Abstract: Systems, methods and media of optimization of temporary read errors (TRE) in three-dimensional (3D) NAND memory devices are disclosed. A disclosed memory device can comprises a plurality of memory cells arranged as an array of NAND memory strings, a plurality of word lines couple to the memory cells, and a controller. The controller is configured to determine whether a next read operation is a first read operation of the memory device after recovering from an idle state, and In response to a positive result of the determination, control the memory device to perform an extended pre-phase of the first read operation before a read-phase of the first read operation.

Abstract: Before a network jitter occurs, a server receives operation instructions from an electronic device, generates a plurality of first rendering instruction streams based on the operation instructions, sends the plurality of first rendering instruction streams to the electronic device, and generates and stores one or more first rendering contexts for one or more first rendering instruction streams in the plurality of first rendering instruction streams. When the network jitter occurs, the server generates a second rendering instruction stream based on received operation instructions, and generates and stores a second rendering context based on the second rendering instruction stream. After the recovery from the network jitter, the server sends a set of rendering context differences to the electronic device. In this way, network transmission traffic after the recovery from the network jitter is reduced, transmission and processing delays are also reduced, and user experience is improved.

Abstract: The cleaning device includes a water injection member and a mopping cloth cleaning member. The water injection member is configured to wet the mopping cloth of a cleaning robot. The mopping cloth cleaning member which reciprocates up and down to flap the mopping cloth of the cleaning robot, in order to clean the mopping cloth of the cleaning robot.

Abstract: The present disclosure provides systems, methods and media of optimization of temporary read errors (TRE) in three-dimensional (3D) NAND memory devices. The (3D) NAND memory devices comprises a plurality of memory cells arranged as an array of NAND memory strings, a plurality of word lines couple to the memory cells, and a controller. The controller is configured to determine whether a next read operation is a first read operation of the memory device after recovering from an idle state, and in response to a positive result of the determination, control the memory device to perform an extended pre-phase of the first read operation before a read-phase of the first read operation.

Abstract: A semiconductor structure includes a stack structure including interleaved conductive layers and dielectric layers, memory strings and slit structures extending through the stack structure, and top select gate cuts extending through the stack structure. At least two of the top select gate cuts are disposed between adjacent slit structures. One of the memory strings includes a memory film and a channel layer. The memory film includes a tunneling layer, a storage layer, and a blocking layer. Each of the top select gate cuts has a width smaller than a diameter of the memory strings.

Abstract: Embodiments of this application provide a synchronization processing method, apparatus, and device, and a medium. The method includes obtaining a skeleton set and a synchronization task set, the skeleton set being used for storing M blocks to be synchronized by the first node device, the skeleton set comprising P skeletons, the synchronization task set comprising P synchronization tasks, the P skeletons being of a one-to-one correspondence to the P synchronization tasks, and M and P both being positive integers; selecting a second node device from the candidate node set, the second node device being a candidate node device in the candidate node set; concurrently transmitting synchronization tasks in the synchronization task set to the second node device, the second node device performing the synchronization tasks and returning the M blocks; and filling the M blocks into the skeletons in the skeleton set.

Abstract: Provided is a compound represented by formula (I) and a pharmaceutical composition thereof, as well as a method for using a compound represented by formula (I) to prevent and/or treat immune-related disorders.

Abstract: This application discloses a method and apparatus for processing data based on a block chain. The method includes adding transaction data in at least two blocks into an execution queue according to a block height sequence and transaction data sequence numbers; executing contract services and acquiring a read cache and a write cache of a parallel transaction generated by executing each contract service, the at least two pieces of transaction data comprising target transaction data; determining a write cache of a merged transaction in block caches, and acquiring a read cache of an updated transaction; determining a write cache of the transaction to be merged of the target transaction data according to the read cache of the updated transaction; and merging the write cache of the transaction to be merged of the target transaction data into a block cache to which the target transaction data belongs.

Abstract: This application discloses a key generation method and apparatus, a device, and a medium. The method includes transmitting a key negotiation request to a consensus node in a blockchain network, to request a key verification parameter; negotiating a collaboration key pair with a second device to obtain a first private key component and a shared public key in the collaboration key pair; signing the key verification parameter by using the first private key component, to obtain a first signature component; generating transaction data according to the first signature component and the key verification parameter; transmitting the transaction data to the second device. The consensus node performs a consensus on the negotiation result data, and after the negotiation result data passes the consensus, submits the negotiation result data to a blockchain of the blockchain network.

Abstract: A housing assembly includes at least two housings, a hinge, a linkage assembly, a locking assembly, a locking key, and a driving assembly. The at least two housings are connected through the hinge, and the at least two housings are able to rotate relative to the hinge. A first end of the linkage assembly is connected to the hinge, and a second end of the linkage assembly extends into the first housing and is movably connected to the first housing. In a case that the locking assembly is in a first state, the locking assembly is separated from the second end; and in a case that the locking assembly is in a second state, the locking assembly abuts against the second end. The driving assembly is connected to the locking key to drive, through the locking key, the locking assembly to switch between the first state and the second state.

Abstract: The present disclosure relates to an integrally formed inductor and a manufacturing method therefor. The method comprises: placing a coil into a mold, wherein the coil is wound in a predetermined shape; bending a lead of the coil onto the mold; and adding soft magnetic metal powder to the mold for pressing, so as to obtain an integrally formed inductor.

Abstract: Embodiments of a three-dimensional (3D) memory device and fabrication methods are disclosed. In some embodiments, the method for forming the 3D memory device includes forming an alternating dielectric stack on a substrate, and forming channel holes that penetrate the alternating dielectric stack and expose at least a portion of the substrate. The method further includes forming top select gate openings that penetrate vertically an upper portion of the alternating dielectric stack and extend laterally. The method also includes forming slit openings parallel to the top select gate openings, wherein the slit openings penetrate vertically the alternating dielectric stack. The method also includes replacing the alternating dielectric stack with a film stack of alternating conductive and dielectric layers, forming top select gate cuts in the top select gate openings, and forming slit structures in the slit openings.

Abstract: A system uses a neural network based model to perform scene text recognition. The system achieves high accuracy of prediction of text from scenes based on a neural network architecture that uses double attention mechanism. The neural network based model includes a convolutional neural network component that outputs a set of visual features and an attention extractor neural network component that determines attention scores based on the visual features. The visual features and the attention scores are combined to generate mixed features that are provided as input to a character recognizer component that determines a second attention score and recognizes the characters based on the second attention score. The system trains the neural network based model by adjusting the neural network parameters to minimize a multi-class gradient harmonizing mechanism (GHM) loss. The multi-class GHM loss varies based on a level of difficulty of the sample.

Abstract: The present application provides a cap assembly and a bottle assembly. The cap assembly is configured to lock a bottle. The bottle includes a bottle head having at least one lock accommodating portion and comprises a bottle body. The cap assembly includes a cap, an elastic arm and at least one lock portion. The cap defines an annular receiving portion having an opening facing downward, and the bottle head can enter the receiving portion through the opening. The elastic arm is annular and is arranged in the receiving portion. The at least one lock portion is arranged on the cap and is configured to be capable of cooperating with the at least one lock accommodating portion of the bottle head, so as to lock the cap assembly to the bottle head.

Abstract: The present application provides a cap assembly and a bottle assembly. The cap assembly includes a cap, an inner cylinder assembly, an elastic member and a blocking member. The cap includes a cap top and a cap side portion. The cap side portion is formed by extending downwardly from an outer edge of the cap top, and the cap top and the cap side portion define a cap cavity. The inner cylinder assembly is connected to the cap top, is arranged in the cap cavity, and forms an annular receiving portion with the cap side portion. The elastic member is arranged in the receiving portion. The blocking member is arranged in the receiving portion below the elastic member. The cap is provided with at least one locking portion for preventing the elastic member and the blocking member from falling from the receiving portion.

Abstract: Aspects concern a method for training a neural network, comprising forming an autoencoder comprising the neural network as encoder and comprising a decoder, for each training image of multiple training images, generating a latent representation of the training image by the encoder, transforming the training image and supplying information about the transformation and at least a part of the latent representation to the decoder to generate a decoder output for the training image and adjusting the encoder and the decoder to reduce a loss between the transformed training images and the decoder outputs.

Abstract: The present invention provides a data loading method. The method comprises: reading data from an initial address in a memory; obtaining a frame header identifier, loading a valid data segment comprising the frame header identifier; verifying the valid data segment; when it is determined that the verification is successful, sending at least part of information within the valid data segment to a cache unit; updating a register according to at least part of the information by the cache unit. Using a specific frame header identifier, loading data in the memory starts only from the specific frame header identifier, which skips the data before the frame header identifier and avoids loading invalid data in the cache unit, thus saving time, preventing lengthy program, and being more scientific and convenient.

Abstract: The present invention discloses a control method, system, electronic device and readable storage medium for a capsule endoscope. The method includes: providing a working apparatus, comprising a capsule endoscope, and an external data recorder for cooperating with and controlling the capsule endoscope; monitoring the received ambient power by the external data recorder before wireless transmission of the capsule endoscope or during an intermittence between two transmissions, and/or monitoring the output power of the capsule endoscope by the external data recorder as data is transmitted during wireless transmission; adjusting the operating state of the working apparatus according to the ambient power and/or output power.

Abstract: Embodiments described herein provide a contrastive learning mechanism with self-labeling refinement, which iteratively employs the network and data themselves to generate more accurate and informative soft labels for contrastive learning. Specifically, the contrastive learning framework includes a self-labeling refinery module to explicitly generate accurate labels, and a momentum mix-up module to increase similarity between a query and its positive, which in turn implicitly improves label accuracy.