Abstract: A neural network system is proposed. The neural network can be trained by model-based reinforcement learning to select actions to be performed by an agent interacting with an environment, to perform a task in an attempt to achieve a specified result. The system may comprise at least one imagination core which receives a current observation characterizing a current state of the environment, and optionally historical observations, and which includes a model of the environment. The imagination core may be configured to output trajectory data in response to the current observation, and/or historical observations. The trajectory data comprising a sequence of future features of the environment imagined by the imagination core. The system may also include a rollout encoder to encode the features, and an output stage to receive data derived from the rollout embedding and to output action policy data for identifying an action based on the current observation.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for environment simulation. In one aspect, a system comprises a recurrent neural network configured to, at each of a plurality of time steps, receive a preceding action for a preceding time step, update a preceding initial hidden state of the recurrent neural network from the preceding time step using the preceding action, update a preceding cell state of the recurrent neural network from the preceding time step using at least the initial hidden state for the time step, and determine a final hidden state for the time step using the cell state for the time step. The system further comprises a decoder neural network configured to receive the final hidden state for the time step and process the final hidden state to generate a predicted observation characterizing a predicted state of the environment at the time step.

Abstract: A neural network system is proposed to select actions to be performed by an agent interacting with an environment to perform a task in an attempt to achieve a specified result. The system may include a controller to receive state data and context data, and to output action data. The system may also include an imagination module to receive the state and action data, and to output consequent state data. The system may also include a manager to receive the state data and the context data, and to output route data which defines whether the system is to execute an action or to imagine. The system may also include a memory to store the context data.

Abstract: A neural network system is proposed. The neural network can be trained by model-based reinforcement learning to select actions to be performed by an agent interacting with an environment, to perform a task in an attempt to achieve a specified result. The system may comprise at least one imagination core which receives a current observation characterizing a current state of the environment, and optionally historical observations, and which includes a model of the environment. The imagination core may be configured to output trajectory data in response to the current observation, and/or historical observations. The trajectory data comprising a sequence of future features of the environment imagined by the imagination core. The system may also include a rollout encoder to encode the features, and an output stage to receive data derived from the rollout embedding and to output action policy data for identifying an action based on the current observation.

Abstract: A neural network system is proposed to select actions to be performed by an agent interacting with an environment to perform a task in an attempt to achieve a specified result. The system may include a controller to receive state data and context data, and to output action data. The system may also include an imagination module to receive the state and action data, and to output consequent state data. The system may also include a manager to receive the state data and the context data, and to output route data which defines whether the system is to execute an action or to imagine. The system may also include a memory to store the context data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for environment simulation. In one aspect, a system comprises a recurrent neural network configured to, at each of a plurality of time steps, receive a preceding action for a preceding time step, update a preceding initial hidden state of the recurrent neural network from the preceding time step using the preceding action, update a preceding cell state of the recurrent neural network from the preceding time step using at least the initial hidden state for the time step, and determine a final hidden state for the time step using the cell state for the time step. The system further comprises a decoder neural network configured to receive the final hidden state for the time step and process the final hidden state to generate a predicted observation characterizing a predicted state of the environment at the time step.

Abstract: A neural network system is proposed. The neural network can be trained by model-based reinforcement learning to select actions to be performed by an agent interacting with an environment, to perform a task in an attempt to achieve a specified result. The system may comprise at least one imagination core which receives a current observation characterizing a current state of the environment, and optionally historical observations, and which includes a model of the environment. The imagination core may be configured to output trajectory data in response to the current observation, and/or historical observations. The trajectory data comprising a sequence of future features of the environment imagined by the imagination core. The system may also include a rollout encoder to encode the features, and an output stage to receive data derived from the rollout embedding and to output action policy data for identifying an action based on the current observation.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for environment simulation. In one aspect, a system comprises a recurrent neural network configured to, at each of a plurality of time steps, receive a preceding action for a preceding time step, update a preceding initial hidden state of the recurrent neural network from the preceding time step using the preceding action, update a preceding cell state of the recurrent neural network from the preceding time step using at least the initial hidden state for the time step, and determine a final hidden state for the time step using the cell state for the time step. The system further comprises a decoder neural network configured to receive the final hidden state for the time step and process the final hidden state to generate a predicted observation characterizing a predicted state of the environment at the time step.

Abstract: A neural network system is proposed. The neural network can be trained by model-based reinforcement learning to select actions to be performed by an agent interacting with an environment, to perform a task in an attempt to achieve a specified result. The system may comprise at least one imagination core which receives a current observation characterizing a current state of the environment, and optionally historical observations, and which includes a model of the environment. The imagination core may be configured to output trajectory data in response to the current observation, and/or historical observations. The trajectory data comprising a sequence of future features of the environment imagined by the imagination core. The system may also include a rollout encoder to encode the features, and an output stage to receive data derived from the rollout embedding and to output action policy data for identifying an action based on the current observation.

Abstract: A neural network system is proposed to select actions to be performed by an agent interacting with an environment to perform a task in an attempt to achieve a specified result. The system may include a controller to receive state data and context data, and to output action data. The system may also include an imagination module to receive the state and action data, and to output consequent state data. The system may also include a manager to receive the state data and the context data, and to output route data which defines whether the system is to execute an action or to imagine. The system may also include a memory to store the context data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for environment simulation. In one aspect, a system comprises a recurrent neural network configured to, at each of a plurality of time steps, receive a preceding action for a preceding time step, update a preceding initial hidden state of the recurrent neural network from the preceding time step using the preceding action, update a preceding cell state of the recurrent neural network from the preceding time step using at least the initial hidden state for the time step, and determine a final hidden state for the time step using the cell state for the time step. The system further comprises a decoder neural network configured to receive the final hidden state for the time step and process the final hidden state to generate a predicted observation characterizing a predicted state of the environment at the time step.

Abstract: Methods, systems, and devices are described that provide for D2D synchronization. The methods, systems, and/or devices may include tools and techniques that provide for synchronizing a mobile device based on detection of a reliability alarm. A reliability alarm may be used between mobile devices, which is transmitted and/or received on specific D2D resources. Since the resources are reserved for the reliability alarm, a mobile device which was previously isolated from network synchronization will be able to receive the reliability alarm that a reliable synchronization signal is close when it moves within range of a reliable device. Once a reliability alarm is received the mobile device may free other resources to allow it to receive synchronization signals from the reliable devices. The mobile device may then synchronize with the network based on the received synchronization signals and transmit its own reliability alarm for subsequent isolated devices to use.

Abstract: A method, an apparatus, and a computer program product for wireless communication are provided in connection with enabling timing source selection and deselection in a decentralized manner for distributed D2D synchronization in densely populated communications systems. In an example, a communications device, functioning in a non-timing source (TS) mode, is equipped to receive a request for timing information during a synchronization channel. The communications device may further be equipped to determine whether to switch to a TS mode based on a selection utility metric value. In another example, a communications device, functioning in a TS mode, is equipped to transmit a TIB during a synchronization channel, and determine whether to switch to a non-TS mode based on a deselection utility metric value. In the TS mode, the UE is configured to transmit a TIB, while in the non-TS mode the UE is configured not to transmit the TIB.

Abstract: A method, an apparatus, and a computer program product for wireless communication are provided in connection with improving resource allocation for distributed D2D synchronization in densely populated communications systems. In an example, a communications device is equipped to transmit a synchronization signal during a beacon period of a synchronization channel. In an aspect, the synchronization channel may include the beacon period, a paging period, and a TIB period. The communications device may further be equipped to monitor at least one of the beacon period, the paging period, or the TIB period of the synchronization channel for one or more signals from one or more UEs in a D2D network, and determine whether to transmit information during at least one of the beacon period, the paging period, or the TIB period based at least in part on the monitoring.

Abstract: Methods, systems, and devices are described that provide for D2D synchronization. The methods, systems, and/or devices may include tools and techniques that provide for synchronizing a mobile device based on detection of a reliability alarm. A reliability alarm may be used between mobile devices, which is transmitted and/or received on specific D2D resources. Since the resources are reserved for the reliability alarm, a mobile device which was previously isolated from network synchronization will be able to receive the reliability alarm that a reliable synchronization signal is close when it moves within range of a reliable device. Once a reliability alarm is received the mobile device may free other resources to allow it to receive synchronization signals from the reliable devices. The mobile device may then synchronize with the network based on the received synchronization signals and transmit its own reliability alarm for subsequent isolated devices to use.

Abstract: A method, an apparatus, and a computer program product for wireless communication are provided in connection with improving resource allocation for distributed D2D synchronization in densely populated communications systems. In an example, a communications device is equipped to transmit a synchronization signal during a beacon period of a synchronization channel. In an aspect, the synchronization channel may include the beacon period, a paging period, and a TIB period. The communications device may further be equipped to monitor at least one of the beacon period, the paging period, or the TIB period of the synchronization channel for one or more signals from one or more UEs in a D2D network, and determine whether to transmit information during at least one of the beacon period, the paging period, or the TIB period based at least in part on the monitoring.

Abstract: A method, an apparatus, and a computer program product for wireless communication are provided in connection with enabling timing source selection and deselection in a decentralized manner for distributed D2D synchronization in densely populated communications systems. In an example, a communications device, functioning in a non-timing source (TS) mode, is equipped to receive a request for timing information during a synchronization channel. The communications device may further be equipped to determine whether to switch to a TS mode based on a selection utility metric value. In another example, a communications device, functioning in a TS mode, is equipped to transmit a TIB during a synchronization channel, and determine whether to switch to a non-TS mode based on a deselection utility metric value. In the TS mode, the UE is configured to transmit a TIB, while in the non-TS mode the UE is configured not to transmit the TIB.

Abstract: A method, an apparatus, and a computer program product for wireless communication are provided in connection with improving convergence to a common timing structure for devices in a distributed synchronization D2D network. In an example, a communications device is equipped to detect, by a UE, a synchronization signal during a listening slot duration scan of a communication channel. In an aspect, the listening slot duration may be defined based on a first timing structure, and the synchronization signal may be defined based on a second timing structure. The communications device may further be equipped to obtain timing information associated with the second timing structure from the synchronization signal, and determine whether the first timing structure or the second timing structure is a preferred timing structure.

Abstract: A system, method, and computer program for determining a descriptor, comprising calculating a maximum distance for a plurality of points in a sector between each of said plurality of points and an origin; calculating a minimal distance from one of said plurality of points and a target line, wherein said maximum distance is an initial value; computing a plurality of Fourier coefficients from said minimal distances; and defining an invariant descriptor from said Fourier coefficients, and appropriate means and computer-readable instructions.