Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for distributing content are disclosed. In one aspect, a method includes storing, in a data structure, data specifying a future live event. An opportunity to provide the specified content to a user at a user device is identified. It is determined that (i) a time of the opportunity is between a start time and an end time for the live event, (ii) that a user device is located in a same geographic region as the live event based on geographical data provided by the user device, and (iii) the user is interested in the live event based on attributes of the user matching attributes of other users that were identified as interested in the live event (e.g., based on evaluation of online search data). The content is provided for display at the user device.

Abstract: A method and system for adjusting ads auction using predicted user responses to an in-ad survey is provided. The method includes (1) providing a content item associated with an actionable object, which when selected, causes a computing device to present a plurality of interactive elements each corresponding to a different one of a plurality of reasons for restricting the content item; (2) receiving, from the computing device, data indicating a particular reason, of the plurality of reasons, for restricting the content item, and the particular reason corresponding to a particular interactive element, of the plurality of interactive elements, that was selected by the user; and (3) updating, using the received data, a content selection model for selecting content items, wherein the content selection model is associated with the user.

Abstract: A deep state space generative model is augmented with intervention prediction. The state space model provides a principled way to capture the interactions among observations, interventions, critical event occurrences, true states, and associated uncertainty. The state space model can include a discrete-time hazard rate model that provides flexible fitting of general survival time distributions. The state space model can output a joint prediction of event risk, observation and intervention trajectories based on patterns in temporal progressions, and correlations between past measurements and interventions.

Abstract: A method and system for adjusting ads auction using predicted user responses to an in-ad survey is provided. The method includes (1) providing a content item associated with an actionable object, which when selected, causes a computing device to present a plurality of interactive elements each corresponding to a different one of a plurality of reasons for restricting the content item; (2) receiving, from the computing device, data indicating a particular reason, of the plurality of reasons, for restricting the content item, and the particular reason corresponding to a particular interactive element, of the plurality of interactive elements, that was selected by the user; and (3) updating, using the received data, a content selection model for selecting content items, wherein the content selection model is associated with the user.

Abstract: A method includes providing a first set of data objects to a first skip router of a neural network (NN). The NN includes a first NN layer and a second NN layer. The first set of data objects is subdivided into a first set of skip objects and a first set of non-skip objects based on a first skip logic implemented by the first skip router and a first context of each data object in the first set of data objects. A first set of processed objects is generated based on the first set of non-skip objects and a first layer logic implemented by the first NN layer. Predictions are generated based on a second set of data objects and a second layer logic implemented by the second NN layer. The second set of data objects includes the first set of processed objects and the first set of skip objects.

Abstract: The present disclosure selects third party content based on feedback. A selector identifies several content items including first and second content items (or more) responsive to a request. A machine learning engine determines a first feature of the first content item, a second feature of the second content item, and a third feature of the web page or a device associated with the request. The machine learning engine determines, responsive to the first feature and the third feature, a first score for the first content item based on a machine learning model generated using historical signals received from devices via a metadata channel formed from an electronic feedback interface. The machine learning engine determines a second score for the second content item responsive to the second feature and the third feature. A bidding module determines a price for the first content item based on the first and second scores.

Abstract: Systems and methods of evaluating information in a computer network environment are provided. A data processing system can obtain or receive a content placement criterion, such as a keyword, associated with a content item and can determine a quality metric of the content placement criterion. The data processing system can identify a candidate content placement criterion and expand placement criteria associated with the content item to include the content placement criterion and the candidate content placement criterion based at least in part on an evaluation of the quality metric of the content placement criterion. The data processing system can expand placement criteria based in part on a throttling parameter. The data processing system can identify a correlation between a document and the placement criteria to identify appropriate content items for the document.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for performing neural architecture search for machine learning models. In one aspect, a method comprises receiving training data for a machine learning, generating a plurality of candidate neural networks for performing the machine learning task, wherein each candidate neural network comprises a plurality of instances of a layer block composed of a plurality of layers, for each candidate neural network, selecting a respective type for each of the plurality of layers from a set of layer types that comprises, training the candidate neural network and evaluating performance scores for the trained candidate neural networks as applied to the machine learning task, and determining a final neural network for performing the machine learning task based at least on the performance scores for the candidate neural networks.

Abstract: A system for predicting and summarizing medical events from electronic health records includes a computer memory storing aggregated electronic health records from a multitude of patients of diverse age, health conditions, and demographics including medications, laboratory values, diagnoses, vital signs, and medical notes. The aggregated electronic health records are converted into a single standardized data structure format and ordered arrangement per patient, e.g., into a chronological order. A computer (or computer system) executes one or more deep learning models trained on the aggregated health records to predict one or more future clinical events and summarize pertinent past medical events related to the predicted events on an input electronic health record of a patient having the standardized data structure format and ordered into a chronological order.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for distributing content are disclosed. In one aspect, a method includes storing, in a data structure, data specifying a future live event. An opportunity to provide the specified content to a user at a user device is identified. It is determined that (i) a time of the opportunity is between a start time and an end time for the live event, (ii) that a user device is located in a same geographic region as the live event based on geographical data provided by the user device, and (iii) the user is interested in the live event based on attributes of the user matching attributes of other users that were identified as interested in the live event (e.g., based on evaluation of online search data). The content is provided for display at the user device.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for distributing content are disclosed. In one aspect, a method includes storing, in a data structure, data specifying a future live event. An opportunity to provide the specified content to a user at a user device is identified. It is determined that (i) a time of the opportunity is between a start time and an end time for the live event, (ii) that a user device is located in a same geographic region as the live event based on geographical data provided by the user device, and (iii) the user is interested in the live event based on attributes of the user matching attributes of other users that were identified as interested in the live event (e.g., based on evaluation of online search data). The content is provided for display at the user device.

Abstract: Systems and methods of evaluating information in a computer network environment are provided. A data processing system can obtain or receive a content placement criterion, such as a keyword, associated with a content item and can determine a quality metric of the content placement criterion. The data processing system can identify a candidate content placement criterion and expand placement criteria associated with the content item to include the content placement criterion and the candidate content placement criterion based at least in part on an evaluation of the quality metric of the content placement criterion. The data processing system can expand placement criteria based in part on a throttling parameter. The data processing system can identify a correlation between a document and the placement criteria to identify appropriate content items for the document.

Abstract: A system for predicting and summarizing medical events from electronic health records includes a computer memory storing aggregated electronic health records from a multitude of patients of diverse age, health conditions, and demographics including medications, laboratory values, diagnoses, vital signs, and medical notes. The aggregated electronic health records are converted into a single standardized data structure format and ordered arrangement per patient, e.g., into a chronological order. A computer (or computer system) executes one or more deep learning models trained on the aggregated health records to predict one or more future clinical events and summarize pertinent past medical events related to the predicted events on an input electronic health record of a patient having the standardized data structure format and ordered into a chronological order.

Abstract: A system for predicting and summarizing medical events from electronic health records includes a computer memory storing aggregated electronic health records from a multitude of patients of diverse age, health conditions, and demographics including medications, laboratory values, diagnoses, vital signs, and medical notes. The aggregated electronic health records are converted into a single standardized data structure format and ordered arrangement per patient, e.g., into a chronological order. A computer (or computer system) executes one or more deep learning models trained on the aggregated health records to predict one or more future clinical events and summarize pertinent past medical events related to the predicted events on an input electronic health record of a patient having the standardized data structure format and ordered into a chronological order.

Abstract: Systems and methods of providing information via a computer network are provided. A data processing system can identify a cluster that includes a plurality of online content items having a semantic or user similarity. The data processing system determines a plurality of cluster placement criteria of the cluster, and receives content configured for display with a web page. The content can be associated with the cluster based on the semantic or user similarity. A cluster placement criterion of the plurality of cluster placement criteria can be selected based on a quality metric of the selected cluster placement criterion, and the selected cluster placement criterion can be provided as a supplemental criterion used to select the content for display with the web page.

Abstract: Systems and methods of evaluating information in a computer network environment are provided. A data processing system can obtain or receive a content placement criterion, such as a keyword, associated with a content item and can determine a quality metric of the content placement criterion. The data processing system can identify a candidate content placement criterion and expand placement criteria associated with the content item to include the content placement criterion and the candidate content placement criterion based at least in part on an evaluation of the quality metric of the content placement criterion. The data processing system can expand placement criteria based in part on a throttling parameter. The data processing system can identify a correlation between a document and the placement criteria to identify appropriate content items for the document.

Abstract: A deep state space generative model is augmented with intervention prediction. The state space model provides a principled way to capture the interactions among observations, interventions, critical event occurrences, true states, and associated uncertainty. The state space model can include a discrete-time hazard rate model that provides flexible fitting of general survival time distributions. The state space model can output a joint prediction of event risk, observation and intervention trajectories based on patterns in temporal progressions, and correlations between past measurements and interventions.

Abstract: Methods and systems for displaying content items on an information resource include identifying a supplemental content item to append to a DOM tree of the information resource and identifying a first content item displayed in a first container of the information resource. The methods also includes monitoring a position of a predetermined portion of the first content item relative to a viewport of the application and determining, responsive to monitoring the position of the predetermined portion of the first content item, that the first content item is visible within the viewport of the application. The method also includes appending the supplemental content item in a second container at a first position between the first container and an end of the information resource and displaying the supplemental content item within the viewport of the application responsive to detecting a scroll action towards the end of the information resource.

Abstract: The present disclosure selects third party content based on feedback. A selector identifies several content items including first and second content items (or more) responsive to a request. A machine learning engine determines a first feature of the first content item, a second feature of the second content item, and a third feature of the web page or a device associated with the request. The machine learning engine determines, responsive to the first feature and the third feature, a first score for the first content item based on a machine learning model generated using historical signals received from devices via a metadata channel formed from an electronic feedback interface. The machine learning engine determines a second score for the second content item responsive to the second feature and the third feature. A bidding module determines a price for the first content item based on the first and second scores.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for distributing content are disclosed. In one aspect, a method includes storing, in a data structure, data specifying a future live event. An opportunity to provide the specified content to a user at a user device is identified. It is determined that (i) a time of the opportunity is between a start time and an end time for the live event, (ii) that a user device is located in a same geographic region as the live event based on geographical data provided by the user device, and (iii) the user is interested in the live event based on attributes of the user matching attributes of other users that were identified as interested in the live event (e.g., based on evaluation of online search data). The content is provided for display at the user device.