Abstract: A method for receiving a full training data set including a plurality of individual training data set, dividing the plurality of individual training sets into N classes, where N is an integer greater than three, dividing the N classes into M full data classes and N-M partial data classes, performing training to obtain a trained fixed size machine learning (ML) classification model and a trained in-class confidence model, outputting a first set of prediction value(s) based on the performance of training, distributing each class of the N classes of individual training data sets to a different node of a distributed machine learning system; and outputting, from the nodes of the distributed machine learning system, a second set of prediction value(s) for each class of the N classes.

Abstract: The present invention provides a processor including a core circuit, a plurality of clock signal generation circuits, a multiplexer and a detection circuit is disclosed. The core circuit is supplied by a supply voltage. The plurality of clock signal generation circuits are configured to generate a plurality of clock signals with different frequencies, respectively, wherein a number of the plurality of clock signals is equal to or greater than three. The multiplexer is configured to receive the plurality of clock signals, and to select one of the plurality of clock signals to serve as an output clock signal according to a control signal, wherein the core circuit uses the output clock signal to serve as an operating clock. The detection circuit is configured to detect a level of the supply voltage received by the core circuit in a real-time manner, to generate the control signal.

Abstract: Net-based checking of a circuit design includes obtaining a circuit design comprising a plurality of polygons. Further, a shape of a first polygon of the plurality of polygons, and a shape of a second polygon of the plurality of polygons is determined. The shape of the first polygon differs from a shape of the second polygon. Violations within the circuit design are detected based on a comparison of the first polygon with the second polygon.

Abstract: An impact tool is provided, including a working portion and a handle. The working portion includes a housing which is made of rubber and a filler. The housing defines an inner space, and the filler is stuffed with the inner space. The filler is shear thickening non-Newtonian fluid. The handle is connected with the working potion.

Abstract: The present disclosure discloses an organic light-emitting device, which includes a first electrode and a hole injection layer laminated with each other and formed an ohmic contact therebetween. The hole injection layer is characterized by a carrier mobility of less than 2×10?5 CM2V?1S?1.

Abstract: A package substrate and a chip package structure using the same are provided. The package substrate includes a laminated board including first to third wiring layers, a pad array, a plurality of ground conductive structures, and a plurality of power conductive structures. At least one of the ground (or power) conductive structures includes two first ground (or power) conductive posts and a second ground (or power) conductive post. The two first ground (or power) conductive posts and the second ground (or power) conductive post are arranged along a first direction, and the second ground (or power) conductive post is located between two orthographic projections of the two first ground (or power) conductive posts. Each of the ground conductive structures in a first column and each of the power conductive structures in a second column are offset from each other in a second direction.

Abstract: An analog circuit is calibrated to perform neural network computing. Calibration input is provided to a pre-trained neural network that includes at least a given layer having pre-trained weights stored in the analog circuit. The analog circuit performs tensor operations of the given layer using the pre-trained weights. Statistics of calibration output from the analog circuit is calculated. Normalization operations to be performed during neural network inference are determined. The normalization operations incorporate the statistics of the calibration output and are performed at a normalization layer that follows the given layer. A configuration of the normalization operations is written into memory while the pre-trained weights stay unchanged.

Abstract: An audio receiver includes a first signal port, a second signal port, a power supply port, a power ground port, and an amplifier circuit. The first signal port is coupled to an audio signal line of a transmission interface. The second signal port is coupled to an audio ground line of the transmission interface. The power supply port is coupled to a power supply line of the transmission interface to generate a power supply level to the power supply line. The power ground port is connected to the ground level and to a power ground line of the transmission interface. When the audio receiver is outputting a power supply current to the audio source device through the power supply port via the power supply line, a connection state between the second signal port and the power ground port is at a high impedance state.

Abstract: A power supply circuit includes a first regulator and a second regulator. The first regulator is configured to generate a first output signal according to an input signal. A voltage value of the first output signal decreases according to the input signal and a first voltage threshold value at a power-off stage. The second regulator is configured to be enabled according to the first output signal to generate a second output signal according to the input signal. A voltage value of the second output signal decreases according to the input signal and a second voltage threshold value at the power-off stage. The second voltage threshold value is greater than the first voltage threshold value.

Abstract: A processor may generate an enforcement point. The enforcement point may include one or more adversarial detection models. The processor may receive user input data. The processor may analyze, at the enforcement point, the user input data. The processor may determine, from the analyzing, whether there is an adversarial attack in the user input data. The processor may generate an alert based on the determining.

Abstract: A circuitry applied to an electronic device having a Universal Serial Bus (USB) type-C connector is provided. The circuitry includes a transceiver circuit, a physical layer circuit and a processing circuit. In operations of the circuitry, the transceiver circuit is coupled to the USB type-C connector. The physical layer circuit is configured to directly utilize a plurality of first signals from the USB type-C connector as at least one portion of Ethernet signals, and process the first signals to generate a plurality of processed first signals. The processing circuit is configured to process the processed first signals to generate an output signal.

Abstract: An audio receiver includes a first signal port, a second signal port, a power supply port, a power ground port, and an amplifier circuit. The first signal port is coupled to an audio signal line of a transmission interface. The second signal port is coupled to an audio ground line of the transmission interface. The power supply port is coupled to a power supply line of the transmission interface to generate a power supply level to the power supply line. The power ground port is connected to the ground level and to a power ground line of the transmission interface. When the audio receiver is outputting a power supply current to the audio source device through the power supply port via the power supply line, a connection state between the second signal port and the power ground port is at a high impedance state.

Abstract: An approach for a fast and accurate word embedding model, “desc2vec,” for out-of-dictionary (OOD) words with a model learning from the dictionary descriptions of the word is disclosed. The approach includes determining that a target text element is not in a set of reference text elements, information describing the target text element is obtained. The information comprises a set of descriptive text elements. A set of vectorized representations for the set of descriptive text elements is determined. A target vectorized representation for the target text element is determined based on the set of vectorized representations using a machine learning model. The machine learning model is trained to represent a predetermined association between the set of vectorized representations for the set of descriptive text elements describing the target text element and the target vectorized representation.

Abstract: The present disclosure relates to an organic electroluminescent device and a preparation method thereof. The organic electroluminescent device includes a hole injection layer including a first doped layer and/or a second doped layer. The first doped layer includes a P-type dopant, and the second doped layer includes a P-type dopant and a hole transport material. The organic electroluminescent device includes a hole transport layer formed on the hole injection layer, and an electron blocking layer formed on the hole transport layer. A difference in HOMO energy level between the electron blocking layer and the hole transport layer is less than or equal to 0.2 eV. The power consumption of the organic electroluminescent device can be reduced. The lifetime of the mass production device therefore can be increased.

Abstract: A performance management method and an electronic device are provided. The method is applied to the electronic device with a system processor and includes: sensing a temperature of the electronic device and determining whether the temperature is greater than a first temperature setting value; when the temperature is not greater than the first temperature setting value, initiating a frequency increasing procedure; when the temperature is greater than the first temperature setting value, determining whether the temperature is greater than a second temperature setting value, where the second temperature setting value is greater than the first temperature setting value; when the temperature is greater than the first temperature setting value and is not greater than the second temperature setting value, initiating a first frequency reducing procedure; and when the temperature is greater than the second temperature setting value, initiating a second frequency reducing procedure or turning off the system processor.

Abstract: A control device and a circuit board are provided. The control device can cooperate with the circuit board, and includes a ball grid array. The ball grid array includes a plurality of power balls and a plurality of ground balls, which are jointly arranged in a ball region. The power balls and the ground balls are respectively divided into a plurality of power ball groups and a plurality of ground ball groups. One of the ground ball groups includes two ground balls and is adjacent to a power ball group. A ball pitch between the two ground balls is greater than that between one of the power balls and one of the ground balls adjacent to each other. The circuit board includes a contact pad array corresponding to the ball grid array of the control device so that the control device can be disposed on the circuit board.

Abstract: A computer-implemented method for automatically adjusting a Uniform Resource Locator (URL) seed list. The method includes crawling for documents based on a seed URL list. The method generates relations data from the documents using a Natural Language Processing (NLP) model. The method analyzes the relations data using an auto-seed model. The method modifies the seed URL list.

Abstract: A wireless communication device including a plurality of wireless transceivers and a controller is provided. Each wireless transceiver performs wireless transmission and reception using a respective Radio Access Technology (RAT). The controller enables the wireless transceivers to provide a plurality of wireless connections using a plurality of subscriber identities, and selects a first wireless connection from the plurality of wireless connections for communications between a user device and the wireless communication device. Also, the controller selects a second wireless connection from the plurality of wireless connections according to a service type of a packet flow associated with the user device, for communications between a service network and the wireless communication device, and assigns the second wireless connection for routing packets of the packet flow between the user device and the service network.

Abstract: The present disclosure discloses an organic light-emitting device, which includes a first electrode and a hole injection layer laminated with each other and formed an ohmic contact therebetween. The hole injection layer is characterized by a carrier mobility of less than 2×10?5 CM2V?1S?1.

Abstract: The present application provides an organic electroluminescent device and a manufacturing method thereof. A host material of a light-emitting layer of the organic electroluminescent device is co-evaporated by a donor host material and a acceptor host material in a same evaporation source to form an exciplex, which solves the technical problems such as low luminous efficiency, short service life or complicated operation process of the organic electroluminescent device in the prior art. The organic electroluminescent device according to the present application includes an anode, a cathode and a light-emitting layer disposed between the anode and the cathode, a host material of the light-emitting layer is formed by premixing a donor host material and a acceptor host material, the donor host material and the acceptor host material are co-evaporated in a same evaporation source to form an exciplex, and the host material is doped with a guest material.