Abstract: A gate voltage bootstrap switching circuit includes an LDO, a first MOS transistor, a second MOS transistor, a third MOS transistor, and a voltage control unit. The LDO has an inverting input terminal of connected to its output terminal. Drain of the first MOS transistor and source of the third MOS transistor are connected to the output terminal, source of the first MOS transistor is connected to drain of the second MOS transistor, and source of the second MOS transistor is connected to drain of the third MOS transistor. Capacitor arrays are connected. The voltage control unit is connected to gates of the first, second, and third MOS transistors to input an external power supply voltage as a gate-source voltage to the MOS transistors. The circuit eliminates the non-linearity of the impedance of the switch MOS transistors and improves the conversion speed of the ADC.

Abstract: Systems and methods for alignment of in vivo and ex vivo images in accordance with embodiments of the invention are illustrated. One embodiment includes a method for aligning in vivo with ex vivo captured tissue images, including obtaining a first image captured in vivo, obtaining a second image captured ex vivo, identifying cells in the first image and the second image, generating a soma-print for cells in the images, where each soma-print includes a plurality of vectors from a cell to each of its n nearest neighboring cells, computing a pair-wise soma-print score for each pairing of cells between the first plurality of cells and the second plurality of cells, identifying matched cell pairings between the first plurality of cells and the second plurality of cells based on their soma-print score, and annotating at least one of the images with the matched cell pairings.

Abstract: A segmented capacitance calibration circuit applied in a pure capacitor array structure includes a first calibration unit, a second calibration unit and a selection switch which are all connected with a scaling capacitor. The first and second calibration units include at least one capacitor, and the selection switch is configured to select the first or second calibration unit to be connected to the pure capacitor array structure. When the first calibration unit is connected to the pure capacitor array structure and in parallel with the scaling capacitor, a negative error calibration is performed on the scaling capacitor. When the second calibration unit is connected to the pure capacitor array structure and in series with the scaling capacitor, a positive error calibration is performed on the scaling capacitor. The nonlinear problem caused by the precision error of the scaling capacitor in the pure capacitor array structure is solved.

Abstract: A method includes pre-training an audio encoder on a public training utterance set and from a corpus of text utterances, sampling a predetermined number of most frequent words that appear in the corpus of text utterances. The method also includes randomly generating a predetermined number of transcripts, and for each corresponding transcripts, processing, using a TTS system, the corresponding transcript to generate a corresponding synthetic speech utterance. The corresponding transcript and the corresponding synthetic speech utterance form a corresponding synthetic training sample. During a first fine-tuning stage, the method also includes fine-tuning an ASR model on the synthetic training samples.

Abstract: Implementations described herein are directed to techniques for eliminating and/or mitigating memorization by machine learning (ML) model(s). Processor(s) can obtain a plurality of training instances to be utilized in training a ML model, identify a plurality of compute cores (e.g., TPUs, GPUs, CPUs, FPGAs, ASICs, etc.), and generate a corresponding per-core gradient at each of the plurality of compute cores. Further, the processor(s) can update the ML model based on the corresponding per-core gradients. In generating the corresponding per-core gradient at a given compute core, the processor(s) can generate corresponding gradients based on a subset of the training instances, determine, based on the corresponding gradients, a corresponding mean gradient, and clip, based on a clipping bound, the corresponding mean gradient for the given compute core to generate the corresponding per-core gradient.

Abstract: A selector with a superlattice-like structure and a preparation method thereof are provided, which belong to the technical field of micro-nano electronics. The selector includes a substrate, and a first metal electrode layer, a superlattice-like layer, and a second metal electrode layer sequentially stacked on the substrate. The superlattice-like layer includes n+1 first sublayers and n second sublayers alternately stacked periodically. A material of the first sublayer is amorphous carbon, and a material of the second sublayer is a chalcogenide with gating property.

Abstract: A method includes obtaining an automatic speech recognition (ASR) model pre-trained on an initial training dataset, creating a set of canary speech utterances, and speeding up each canary speech utterance in the set of canary speech utterances. The operations also include fine-tuning the ASR model on the set of sped-up canary speech utterances and measuring un-intended memorization of the fine-tuned ASR model based on speech recognition results performed by the fine-tuned ASR model on the sped-up canary speech utterances.

Abstract: A gate voltage bootstrap switching circuit includes an LDO, a first MOS transistor, a second MOS transistor, a third MOS transistor, and a voltage control unit. The LDO has an inverting input terminal of connected to its output terminal. Drain of the first MOS transistor and source of the third MOS transistor are connected to the output terminal, source of the first MOS transistor is connected to drain of the second MOS transistor, and source of the second MOS transistor is connected to drain of the third MOS transistor. Capacitor arrays are connected. The voltage control unit is connected to gates of the first, second, and third MOS transistors to input an external power supply voltage as a gate-source voltage to the MOS transistors. The circuit eliminates the non-linearity of the impedance of the switch MOS transistors and improves the conversion speed of the ADC.

Abstract: An integrated joint module includes a housing, a driving assembly, a speed reduction assembly, a braking assembly and an encoding assembly. The housing includes a first housing and a second housing, an annular supporting platform is arranged on an inner side of the first housing. The driving assembly includes an output shaft, a stator embedded in the annular supporting platform, and a rotor connected with the output shaft and arranged on an inner side of the stator, the speed reduction assembly and the braking assembly are connected with two ends of the output shaft. The encoding assembly is arranged on a side of the braking assembly away from the driving assembly and connected with the output shaft, the second housing is sleeved on the encoding assembly and connected with the first housing. The integrated joint module helps to simplify the structure of the joint module and reduce the cost.

Abstract: The present disclosure provides a collaborative robot arm and a joint module. The joint module includes a housing, a driving assembly, and a multi-turn absolute encoder. The joint module detects the angular position of the output shaft and records a number of rotating revolutions of the output shaft only by means of the multi-turn absolute encoder. The multi-turn absolute encoder includes a base, a bearing, a rotating shaft, an encoding disk, and a circuit board, the encoding disk is rotatably connected with the base by the rotating shaft and the bearing, the circuit board is fixedly connected with the base, and the reading head on the circuit board detects the angular position of the output shaft cooperatively with the encoding disk, making the multi-turn absolute encoder be an integrated structure. The base and the rotating shaft are detachably connected with the housing and the output shaft respectively.

Abstract: The disclosure belongs to the field of micro-nano electronic materials, and in particular to a Se-based selector material, a selector unit, and a preparation method thereof. The Se-based selector material is a compound including Ge, Se, and B elements. The chemical formula of the Se-based selector material is (GexSe1?x)1?yByMz, in which the M element is at least one of In, Ga, Al, and Zn, and 0.1?x?0.9, 0.02?y?0.15, and 0?z?0.2. The problems of safety and stability of the existing material selection for the selector are solved by the selector material, the selector unit, and the preparation method thereof provided by the disclosure. In addition, the threshold voltage of the selector device prepared based on the Se-based selector material is adjustable, and the comprehensive performance is good.

Abstract: A segmented capacitance calibration circuit applied in a pure capacitor array structure includes a first calibration unit, a second calibration unit and a selection switch which are all connected with a scaling capacitor. The first and second calibration units include at least one capacitor, and the selection switch is configured to select the first or second calibration unit to be connected to the pure capacitor array structure. When the first calibration unit is connected to the pure capacitor array structure and in parallel with the scaling capacitor, a negative error calibration is performed on the scaling capacitor. When the second calibration unit is connected to the pure capacitor array structure and in series with the scaling capacitor, a positive error calibration is performed on the scaling capacitor. The nonlinear problem caused by the precision error of the scaling capacitor in the pure capacitor array structure is solved.

Abstract: A printing head and a method for applying a correction for mounting deviation of light-emitting chips are provided. The printing head includes a plurality of light-emitting chips. Each light-emitting chip includes a plurality of primary light-emitting elements that are continuously arranged. At least one of two adjacent light-emitting chips further includes at least one spare light-emitting element continuously and linearly arranged after the primary light-emitting elements. If the two adjacent light-emitting chips are both at a target mounting position, first N light-emitting units of one of the two light-emitting chips respectively face to first N light-emitting units of the other one of the two light-emitting chips, where N?1. Each two of the light-emitting units facing to each other form a group. One of the two light-emitting units in each of the groups is set to a light emission disabled state.

Abstract: The present disclosure provides a collaborative robot arm and a joint module. The joint module includes a housing, a driving assembly, and a multi-turn absolute encoder. The joint module detects the angular position of the output shaft and records a number of rotating revolutions of the output shaft only by means of the multi-turn absolute encoder. The multi-turn absolute encoder includes a base, a bearing, a rotating shaft, an encoding disk, and a circuit board, the encoding disk is rotatably connected with the base by the rotating shaft and the bearing, the circuit board is fixedly connected with the base, and the reading head on the circuit board detects the angular position of the output shaft cooperatively with the encoding disk, making the multi-turn absolute encoder be an integrated structure. The base and the rotating shaft are detachably connected with the housing and the output shaft respectively.

Abstract: An integrated joint module includes a housing, a driving assembly, a speed reduction assembly, a braking assembly and an encoding assembly. The housing includes a first housing and a second housing, an annular supporting platform is arranged on an inner side of the first housing. The driving assembly includes an output shaft, a stator embedded in the annular supporting platform, and a rotor connected with the output shaft and arranged on an inner side of the stator, the speed reduction assembly and the braking assembly are connected with two ends of the output shaft. The encoding assembly is arranged on a side of the braking assembly away from the driving assembly and connected with the output shaft, the second housing is sleeved on the encoding assembly and connected with the first housing. The integrated joint module helps to simplify the structure of the joint module and reduce the cost.

Abstract: A drive structure of a desktop robotic arm is disclosed, including a base and a turntable. The base is internally provided with a turntable drive motor and a turntable drive shaft, the turntable drive motor is drive-connected to the turntable drive shaft, and the turntable drive shaft is drive-connected to the turntable. The turntable is provided with an upper arm drive motor and a forearm drive motor. The turntable drive motor, the upper arm drive motor and the forearm drive motor are all servo motors with absolute value encoders. According to the drive structure of the desktop robotic arm, by using servo motors as the drive motors for controlling the turntable, an upper arm and a forearm, for which the absolute value encoders are correspondingly configured, control accuracy and driving power can be improved. Further, the present invention also discloses a desktop robotic arm and a robot.

Abstract: A joint module is provided which includes a housing, a driving assembly, a speed reduction assembly, a first fastener, a second fastener and a third fastener. The driving assembly includes an output shaft, a lower bearing seat and a lower bearing, an inner ring and an outer ring of the lower bearing are respectively connected with the output shaft and the lower bearing seat. The first fastener fixes the lower bearing seat on the housing, and the second fastener fixes the speed reduction assembly on the housing. The third fastener fixes the input shaft of the speed reduction assembly to the output shaft. The output shaft of the driving assembly and the input shaft of the speed reduction assembly are separate structural members, which not only allows the driving assembly and the speed reduction assembly to be tested separately before assembly, but also facilitates the later maintenance.

Abstract: Disclosed are a Bluetooth intelligent unlocking method, an apparatus, a device and a storage medium. The method includes: scanning terminal devices within a preset range based on a Bluetooth signal, and determining whether there is a terminal device according to a scanning result; in responding to that there is a terminal device, obtaining device information of the terminal device and verifying the device information; in responding to that the device information is validated, unlocking a Bluetooth lock upon receiving an unlocking command triggered based on a preset key; further, when there is the terminal device within the preset range, the present application unlocks the Bluetooth intelligent lock based on the unlocking command triggered by the preset key, so that the user does not need to unlock through the app of the mobile phone to improve the user experience.

Abstract: Disclosed are an intelligent unlocking method, an apparatus, a device and a storage medium. The method includes: waking up a Bluetooth module and broadcasting a Bluetooth signal upon receiving an unlocking instruction; determining whether a terminal device receiving the Bluetooth signal is a terminal device paired with the Bluetooth module successfully within a preset time period; and in responding to that the terminal device receiving the Bluetooth signal is the terminal device paired with the Bluetooth module successfully within the preset time period, establishing a Bluetooth connection with the terminal device, and changing the Bluetooth module to a sleep state after a Bluetooth lock is opened.