Xerox Corporation

Conduent Business Services, LLC

Hyun Joo Kim

Gueyoung Jung

Sriganesh Madhvanath

Andres Quiroz Hernandez

Nathan Gnanasambandam

A continuous electronic data stream of unlabeled data instances is received and fed into both a stream-based selection strategy and a pool-based selection strategy. The stream-based selection strategy is continuously applied to each of the unlabeled data instances to continually select stream-based data instances that are to be annotated. Additionally, the pool-based selection strategy is periodically applied to a pool of data obtained from the unlabeled data instances, to periodically select pool-based data instances that are to be annotated. Each time the pool-based selection strategy is applied, these methods automatically replace the stream-based data instances with the pool-based data instances. Also, these methods provide, on demand, access to allow a user to annotate the stream-based data instances and the pool-based data instances.

Hybrid active learning for non-stationary streaming data with asynchronous labeling

A method, non-transitory computer readable medium, and apparatus for adapting resources of the cluster of nodes for a real-time streaming workflow are disclosed. For example, the method receives a notification that a node of the cluster of nodes associated with an instance of a process of the real-time streaming workflow is predicted to be a bottleneck, identifies a number of hops to send a resource statement when the bottleneck is predicted that minimizes a ripple effect associated with transmitting the resource statement, transmits the resource statement to at least one or more nodes of the cluster of nodes within the number of hops, receives a response from one of the at least one or more nodes within the cluster of nodes and adapts a resource usage to the at least one of the one or more nodes within the cluster of nodes that the response was received from.

Method and apparatus for ripple rate sensitive and bottleneck aware resource adaptation for real-time streaming workflows

The present invention generally relates to systems and methods for executing scripts (a sequence of declarative operations) on large data sets. Some implementations store descriptions of previously-executed operations and associated input and output data sets. When executing similar operations on the same, a subset of, a superset of, or any fragment of data subsequently, some implementations detect duplication of operations and access previously-stored output data sets in order to re-use data and reduce the amount of execution, thus avoiding time-consuming duplicative computations.

Systems and methods for managing duplication of operations

A method, non-transitory computer readable medium, and apparatus for configuring a scheduling a job request in a data processing platform are disclosed. The method receives a new job request having a priority selected by a user, submits the new job request to an online job queue comprising a plurality of jobs, wherein each one of the plurality of jobs comprises a respective priority selected by a respective user and schedules the new job request and the plurality of jobs in the online job queue to one or more available worker nodes in a unit time slot based upon a comparison of the priority of the new job and the respective priority of the plurality of jobs in the online job queue, wherein the scheduling algorithm is based on one of: blocks having a variable size and a static processing time or blocks having a static size and a variable processing time.

Growing community of inventors

Monmouth Junction, NJ, United States of America

Hyun Joo Kim