetry data filter model; determining that the output value is above an importance threshold; and in response to the determining, transmitting telemetry data associated with the error to a server computing device.

Abstract: Intelligent collection and analysis of in-app failure data is disclosed herein. Upon an application failure in a client device, the client device may collect failure information uniquely identifying a specific failure and provide the failure information to an analysis system. The analysis system may identify a specific failure that identifies the application and a specific portion of the code in the application, based on the failure information and match an action correlated to the specific failure where the action is uniquely designed to resolve the specific failure in the application. The action may include instructions for the client device used to intelligently lead to a resolution of the specific failure. The analysis system may transmit the action to the client device to perform the action and provide any follow up information to the analysis server. The analysis server may use the information to further analyze the specific failure.

Abstract: A system and method for sampling telemetry events are provided. The method includes receiving, by a cloud-based server, a plurality of telemetry events, related to an application, from a plurality of client devices; generating, by the cloud-based server, a sampling model for collecting a telemetry event based on the plurality of telemetry events, where the sampling model defines under what conditions the telemetry event is to be reported by a client device; generating, by the cloud-based server, an instruction for determining whether or not to report an incoming same telemetry event by the client device based on the sampling model; and providing, by the cloud-based server, the instruction to the client device, to allow the client device to determine whether or not to report the incoming same telemetry event based on the instruction.

Abstract: Intelligent collection and analysis of in-app failure data is disclosed herein. Upon an application failure in a client device, the client device may collect failure information uniquely identifying a specific failure and provide the failure information to an analysis system. The analysis system may identify a specific failure that identifies the application and a specific portion of the code in the application, based on the failure information and match an action correlated to the specific failure where the action is uniquely designed to resolve the specific failure in the application. The action may include instructions for the client device used to intelligently lead to a resolution of the specific failure. The analysis system may transmit the action to the client device to perform the action and provide any follow up information to the analysis server. The analysis server may use the information to further analyze the specific failure.

Abstract: A system and method for sampling telemetry events are provided. The method includes monitoring, using at least one listening module, execution of instructions by a processor, identifying, using the at least one listening module, one or more unique tags, determining, based on one or more trigger conditions, if the one or more unique tags is associated with at least one of the one or more trigger conditions, wherein the one or more trigger conditions are dynamically modifiable via a remote device, obtaining, based on the determination, a portion of a stack data structure associated with the execution of the instructions by the processor; and performing, based on the determination, at least one action.

Abstract: Methods, systems, and computer storage media for providing log files using logging system operations in a logging system. The logging system operations support memory mapping log files and asynchronously managing file operations. The logging system operations support selectively mapping segments of log files during write operations. The logging operations also support performing file operations (e.g., closing, opening, deleting and serializing files) advantageously as background processes. Selective memory mapping specifically includes incrementally mapping new segments of a log file up to a predetermined log file size limit. The logging operations support processing spare files using spare file memory mapping. A spare file replaces an existing log file to continue writing logging data using the spare file.

Abstract: Methods, systems, and computer storage media for providing log files using logging system operations in a logging system. The logging system operations support memory mapping log files and asynchronously managing file operations. The logging system operations support selectively mapping segments of log files during write operations. The logging operations also support performing file operations (e.g., closing, opening, deleting and serializing files) advantageously as background processes. Selective memory mapping specifically includes incrementally mapping new segments of a log file up to a predetermined log file size limit. The logging operations support processing spare files using spare file memory mapping. A spare file replaces an existing log file to continue writing logging data using the spare file.

Abstract: The invention relates to an apparatus and a method therewith to quantify visual field in infants. The invention provides a novel solution to the unmet need of detection of defects in the visual field in infants and/or babies and providing an apparatus with a software framework to minimize the errors of recording the readings and increasing their precision and analyzing said results automatically using the software embedded in the system. Said apparatus to quantify visual field in infants comprise of a light projection device onto a dome-shaped hemispherical structure with an infra-red camera along with fixation lights, a means for monitoring responses of the subject during testing, a means for recording and analyzing the readings. Said means to detect and record the eye/head movement in the subject during testing comprises said infra red (IR) camera connected to a device wherein software application is embedded.

Abstract: The invention relates to an apparatus and a method therewith to quantify visual field in infants. The invention provides a novel solution to the unmet need of detection of defects in the visual field in infants and/or babies and providing an apparatus with a software framework to minimize the errors of recording the readings and increasing their precision and analyzing said results automatically using the software embedded in the system. Said apparatus to quantify visual field in infants comprise of a light projection device onto a dome-shaped hemispherical structure with an infra-red camera along with fixation lights, a means for monitoring responses of the subject during testing, a means for recording and analyzing the readings. Said means to detect and record the eye/head movement in the subject during testing comprises said infra red (IR) camera connected to a device wherein software application is embedded.