Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for determining neural network architectures. One of the methods includes generating, using a controller neural network, a batch of output sequences, each output sequence in the batch defining a respective architecture of a child neural network that is configured to perform a particular neural network task; for each output sequence in the batch: training a respective instance of the child neural network having the architecture defined by the output sequence; evaluating a performance of the trained instance of the child neural network on the particular neural network task to determine a performance metric for the trained instance of the child neural network on the particular neural network task; and using the performance metrics for the trained instances of the child neural network to adjust the current values of the controller parameters of the controller neural network.

Abstract: Methods and systems for dissolving an iron-containing ore are disclosed. For example, a method of processing and dissolving an iron-containing ore comprises: thermally reducing one or more non-magnetite iron oxide materials in the iron-containing ore to form magnetite in the presence of a reductant, thereby forming thermally-reduced ore; and dissolving at least a portion of the thermally-reduced ore using an acid to form an acidic iron-salt solution; wherein the acidic iron-salt solution comprises protons electrochemically generated in an electrochemical cell.

Abstract: A system for reducing water usage of hand wash station includes a Venturi nozzle contained within a faucet, an air self-closing valve connected to both Venturi nozzle and a first 1-way valve, a bellow pump connected to both the first 1-way valve and a second 1-way valve. A spring or multiple springs are attached to the bellow pump. A solid plate attached on the top of the bellow pump for user standing during hands washing, the weight of the user on the bellow pump pushing air to the Venturi nozzle. A water spring return valve connected to both Venturi nozzle and a fresh water tank, the water spring return valve using to turn off water, allowing only air blow out of the Venturi nozzle for hand drying.

Abstract: Assemblies including a bulk acoustic wave acoustic sensor die having a first and an opposing second major surface, the die including a piezoelectric structure, a first and a second electrode electrically connected to the piezoelectric structure, and an active surface on the first major surface of the die; a printed circuit board (PCB), the PCB having a first major surface and an opposing second major surface and including a slot spanning from the first major surface to the second major surface through the PCB; a first bond electrically and mechanically connecting the die to the PCB; and a second bond electrically and mechanically connecting the die to the PCB, wherein the first and the second bonds are located on either side of the slot through the PCB and the active surface of the die is above the slot in the PCB.

Abstract: A stationary electromagnetic inspection (EMI) apparatus and method are disclosed which includes a base; a plurality of hydraulic clamping rollers, connected to that base, operable to move a test pipe forward and backward in a translational direction only; a transverse flaw detection module comprising a plurality of sensor arms each equipped with rows of magnetic sensor arrays arranged in a stagger manner along an axis of a test pipe; and a longitudinal flaw detection module comprising magnetizers and magnetic sensor arrays each respectively arranged orthogonal to one another.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for determining neural network architectures. One of the methods includes generating, using a controller neural network, a batch of output sequences, each output sequence in the batch defining a respective architecture of a child neural network that is configured to perform a particular neural network task; for each output sequence in the batch: training a respective instance of the child neural network having the architecture defined by the output sequence; evaluating a performance of the trained instance of the child neural network on the particular neural network task to determine a performance metric for the trained instance of the child neural network on the particular neural network task; and using the performance metrics for the trained instances of the child neural network to adjust the current values of the controller parameters of the controller neural network.

Abstract: Methods and systems for producing are disclosed. A method for producing iron, for example, comprises: providing an iron-containing ore to a dissolution subsystem comprising a first electrochemical cell; wherein the first anolyte has a different composition than the first catholyte; dissolving at least a portion of the iron-containing ore using an acid to form an acidic iron-salt solution having dissolved first Fe3+ ions; providing at least a portion of the acidic iron-salt solution to the first cathodic chamber; first electrochemically reducing said first Fe3+ ions in the first catholyte to form Fe2+ ions; transferring the formed Fe2+ ions from the dissolution subsystem to an iron-plating subsystem having a second electrochemical cell; second electrochemically reducing a first portion of the transferred formed Fe2+ ions to Fe metal at a second cathode of the second electrochemical cell; and removing the Fe metal.

Abstract: Disclosed are systems, methods, and non-transitory computer-readable media for protocol independent data unit forwarding. A packet forwarding system receives a data unit comprising a header byte string via an input port. The packet forwarding system parses the data unit based on a header type determined based on the input port, yielding a parsing output describing the header byte string of the data unit. The packet forwarding system updates a metadata item associated with the data unit based on the parsing output and determines a packet forwarding instruction for forwarding the data unit to a destination based on the metadata item associated with the data unit. The packet forwarding system forwards the data unit to the destination based on the packet forwarding instruction and the metadata item associated with the data unit.

Abstract: A network appliance can be configured for storing a plurality of flow table entries in a flow table of a match-action pipeline, wherein the match-action pipeline is implemented via a packet processing circuit configured to process a plurality of network traffic flows associated with the plurality of flow table entries. An extended packet processing pipeline of the network appliance can read a flow table entry of the flow table. The extended packet processing pipeline can be implemented via a pipeline circuit. The extended packet processing pipeline can determine that a network traffic flow associated with the flow table entry is expired or terminated. The network appliance can delete the flow table entry from the flow table by processing a traffic flow deletion operation after determining that the network traffic flow is expired or terminated.

Abstract: A box-shaped air intake silencer with vertical baffles for a gas turbine system. The box-shaped silencer has a silencer housing having opposing ends with a first end closed and a second end opened to permit passage of air. Opposing sidewalls extend between the first end and the second end with each of the sidewalls operative to receive a perpendicular flow of intake air. A silencer having vertically, spaced baffles are disposed within the housing. A first end of the baffles face the first end of the housing and a second end of the baffles faces the second end of the housing. Each of the baffles extend horizontally between the first end and the second end of the housing. The baffles receive the perpendicular flow of intake air proximate the first end of the housing and direct the intake air towards the second end of the housing for passage therethrough.

Abstract: Generally, the present disclosure is directed to systems and methods that generate augmented training data for machine-learned models via application of one or more augmentation techniques to audiographic images that visually represent audio signals. In particular, the present disclosure provides a number of novel augmentation operations which can be performed directly upon the audiographic image (e.g., as opposed to the raw audio data) to generate augmented training data that results in improved model performance. As an example, the audiographic images can be or include one or more spectrograms or filter bank sequences.

Abstract: A computer-implemented method for performing computer vision with reduced computational cost and improved accuracy can include obtaining, by a computing system including one or more computing devices, input data comprising an input tensor having one or more dimensions, providing, by the computing system, the input data to a machine-learned convolutional attention network, the machine-learned convolutional attention network including two or more network stages, and, in response to providing the input data to the machine-learned convolutional attention network, receiving, by the computing system, a machine-learning prediction from the machine-learned convolutional attention network. The convolutional attention network can include at least one attention block, wherein the attention block includes a relative attention mechanism, the relative attention mechanism including the sum of a static convolution kernel with an adaptive attention matrix.

Abstract: Generally, the present disclosure is directed to systems and methods that generate augmented training data for machine-learned models via application of one or more augmentation techniques to audiographic images that visually represent audio signals. In particular, the present disclosure provides a number of novel augmentation operations which can be performed directly upon the audiographic image (e.g., as opposed to the raw audio data) to generate augmented training data that results in improved model performance. As an example, the audiographic images can be or include one or more spectrograms or filter bank sequences.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a neural network to perform a machine learning task using a momentum and sign based optimizer.

Abstract: Systems and methods for embodiments of graph based and machine learning artificial intelligence systems for generating access item recommendations in an identity management system are disclosed. Embodiments of the identity management systems disclosed herein may utilize a graph based approach, a machine learning based approach, and hybrid combinations thereof for generating access item recommendations.

Abstract: A reactor for coating particles includes a stationary vacuum chamber to hold a bed of particles to be coated, a vacuum port in an upper portion of the chamber, a chemical delivery system configured to inject a reactant or precursor gas into a lower portion of the chamber, a paddle assembly, and a motor to rotate a drive shaft of the paddle assembly. The lower portion of the chamber forms a half-cylinder. The paddle assembly includes a rotatable drive shaft extending through the chamber along the axial axis of the half cylinder, and a plurality of paddles extending radially from the drive shaft such that rotation of the drive shaft by the motor orbits the plurality of paddles about the drive shaft.

Abstract: The present disclosure provides an air conditioner comprising: an indoor heat exchanger to exchange heat between underground water and indoor air; an outdoor heat exchanger to exchange heat between underground water and outdoor air; a well; an underground water tank to store underground water; and a water pump to pump water from the well to the underground water tank. The indoor heat exchanger comprises a copper tube running through a row of parallel aluminum panels. Air is sucked by a fan through the indoor heat exchanger in the direction opposite to the underground water flow direction in the copper tube. Air is then passed through an evaporator unit of a compressor to be dehumidified. The underground water, after exchanging heat with the air to reach approximate room temperature, goes through a condensing unit of the compressor to release heat.

Abstract: The present disclosure relates to a method for detecting a mutation and a related non-transitory computer storage medium. Some embodiments of the present disclosure relate to a method for detecting a mutation. The method includes: receiving a computed tomography (CT) image of a lung; generating a first set of radiomics features based on the CT image through a first image processing model; determining a first region of the CT image through a segmentation model; generating a second set of radiomics features based on the first region of the CT image; and determining whether a mutation occurs based on the first and second sets of radiomics features through a classifier model.

Abstract: Integration driving mechanism for fin control assembly applied for UAV, aerial observation equipment, comprises: the control assembly, the fin root assembly, the fin shaft. The control assembly is small size, high temperature operation, waterproof ability, vibration resistance, easily replacement and reparation that is still warrant the operation condition.

Abstract: A method for determining a placement for machine learning model operations across multiple hardware devices is described.