Abstract: A method of audio processing includes capturing a binaural audio signal, calculating noise reduction gains using a machine learning model, and generating a modified binaural audio signal. The method may further including performing various corrections to the audio to account for video captured by different cameras such as a front camera and a rear camera. The method may further include performing smooth switching of the binaural audio when switching between the front camera and the rear camera. In this manner, noise may be reduced in the binaural audio, and the user perception of the combined video and binaural audio may be improved.

Abstract: The irregularly-shaped display panel comprises a plurality of pixel units arranged in an array and a shielding layer. some of the pixel units are shielded by the shielding layer to form irregularly-shaped pixel units, and each of the irregularly-shaped pixel unit at least comprises a first irregularly-shaped pixel unit; in the first irregularly-shaped pixel unit, the areas of opening regions of at least two sub-pixels shielded by the shielding layer are different, and the sub-pixel having a larger area of an opening region shielded by the shielding layer has a greater pixel pressure difference.

Abstract: Embodiments described herein provide systems and methods for training a sequential recommendation model. Methods include determining a difficulty and quality (DQ) score associated with user behavior sequences from a training dataset. User behavior sequences are sampled during training based on their DQ scores. A meta-extrapolator may also be trained based on user behavior sequences sampled according to DQ score. The meta-extrapolator may be trained with high quality low difficulty sequences. The meta-extrapolator may then be used with an input of high quality high difficulty sequences to generate synthetic user behavior sequences. The synthetic user behavior sequences may be used to augment the training dataset to fine-tune the sequential recommendation model, while continuing to sample user behavior sequences based on DQ score. As the DQ score is based on current model predictions, DQ scores iteratively update during the training process.

Abstract: A capsule endoscope image recognition method based on deep learning, and a device and a medium is provided. After a plurality of frames of continuously captured images are processed to form an image sequence of a specific format, multi-channel recognition is performed on the plurality of frames of images by means of a 3D convolutional neural network model, and a recognition result is then output in combination with a recognition probability of each channel, such that the image recognition precision is improved.

Abstract: A chip slot is disclosed, which includes a slot, where a plurality of terminal groups are disposed in the slot, terminals in each terminal group include metal sheets that are symmetrically disposed on two opposite inner side walls of the slot, and each metal sheet has a bending pin that extends outside the slot; bending directions of bending pins on the terminals in each terminal group are same; and for any row of metal sheets in any two adjacent terminal groups, along an arrangement direction of the row of metal sheets, bending pins of the metal sheets in the adjacent terminal groups are alternately arranged on both sides of the row of metal sheets. The bending pins of the row of metal sheets are bent toward two different directions.

Abstract: The present disclosure relates to a radar ranging system based on a true random generator and a ranging method thereof The present disclosure adopts the following technical scheme: the system comprises a radar signal modulation and transmission unit and a radar signal receiving and processing unit, wherein the radar signal modulation and transmission unit inputs and stores a true random binary sequence generated by the true random generator in a memory, and cyclically outputs the true random binary sequence to a BPSK modulator for phase modulation. The receiving and amplifying unit receives the reflected radar signal. The down-sampling unit down samples the intermediate frequency signal generated by the mixers. The data processing unit converts the sampled analog signal into a digital signal and cross-correlates the signal with the true random binary sequence in the memory, so as to obtain the distance to the object side to be measured.

Abstract: The present invention relates to a method of improving soybean transformation efficiency, which comprises: transforming a plant cell using a recombinant vector containing a gene of interest and a gene encoding a sulfonylurea herbicide hydrolase; screening and culturing the transformed plant cell by external application of an ALS inhibitor, using the gene encoding the sulfonylurea herbicide hydrolase as a selective marker; selecting a plant cell that has not been killed and/or not been inhibited.

Abstract: An intelligent control method for engine initiation, including: determining whether a current driving mode satisfies a predetermined driving mode; when the current driving mode satisfies the predetermined driving mode, acquiring a current remaining capacity of a battery; determining whether the current remaining capacity is greater than a predetermined capacity threshold value; when the current remaining capacity is greater than the predetermined capacity threshold value, not starting an engine, otherwise, acquiring a catalyst temperature and a vehicle operating state; determining, according to the vehicle operating state, whether the catalyst temperature is less than a predetermined temperature threshold value; when the catalyst temperature is not less than the predetermined temperature threshold value, driving the engine to rotate clockwise to start same in a fuel cut-off manner, otherwise, heating a catalyst by a heating plate until the catalyst temperature is not less than the predetermined temperature th

Abstract: Embodiments described herein provide a cascade-guided adversarial training method for sequential recommendation models. A system may compute cascade scores for items in a user interaction sequence. The cascade scores may be based on the position in the sequence, as well as the appearance of the same item in other sequences. Based on the computed cascade score, the system may perturb item embeddings. The perturbed user interaction sequences with perturbed item embeddings may be used to train a sequential recommendation model.

Abstract: Embodiments described herein provide a universal item learning framework that generates universal item embeddings for zero-shot items. Specifically, the universal item learning framework performs generic features extraction of items and product knowledge characterization based on a product knowledge graph (PKG) to generate embeddings of input items. A pretrained language model (PLM) may be adopted to extract features from generic item side information, such as titles, descriptions, etc., of an item. A PKG may be constructed to represent recommendation-oriented knowledge, which comprise a plurality of nodes representing items and a plurality of edges connecting nodes represent different relations between items. As those relations in PKG are usually retrieved from user-item interactions, the PKG adapts the universal representation for recommendation with knowledge of user-item interactions.

Abstract: A method for heat preservation of a battery of a vehicle by feeding electric power is provided. This method includes: determining an expected vehicle usage time after charging of the battery of the vehicle has been completed and an expected activation time of a function of heating by feeding electric power for heating the battery; and controlling, when the expected activation time is reached, the vehicle to activate the function of heating by feeding electric power according to a current temperature of the battery and the expected vehicle usage time in order that the current temperature of the battery reaches a safety temperature upper threshold to complete heat preservation for the battery. According to the method in the present disclosure, an influence of low temperature on the battery will be avoided, and a waste of electric energy caused due to continuous heating of the battery may be avoided.

Abstract: The present disclosure provides a method and system for determining a pure electric available power and a vehicle, the method being applied to a vehicle including a fuel cell, the vehicle further including a battery pack and a motor, the fuel cell and the battery pack being electrically connected to the motor, wherein the method includes: monitoring a current travelling state of the vehicle and a current on-off state of the fuel cell; acquiring a maximum peak power outputted by the motor, a maximum output power of the battery pack and a starting-up power of the fuel cell; and according to the current travelling state, the current on-off state of the fuel cell, the maximum peak power, the maximum output power and the starting-up power, determining the pure electric available power.

Abstract: A wheel-leg mechanism is provided. The mechanism comprises a thigh, the upper end of the thigh is movably arranged in a mounting seat for a thigh motor, and is in transmission connection with a thigh motor, and the thigh motor is fixedly provided on one side of the mounting seat for a thigh motor; a shank motor is arranged at the side, away from the thigh motor, of the thigh, a suspension shock absorption system is connected to the shank motor, the shank motor is in transmission connection with a shank by a synchronous belt, the shank is movably connected to the tail end of the thigh, a wheel is movably mounted at the tail end of the shank, and the wheel is in transmission connection with a hub motor; and a braking system is provided on the wheel. A wheel-legged vehicle having the wheel-leg mechanism is further provided.

Abstract: Embodiments of this disclosure include an image obtaining method and apparatus, a server, and a storage medium. In the method, a target application process corresponding to a user identifier is obtained, by processing circuitry, from an application process set. A plurality of window image data that is currently generated is obtained, via a data obtaining module, when an image rendering function in the target application process is called. Image synthesis processing is performed on the plurality of window image data, to obtain a user interface image to be displayed. Further, a notification message that includes the user interface image is transmitted to a user terminal corresponding to the user identifier for display on a user interface.

Abstract: Embodiments described herein provides an intent prototypical contrastive learning framework that leverages intent similarities between users with different behavior sequences. Specifically, user behavior sequences are encoded into a plurality of user interest representations. The user interest representations are clustered into a plurality of clusters based on mutual distances among the user interest representations in a representation space. Intention prototypes are determined based on centroids of the clusters. A set of augmented views for user behavior sequences are created and encoded into a set of view representations. A contrastive loss is determined based on the set of augmented views and the plurality of intention prototypes. Model parameters are updated based at least in part on the contrastive loss.

Abstract: A method for providing a neural network system includes performing contrastive learning to the neural network system to generate a trained neural network system. The performing the contrastive learning includes performing first model augmentation to a first encoder of the neural network system to generate a first embedding of a sample, performing second model augmentation to the first encoder to generate a second embedding of the sample, and optimizing the first encoder using a contrastive loss based on the first embedding and the second embedding. The trained neural network system is provided to perform a task.

Abstract: A wheel-leg mechanism is provided. The mechanism comprises a thigh, the upper end of the thigh is movably arranged in a mounting seat for a thigh motor, and is in transmission connection with a thigh motor, and the thigh motor is fixedly provided on one side of the mounting seat for a thigh motor; a shank motor is arranged at the side, away from the thigh motor, of the thigh, a suspension shock absorption system is connected to the shank motor, the shank motor is in transmission connection with a shank by a synchronous belt, the shank is movably connected to the tail end of the thigh, a wheel is movably mounted at the tail end of the shank, and the wheel is in transmission connection with a hub motor; and a braking system is provided on the wheel. A wheel-legged vehicle having the wheel-leg mechanism is further provided.

Abstract: A code reuse processing method, apparatus and electronic device related to a software development technology in the technical field of computers. The method includes: receiving a target request sent by a client of a target unit; in response to the target request, determining a first code line quantity per capita submitted within a target unit time and a reuse code base of the target unit used by a first code base in a dependency manner, the first code base is a code base except the reuse code base; determining reference information configured to indicate an amount of labor capable of being saved for the target unit when the target unit performs a code reuse within a target duration, the reference information based on the first code line quantity, a code line quantity of the reuse code base and a preset reuse proportion; and outputting the reference information to the client.

Abstract: A yaw motion control method for a four-wheel distributed vehicle includes: calculating the steering response of the vehicle in a steady state using a nonlinear vehicle model in reference with an understeering degree while constraining by the limit value of the road surface adhesion condition according to the sideslip angle response and the vertical load change in the steady state, calculating the lateral force response and the self-aligning moment response of the tires in the steady state by a magic tire formula, calculating the required additional yaw moment by using the yaw motion balance equation, reasonably distributing the generalized control force to the four drive motors through the optimization algorithm in combination with the current driving conditions; finally, off-line storing and retrieving the calculation results of the off-line distribution of different vehicle parameters required by different upper layers to distribute the torques to the four drive wheels.

Abstract: The present disclosure relates to the technical field of beverage mixers, in particular to a portable smart decanter, including a tray, a ball cover, balls, a rotary disc, a cam, a bearing, a motor, a motor cover, a diffusion ring, a USB charging port, a display, a key, a smart control drive, an upper cover, a lower cover, foot pads and a battery cover. The bottom center buckle of the lower cover is used to fix the battery cover. Through structure and function combination of various modules, when an object is placed on the tray of the on-off switch module, the force-bearing spring on the on-off switch module will touch the switch and make the internal circuit connected, and then the motor on the eccentric cam module begins to rotate and drive the ball sliding structure to slide.