Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for controlling agents using reporter neural networks.

Abstract: Systems and methods are provided for learning world graphs to accelerate hierarchical reinforcement learning (HRL) for the training of a machine learning system. The systems and methods employ or implement a two-stage framework or approach that includes (1) unsupervised world graph discovery, and (2) accelerated hierarchical reinforcement learning by integrating the graph.

Abstract: A method for generating a synthetic image with predefined properties. The method includes the steps of providing first values which characterize the predefined properties of the image that is to be generated and attention weights which characterize a weighting of one of the first values and feeding sequentially the first values and assigned attention weights as input value pairs into an generative automated learning system that includes at least a recurrent connection. An image generation system and a computer program that are configured to carry out the method are also described.

Abstract: Systems and methods are provided for learning world graphs to accelerate hierarchical reinforcement learning (HRL) for the training of a machine learning system. The systems and methods employ or implement a two-stage framework or approach that includes (1) unsupervised world graph discovery, and (2) accelerated hierarchical reinforcement learning by integrating the graph.

Abstract: A smart wearable apparatus (102) includes a processor and a memory having a set of instructions that when executed by the processor causes the smart wearable apparatus to receive activity sensor data of an activity performed by a user. Further, the smart wearable apparatus is caused to send the activity sensor data to a content selection device (103) that selects content that is matched to the activity performed by the user so that the content is played in synchronization with the activity. Further, a process receives activity sensor data performed by a user. The process also sends the activity sensor data to a content selection device (103) that selects content that is matched to the activity performed by the user so that the content is played in synchronization with the activity.

Abstract: A method for generating a synthetic image with predefined properties. The method includes the steps of providing first values which characterize the predefined properties of the image that is to be generated and attention weights which characterize a weighting of one of the first values and feeding sequentially the first values and assigned attention weights as input value pairs into an generative automated learning system that includes at least a recurrent connection. An image generation system and a computer program that are configured to carry out the method are also described.

Abstract: The present principles generally relate to video processing and viewing, and particularly, to concurrent viewing of a video with other users and processing of user-added, real-time content. The present principles provide capabilities to create a shared video viewing experience which merge concurrent video watching with u user-provided real-time commenting and content. Users watching the same content at the same time may overlay graphical elements on the shared video to communicate with other concurrent viewers of the video. These graphical elements are annotations used to communicate with another viewer, or among a group of viewers, and are overlaid onto the video itself in real time during an interactive session as though the users are in concurrent conversations.

Abstract: An apparatus comprises a processor configured to play back a first media content segment; determine if a first source media content segment exists in a transition history, which is one of the first media content segment and a segment similar to the first media content segment, the transition history including transitions from at least one source media content segment to at least one destination media content segment; and if the first source media content segment exists, play back a first destination media content segment transitioned from the first source media content segment in the transition history.

Abstract: An apparatus comprises a processor configured to receive first and second ratings of a first segment of a media content respectively from second and third users; assign first and second weights respectively to the first and second ratings according to relationship in a social network between the first and second users and between the first and third users respectively; adjust the first and second ratings using the first and second weights, respectively; combine the adjusted first and second ratings to form the customized rating data of the first segment for the first user; and deliver the customized rating data of the first segment to the first user; and a memory storing the customized rating data.