Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network. One of the methods includes maintaining a replay memory that stores trajectories generated as a result of interaction of an agent with an environment; and training an action selection neural network having policy parameters on the trajectories in the replay memory, wherein training the action selection neural network comprises: sampling a trajectory from the replay memory; and adjusting current values of the policy parameters by training the action selection neural network on the trajectory using an off-policy actor critic reinforcement learning technique.

Abstract: An accelerator architecture is presented for a Transformer machine learning model. The accelerator is comprised of: one or more memory devices, each memory device has a random access memory and is configured for processing in memory, where a key matrix, a value matrix, a query weight matrix, a key weight matrix, and a value weight matrix for an attention mechanism of the Transformer machine learning model are stored in the one or more memory devices; and a control device interfaced with each of the one or more memory devices, the control device is used to coordinate the vector matrix multiplication operations performed on the memory devices, perform other arithmetic and logic operations used in attention blocks that are not suited for the memory devices, and coordinate the updates of the key and value matrices in the one or more memory devices.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection neural network. One of the methods includes maintaining a replay memory that stores trajectories generated as a result of interaction of an agent with an environment; and training an action selection neural network having policy parameters on the trajectories in the replay memory, wherein training the action selection neural network comprises: sampling a trajectory from the replay memory; and adjusting current values of the policy parameters by training the action selection neural network on the trajectory using an off-policy actor critic reinforcement learning technique.

Abstract: An image sensor includes a pixel array, and the pixel array includes a first number of first pixel units and a second number of second pixel units, where the first pixel unit includes photosensitive elements stacked and configured to acquire photo signals of at least two colors. The second pixel unit includes photosensitive elements configured to acquire photo signals of a single color.

Abstract: A ferroelectric film phase shifter includes a substrate layer; an isolated signal layer located on the substrate layer; first, second and third top transmission line electrodes distributed on the isolated signal layer at intervals; the first and second top transmission line electrodes located at both ends of the isolated signal layer, and the third top transmission line electrode located on a middle region of the isolated signal layer; a bottom transmission line electrode located in the isolated signal layer; an intermediate transmission line structure located in a middle region of the bottom transmission line electrode and adjacent to the third top transmission line electrode; MIM hafnium oxide-based ferroelectric capacitor structures located at two ends of the bottom transmission line electrode; and metal transmission line structures located between each MIM hafnium oxide-based ferroelectric capacitor structure and each of the first top transmission line electrode and the second top transmission line electr

Abstract: A thermoelectric array display and a manufacturing method thereof are provided. The thermoelectric array display includes at least a first pixel, where the first pixel includes a bottom electrode, a P-type thermoelectric leg, an N-type thermoelectric leg, and a top electrode; the P-type thermoelectric leg is arranged on the bottom electrode; and the P-type thermoelectric leg is connected in series to the N-type thermoelectric leg by the top electrode. The thermoelectric array display has strong concealment of information transmission, can effectively reduce heat generation of a device, and can implement long-distance signal transmission.

Abstract: The present invention discloses a method for preparing an ibuprofen spherical crystal having high bulk density and a product thereof. The method includes: (1) heating ibuprofen until a molten liquid state is reached; and (2) pressurizing the molten liquid state of ibuprofen, dropping the molten liquid state of ibuprofen into a crystallizer filled with water through a liquid distributor for crystallization under stirring conditions until the molten liquid state of ibuprofen is completely dropped, lowering the rotation speed of stirring to allow crystal growing, and finally, subjecting a resulting crystal slurry to post-treatment to obtain the ibuprofen spherical crystal having high bulk density. The ibuprofen spherical crystal has a bulk density of 0.50-0.70 g/mL, a tap density of 0.63-0.89 g/mL, a median particle size of 300-1,000 ?m and an angle of repose of 22-29°.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for data-driven robotic control. One of the methods includes maintaining robot experience data; obtaining annotation data; training, on the annotation data, a reward model; generating task-specific training data for the particular task, comprising, for each experience in a second subset of the experiences in the robot experience data: processing the observation in the experience using the trained reward model to generate a reward prediction, and associating the reward prediction with the experience; and training a policy neural network on the task-specific training data for the particular task, wherein the policy neural network is configured to receive a network input comprising an observation and to generate a policy output that defines a control policy for a robot performing the particular task.

Abstract: A neural network control system for controlling an agent to perform a task in a real-world environment, operates based on both image data and proprioceptive data describing the configuration of the agent. The training of the control system includes both imitation learning, using datasets generated from previous performances of the task, and reinforcement learning, based on rewards calculated from control data output by the control system.

Abstract: A method of generating an output image having an output resolution of N pixels×N pixels, each pixel in the output image having a respective color value for each of a plurality of color channels, the method comprising: obtaining a low-resolution version of the output image; and upscaling the low-resolution version of the output image to generate the output image having the output resolution by repeatedly performing the following operations: obtaining a current version of the output image having a current K×K resolution; and processing the current version of the output image using a set of convolutional neural networks that are specific to the current resolution to generate an updated version of the output image having a 2K×2K resolution.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for data-driven robotic control. One of the methods includes maintaining robot experience data; obtaining annotation data; training, on the annotation data, a reward model; generating task-specific training data for the particular task, comprising, for each experience in a second subset of the experiences in the robot experience data: processing the observation in the experience using the trained reward model to generate a reward prediction, and associating the reward prediction with the experience; and training a policy neural network on the task-specific training data for the particular task, wherein the policy neural network is configured to receive a network input comprising an observation and to generate a policy output that defines a control policy for a robot performing the particular task.

Abstract: A system for performing real-time quantum-safe computation of a digital transaction using in a blockchain consensus protocol, comprising a plurality of permissioned verification servers being a plurality of distributed participants that are adapted to create common randomization to all of said participants which remains unrevealed until being used by said participants, by assigning to each participant a unique polynomial having a maximal degree being common to all participants; allowing each participant to select a random value; allowing each participant to send his selected random value to all other participants using a secret sharing scheme based on points on his unique polynomial, such that said secret hides the details of said selected random value and all other participants that receive shares of said selected random value will not be able to reconstruct said selected random value from the received shares; create a pool of all shares of all participants; build a quantum-safe consensus of honest participa

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an action selection policy neural network, wherein the action selection policy neural network is configured to process an observation characterizing a state of an environment to generate an action selection policy output, wherein the action selection policy output is used to select an action to be performed by an agent interacting with an environment. In one aspect, a method comprises: obtaining an observation characterizing a state of the environment subsequent to the agent performing a selected action; generating a latent representation of the observation; processing the latent representation of the observation using a discriminator neural network to generate an imitation score; determining a reward from the imitation score; and adjusting the current values of the action selection policy neural network parameters based on the reward using a reinforcement learning training technique.

Abstract: Disclosed herein is a hybrid polymeric material comprising a tropoelastin and a copolymer of a polyol monomer and a polycarboxylic acid monomer. The hybrid polymeric material is suitable for use as a tissue scaffold.

Abstract: A method of generating an output image having an output resolution of N pixels×N pixels, each pixel in the output image having a respective color value for each of a plurality of color channels, the method comprising: obtaining a low-resolution version of the output image; and upscaling the low-resolution version of the output image to generate the output image having the output resolution by repeatedly performing the following operations: obtaining a current version of the output image having a current K×K resolution; and processing the current version of the output image using a set of convolutional neural networks that are specific to the current resolution to generate an updated version of the output image having a 2K×2K resolution.

Abstract: A method is proposed of training a neural network to generate action data for controlling an agent to perform a task in an environment. The method includes obtaining, for each of a plurality of performances of the task, one or more first tuple datasets, each first tuple dataset comprising state data characterizing a state of the environment at a corresponding time during the performance of the task; and a concurrent process of training the neural network and a discriminator network. The training process comprises a plurality of neural network update steps and a plurality of discriminator network update steps.

Abstract: A method of generating an output image having an output resolution of N pixels×N pixels, each pixel in the output image having a respective color value for each of a plurality of color channels, the method comprising: obtaining a low-resolution version of the output image; and upscaling the low-resolution version of the output image to generate the output image having the output resolution by repeatedly performing the following operations: obtaining a current version of the output image having a current K×K resolution; and processing the current version of the output image using a set of convolutional neural networks that are specific to the current resolution to generate an updated version of the output image having a 2K×2K resolution.

Abstract: A high-sensitivity, high-selectivity and portable detection method for trace uranyl ion is described. The method has an ultralow detection limit of 11 pM/2.6 ppt and is useful in precise monitoring of the uranium content in agricultural and sideline products, foods, environments and so on. The test instrument is miniaturized and low in cost to achieve high-precision portable measurement in the field. A conjugated polymer with aggregation-induced emission (AIE) activity is synthesized, and prepared into Pdots, and a uranyl-responsive electrochemiluminescence (ECL) probe is developed by modifying the Pdots with DNA or RNA, which serves as an adsorption ligand of uranyl ion. The probe exhibits good biocompatibility. The ECL technology can be used in uranyl ion detection and the method has extremely high sensitivity. A uranyl ion probe with AIE activity is also disclosed, which can be applied in portable precise monitoring of trace uranyl ion by means of the ECL technology.