Abstract: An online system performs predictions for real-time tasks and near real-time tasks based on available network bandwidth. A client device receives a regression based machine learning model. Responsive to receiving a task, the client device determines an available network bandwidth for the client device. If the available network bandwidth is below a threshold, the client device uses the regression based machine learning model to perform the task. If the client device determines that the network bandwidth is above the threshold, the client device extracts features of the task, serializes the extracted features, and transmits the serialized features to an online system, causing the online system to use a different machine learning model to perform the task based on the serialized features.

Abstract: Examples include maintaining a virtual pool of containers; receiving a request from a client for one of a plurality of services to performed; when the request includes client code, determining whether the request belongs to regular or priority queue based on two models; adding the request to an appropriate shard in the queue; getting the request from the selected one of the plurality of queues and assigning a container for the request from the virtual pool of containers, the client code to be executed in the container; and after the client code is executed in the container, deleting the container from the virtual pool.

Abstract: A contextual processing engine architecture. The architecture utilizes data objects retrieved from a database to form a new transactional item data structure as input into a contextual processing engine. The transactional data structure includes a prior context pointer to point to historical context. The historical context can be null for new transactions or one or more basis transaction item data structures for contextual transactions. The processing engine processes the input using process functions lists and aggregates the results for output.

Abstract: Examples include a method of predictive rate limiting for performing services requested by a client in a cloud computing system. The method includes receiving a request from a client for one of a plurality of services to be performed, the client belonging to an organization; and determining a current threshold for the organization by applying a real time data model and a historical data model, the real time data model generating a first threshold at least in part by determining a number of requests received from the organization over a first preceding period of time; the historical data model generating a second threshold, the historical data model being generated by applying a machine learning model to historical data stored during processing of previous requests for the plurality of services from the organization over a second preceding period of time, the current threshold being the average of the first threshold and the second threshold.

Abstract: An online system performs predictions for real-time tasks and near real-time tasks that need to be performed by a deadline. A client device receives a real-time machine learning based model associated with a measure of accuracy. If the client device determines that a task can be performed using predictions having less than the specified measure of accuracy, the client device uses the real-time machine learning based model. If the client device determines that a higher level of accuracy of results is required, the client device sends a request to an online system. The online system provides a prediction along with a string representing a rationale for the prediction.

Abstract: An online system performs predictions for real-time tasks and near real-time tasks based on available network bandwidth. A client device receives a regression based machine learning model. Responsive to receiving a task, the client device determines an available network bandwidth for the client device. If the available network bandwidth is below a threshold, the client device uses the regression based machine learning model to perform the task. If the client device determines that the network bandwidth is above the threshold, the client device extracts features of the task, serializes the extracted features, and transmits the serialized features to an online system, causing the online system to use a different machine learning model to perform the task based on the serialized features.

Abstract: An online system performs predictions for real-time tasks and near real-time tasks that need to be performed by a deadline. A client device receives a real-time machine learning based model associated with a measure of accuracy. If the client device determines that a task can be performed using predictions having less than the specified measure of accuracy, the client device uses the real-time machine learning based model. If the client device determines that a higher level of accuracy of results is required, the client device sends a request to an online system. The online system provides a prediction along with a string representing a rationale for the prediction.

Abstract: Examples include maintaining a virtual pool of containers; receiving a request from a client for one of a plurality of services to performed; when the request includes client code, determining whether the request belongs to regular or priority queue based on two models; adding the request to an appropriate shard in the queue; getting the request from the selected one of the plurality of queues and assigning a container for the request from the virtual pool of containers, the client code to be executed in the container; and after the client code is executed in the container, deleting the container from the virtual pool.

Abstract: An online system performs predictions for real-time tasks and near real-time tasks that need to be performed by a deadline. A client device receives a real-time machine learning based model associated with a measure of accuracy. If the client device determines that a task can be performed using predictions having less than the specified measure of accuracy, the client device uses the real-time machine learning based model. If the client device determines that a higher level of accuracy of results is required, the client device sends a request to an online system. The online system provides a prediction along with a string representing a rationale for the prediction.

Abstract: An online system performs predictions for real-time tasks and near real-time tasks that need to be performed by a deadline. A client device receives a real-time machine learning based model associated with a measure of accuracy. If the client device determines that a task can be performed using predictions having less than the specified measure of accuracy, the client device uses the real-time machine learning based model. If the client device determines that a higher level of accuracy of results is required, the client device sends a request to an online system. The online system provides a prediction along with a string representing a rationale for the prediction.

Abstract: Examples include a method of predictive rate limiting for performing services requested by a client in a cloud computing system. The method includes receiving a request from a client for one of a plurality of services to be performed, the client belonging to an organization; and determining a current threshold for the organization by applying a real time data model and a historical data model, the real time data model generating a first threshold at least in part by determining a number of requests received from the organization over a first preceding period of time; the historical data model generating a second threshold, the historical data model being generated by applying a machine learning model to historical data stored during processing of previous requests for the plurality of services from the organization over a second preceding period of time, the current threshold being the average of the first threshold and the second threshold.

Abstract: A training dataset having training instances is determined. Each training instance comprises first and second records and a second record and a label indicate whether there is a match between the first and second records. A matching score vector is determined for each such training instance, and comprises components storing match scores for extracted features from field values in the first and second records. Based on matching score vectors and a match objective function, match score thresholds are determined for the extracted features. Match rule(s) each of which comprises predicate(s) are generated. Each predicate makes a predication on whether two records match by comparing a match score derived from the two records against a match score threshold.