Abstract: Various arrangements are presented for increasing a link margin of a wireless audio link. A short-range wireless communication link having a first physical layer (PHY) symbol rate is established between an audio source device and an audio output device. An audio stream is transmitted using the communication link, which includes a connected isochronous stream (CIS) link. A number of packet retransmissions are detected on the CIS. Based on the detected number of packet retransmissions on the CIS, the first PHY symbol rate of the CIS can be altered to a second PHY symbol rate for transmitting the audio stream.

Abstract: This application relates to the artificial intelligence field, and discloses a neural network search method and a related apparatus. The neural network search method includes: constructing attention heads in transformer layers by sampling a plurality of candidate operators during model search, to construct a plurality of candidate neural networks, and comparing performance of the plurality of candidate neural networks to select a target neural network with higher performance. In this application, a transformer model is constructed with reference to model search, so that a new attention structure with better performance than an original self-attention mechanism can be generated, and effect in a wide range of downstream tasks is significantly improved.

Abstract: A method for producing an aluminium alloy extrusion, consisting of an alloy with the composition ?0.30% by weight of silicon, ?0.40% by weight of iron, 0.01-1.1% by weight of manganese, ?0.30% by weight of magnesium, ?0.70% by weight of zinc, ?0.35% by weight of chromium, ?0.20% by weight of zirconium, ?0.25% by weight of titanium, ?0.20% by weight vanadium, ?0.10% by weight of copper, up to 0.15% by weight of other impurities, each not greater than 0.03% by weight and the balance aluminium, the method comprising the steps casting the molten metal into extrusion billet a) subjecting the billet to a homogenization treatment at a holding temperature of 550 to 620° C. for 6 to 10 hours b) heating the billet to a temperature of 400 to 550° C. c) extruding the billet to a tube.

Abstract: An audio output method and a system including first and second electronic devices and an audio output device. The first and second devices are connected to a network using a system authentication account, the first device has installed thereon a music application, and the second device has installed thereon a voice call application. When the audio output device outputs an audio signal corresponding to audio data of the music application, in response to the second electronic receiving an incoming call from the voice call application, the first device displays a first prompt corresponding to the incoming call. The first device sends an indication of answering the incoming call to the second device, which answers the incoming call and sends audio data of the incoming call to the first device, and the first device causes the audio output device to output an audio signal corresponding to the audio data of the incoming call.

Abstract: Features described herein pertain to systems and methods for wirelessly providing an audio stream. When audio that is to be output to an audio output device is associated with an application a set of parameters for modifying an established Connected Isochronous Stream (CIS) of a wireless link between an audio source and the audio output device can be determined and the CIS of the wireless link can be modified based on the set of parameters. The audio that is associated with the application can be output to the audio output device using the modified CIS of the wireless link.

Abstract: A method, including: identifying static application features of an application; identifying resource access features of the application; labeling a translation lookaside buffer (TLB) miss threshold of a runtime feature of the application; determining utilization of larger pages during the runtime based on the TLB miss threshold; and setting the TLB miss threshold based on the determined utilization of the larger pages.

Abstract: A torsion balance is provided which includes a twisting wire and a reflector. The twisting wire is a suspended carbon nanotube. The reflector is hung on the twisting wire. The reflector further includes a film, a first reflecting layer, and a second reflecting layer; and the film includes a first surface and a second surface opposite to the first surface, and the first reflecting layer is located on the first surface and the second reflecting layer is located on the second surface.

Abstract: Method, computer program product, and computer system are provided. Traffic is collected against a NoSQL database by an activity collector. A database transaction log is periodically extracted and analyzed. The collected traffic and the analyzed database transaction log are input to building a knowledge base of database access patterns. Current traffic is captured and used to compute an activity threshold. Traffic is directed to a workload processor based on the activity threshold. Traffic is directed to an intensive insert/update/delete (IUD) processor in response to the activity threshold exceeding a configured threshold. A plurality of temporary shards is generated along with an adaptive key and adaptive index in the plurality of temporary shards. The intensive IUD processor traffic is re-directed to the plurality of temporary shards while the activity threshold exceeds the configured threshold.

Abstract: Provided herein are compounds of the Formula (I): (I) and tautomers, stereoisomers and pharmaceutically acceptable salts and solvates thereof, wherein Rx, Ry, W, X, Y, Z, Ring A and (AA) have the meanings given in the specification, which are inhibitors of RET kinase and are useful in the treatment and prevention of diseases which can be treated with a RET kinase inhibitor, including RET-associated diseases and disorders.

Abstract: An electronic device includes a rectangular ceramic package structure having opposite first and second sides, an interior cavity that extends to an opening in the second side, opposite third and fourth sides spaced along a first direction, opposite fifth and sixth sides spaced along an orthogonal second direction, and non-conductive indents extending into the third and fourth sides. The device also includes a semiconductor die in the cavity, a lid that covers the opening and seals the cavity, a conductive terminal having a planar side exposed along the first side that is electrically coupled to a circuit of the semiconductor die and extends to a first one of the non-conductive indents, and conductive pins spaced apart from the conductive terminal and extending outward from the first side of the ceramic package structure along a third direction.

Abstract: A sensor package structure includes a substrate, a sensor chip disposed on the substrate along a predetermined direction for being electrically coupled to each other, a light-permeable layer, an adhesive layer having a ring-shape and sandwiched between the sensor chip and the light-permeable layer, and an encapsulant formed on the substrate. The adhesive layer is formed with at least one type of a buffering cavity, wave-shaped slots, and rectangular slots, which penetrate therethrough along the predetermined direction. The buffering cavity can be located in the adhesive layer, and any one type of the wave-shaped slots and the rectangular slots can be respectively recessed in an inner side and an outer side of the adhesive layer. A minimum width of the adhesive layer is greater than or equal to 50% of a predetermined width between the inner side and the outer side.

Abstract: Techniques are described for linked blockchains that allow computing devices to access information from an acceptance blockchain about acceptance reports stored in an enterprise delivery blockchain that are also linked to a product blockchain that stores information of the products of the acceptance reports. An acceptance report is indicative of operability of a product in a service provided by a service provider. A processor is configured to access the block in the enterprise delivery blockchain, retrieve the acceptance report from the block in the enterprise delivery blockchain, determine an identifier, identified in the block in the enterprise delivery blockchain, to a block in an acceptance blockchain, access the block in the acceptance blockchain via the determined identifier, retrieve from the block in the acceptance blockchain contextual information of the acceptance report, and output the acceptance report and contextual information of the acceptance report.

Abstract: This disclosure relates to the field of artificial intelligence, and discloses a data processing method. The method includes: obtaining a transformer model including a target network layer and a target module; and processing to-be-processed data by using the transformer model, to obtain a data processing result. The target module is configured to: perform a target operation on a feature map output at the target network layer, to obtain an operation result, and fuse the operation result and the feature map output, to obtain an updated feature map output. In this disclosure, the target module is inserted into the transformer model, and the operation result generated by the target module and an input are fused, so that information carried in a feature map output by the target network layer of the transformer model is increased.

Abstract: Provided herein are methods of determining that a subject has or is at risk of having a disease (e.g., cancer) using analysis of gene expression of loci identified in nucleic acid molecules derived from a cell-free biological sample of the subject.

Abstract: The present disclosure provides a neural network-based method for calibration and localization of an indoor inspection robot. The method includes the following steps: presetting positions for N label signal sources capable of transmitting radio frequency (RF) signals; computing an actual path of the robot according to numbers of signal labels received at different moments; computing positional information moved by the robot at a tth moment, and computing a predicted path at the tth moment according to the positional information; establishing an odometry error model with the neural network and training the odometry error model; and inputting the predicted path at the tth moment to a well-trained odometry error model to obtain an optimized predicted path. The present disclosure maximizes the localization accuracy for the indoor robot by minimizing the error of the odometer with the odometry calibration method.

Abstract: A system includes: a ring bus; a plurality of static uninterruptible power supplies (UPSs), each static UPS of the plurality of static UPSs including: at least one battery; an input that is electrically connected to a first external electrical power source; and an output that is electrically connected to a load, and, via a first corresponding choke, to the ring bus; at least one fuel-cell interface converter (FIC) that converts direct current (DC) electrical power to alternating current (AC) electrical power, each FIC of the at least one FIC being electrically connected to the ring bus via a second corresponding choke; and a fuel cell module corresponding to and electrically connected to each FIC, the fuel cell module including a fuel cell.

Abstract: A bearing unit includes a radially inner ring, a radially outer ring comprising a lubricant inlet, a plurality of rolling bodies interposed between the inner ring and the outer ring, and a first seal. The first seal includes an axially outer sealing ring, an axially inner sealing ring, and a first sealing lip formed on an axially inner surface of the axially inner sealing ring and extending axially inward to make sealing contact with an axially outer surface of the axially inner sealing ring. The first sealing lip prevents lubricant from flowing from an interior of the bearing unit to an exterior of the bearing unit at a first axial end of the bearing unit, and the first seal is disposed axially outside the inlet of the radially outer ring.

Abstract: Introduced here are computer programs and computer-implemented techniques for discovering malicious emails and then remediating the threat posed by those malicious emails in an automated manner. A threat detection platform may monitor a mailbox to which employees of an enterprise are able to forward emails deemed to be suspicious for analysis. This mailbox may be referred to as an “abuse mailbox” or “phishing mailbox.” The threat detection platform can examine emails contained in the abuse mailbox and then determine whether any of those emails represent threats to the security of the enterprise. For example, the threat detection platform may classify each email contained in the abuse mailbox as being malicious or non-malicious. Thereafter, the threat detection platform may determine what remediation actions, if any, are appropriate for addressing the threat posed by those emails determined to be malicious.

Abstract: Embodiments are provided for securing data access to machine learning training data at a plurality of distributed computing devices. Electronic content including original data that corresponds to a preferred data security level is divided into a plurality of microsegments. The plurality of microsegments is restrictively distributed to a plurality of computing devices which apply transcription labels to the plurality of microsegments. The labeled microsegments are reconstructed into training data which is then used to train a machine learning model while facilitating an improvement in data security of the original data included with the training data from the reconstructed microsegments.

Abstract: A method is provided for quantizing a neural network model performed by a processing system. The method comprises determining a scaling factor based on a distribution of weights associated with the neural network model, determining quantized weights based on the scaling factor and the weights associated with the distribution, determining a training loss of the neural network model based on the quantized weights during training of the neural network model, and determining an updated scaling factor for the neural network model based on a gradient of the training loss.