Nec Corporation

Nec C Orporation

Nec Laboratories Europe Gmbh

Bin Cheng

Jonathan Fuerst

Mauricio Fadel Argerich

Apostolos Papageorgiou

Flavio Cirillo

Atsushi Kitazawa

Martin Bauer

Masahiro Hayakawa

Gurkan Solmaz

Everton Luis Berz

A method for using knowledge infusion for robust and transferable machine learning models includes receiving a plurality of adaptive and programmable knowledge functions comprising a plurality of strong functions and a plurality of weak functions. A knowledge model is generated based on the plurality of strong functions and the plurality of weak functions. A machine learning model is trained based on the generated knowledge model.

Automated knowledge infusion for robust and transferable machine learning

A method for reinforcement machine learning uses a reinforcement learning system that has an environment and an agent. The agent has a policy providing a mapping between states of the environment and actions. The method includes: determining a current state of the environment; determining, using the policy, a current policy output based on the current state; determining, using a knowledge function, a current knowledge function output based on the current state; determining an action based on the current policy output and the current knowledge function output; applying the action to the environment resulting in updating the current state and determining a reward; and updating the policy based on at least one of the current state and the reward.

Weakly supervised reinforcement learning

Systems and methods for controlling data usage in a distributed environment among multiple entity domains. A method include steps of: receiving, in a local entity domain, a data consuming application comprising or identifying at least a first analytics task, wherein the first analytics task processes data inputs to produce first output data; determining availability of the data inputs; interpreting data usage policies, or data control policies, to determine atomic actions to be executed, wherein at least one of the data usage policies indicates that one of the data inputs for the first analytics task must remain in a remote entity domain; and executing the atomic actions, wherein the atomic actions include dispatching the first analytics task to the remote entity domain for remote execution of the first analytics task and receiving the first output data of the first analytics task from the remote entity domain.

Methods and systems for de-centralized data sharing with enforced data usage control for cross-party data analytics

A method is for matching a set of first classes assigned to a first data set with a set of second classes assigned to a second data set. The method includes constructing, via a set of pre-processing functions, a plurality of alignment profiles such that at least one alignment profile is assigned to each of the first classes and each of the second classes. The method includes generating a comparison matrix for each group of the alignment profiles, such that each group includes at least one of the first classes and at least one of the second classes. The method includes training a first machine learning model, through supervised training, based on the generated comparison matrices and based on probabilistic labels generated by a second machine learning model.

Ontology matching based on weak supervision

A method of the invention manages data stream processing tasks of an application topology in a stream processing system, the stream processing system having one or more data producers, one or more compute nodes, and one or more result consumers, includes: assigning geographic scope granularities to the data stream processing tasks of the application topology; receiving, by a task manager from a result consumer of the one or more result consumers, a subscription in order to trigger data processing for computing a result, the subscription including geographic scope related information concerning the triggered data processing]; generating, by the task manager based on the geographic scope related information of the subscription, an execution plan including task instances for performing the triggered data processing; and generating, by the task manager from the execution plan, a deployment plan for deploying the task instances of the execution plan on the compute nodes.

Method and system for managing data stream processing

A method of providing for cross-device operations between user devices includes maintaining, using an intermediate device connected to the user devices, state information on different elements in different sub-views displayed respectively on first and second ones of the user devices. The first and second user devices are different from each other. The method further includes performing: migrating and synchronizing the state of one of the elements which is moved from the sub-view of the first user device to the sub-view of the second user device; and/or generating a shadow element on the sub-view of the second user device and linking the shadow element to one of the elements on the first user device which is duplicated from the sub-view of the first user device on the sub-view of the second user device, and updating the states of the linked elements.

Method and system of providing for cross-device operations between user devices

A method for supporting stream processing framework functionality in a stream processing system, the stream processing system including one or more input modules, a stream processing platform, and computing nodes, includes deploying, by the stream processing platform using the input modules, tasks of at least one stream processing topology on the computing nodes based on both stream processing topology-related information and stream processing topology-external information. The method additionally includes preparing and executing, by the stream processing platform, the tasks of the at least one stream processing topology on the computing nodes.

Method and system for supporting stream processing framework functionality

A method to provide content for user devices connected to an application server (WAS) via an intermediate device includes one or more of the user devices requesting content from the intermediate device. The intermediate device fetches the requested content, and executes the content. A content tree is generated based on the executed content, and the generated content tree is split into content sub trees for each user device. The respective content sub tree is provided to the respective user device. The content tree and/or content sub-trees are updated in case of a content change on at least one user device.

Growing community of inventors

Sandhausen, Germany

Bin Cheng