Abstract: Embodiments of the present disclosure provide interaction methods and devices. The method can include obtaining a display object comprising a plurality of collage pictures for display; acquiring, in response to a switch operation for at least one collage picture of the plurality of collage pictures of the display object, a first target collage picture, wherein a first target content type of the first target collage picture matches a content type of the at least one collage picture; and replacing the at least one collage picture with the first target collage picture.

Abstract: Methods and devices, the method including receiving a matrix of a neural network model; classifying at least a portion of the matrix as a first section based on a first distribution pattern of non-zero elements of the portion of the matrix; and identifying memory addresses of the non-zero elements in the first section of the matrix for loading, according to a first order determined based on the first distribution pattern, the non-zero elements in the first section into one or more vector registers.

Abstract: Embodiments of the present disclosure present a hyper-square interconnect topology and advanced ring-based AllReduce operations. In some embodiments, a topology is provided that is an improvement over conventional interconnect topologies by eliminating delays associated with long wirings. In some embodiments, computing nodes are divided into sub-sections to better allocate computing tasks, and the system can be optimized to divide up the computing nodes by maximizing the number of square sub-sections in the topology. In some embodiments, the system can be optimized to select square sub-sections first for each computing task. Each sub-section can comprise some computing nodes or all computing nodes in the hyper-square interconnect topology. This flexibility allows the hyper-square interconnect topology to utilize the computing nodes more efficiently by assigning appropriate numbers of computing nodes to each computing task based on the computing need of the computing task.

Abstract: Methods and systems are provided for sequence-to-sequence conversion from unstructured search queries to structured database queries, so that lay persons may retrieve information from relational databases without specialized knowledge of database query languages. An encoder module and a decoder module of a learning model are trained to convert an unstructured search query to an intermediate feature vector by computing co-attention and self-attention based on a context string and a database schema, encoding the database schema in the context string by application of self-attention between the context string containing tokens of the database schema with learned structural attention heads which relate the token to logic of the database. Training is performed using labeled training datasets which include structured database queries which are normalized by parsing into a semantic representation thereof, followed by linearization.

Abstract: Methods and apparatus for reducing a size of a neural network model, the method including: compressing data of the neural network model; identifying structure information of a vector register, wherein the structure information includes a number of registers included in the vector register; comparing a number of elements in the compressed data with a first condition, wherein the first condition is determined based on the number of registers in the vector register; and in response to the number of elements satisfying the first condition, associating the compressed data with the vector register to enable loading the compressed data to the vector register.

Abstract: Methods, apparatuses, and devices for Josephson junction preparation includes: obtaining a first pattern structure for generating a first Josephson junction of a first type and a plurality of second pattern structures for generating a plurality of second Josephson junctions of a second type; evaporating a material on the first pattern structure and the plurality of second pattern structures based on a first evaporation direction to generate a first electrode layer for implementing information transmission; forming an insulating layer on the first electrode layer, the insulating layer including a compound corresponding to the material; evaporating the material on the first pattern structure and the plurality of second pattern structures based on a second evaporation direction to generate a second electrode layer for implementing information transmission; and forming the first Josephson junction and the plurality of second Josephson junctions.

Abstract: Communication methods and devices include opening a plurality of detection windows according to a preset detection period; in response to having detected a preamble of a radio frame in at least a preset number of continuous detection windows of the plurality of detection windows, switching the receiving end device to an awake state; and receiving the radio frame while the receiving end device is operating in the awake state.

Abstract: A superconducting circuit having a Josephson junction includes a first electrode layer for signal transmission; a second electrode layer for signal transmission; and an insulating layer arranged between the first electrode layer and the second electrode layer to form a Josephson junction, wherein, the first electrode layer and the second electrode layer are composed of a preset material, the insulating layer is composed of a compound corresponding to the preset material, and the preset material includes a non-aluminum superconducting material to prolong a coherence time of superconducting qubits.

Abstract: An image recognition method includes: obtaining an image; extracting a target image region corresponding to a target part from the image, wherein the target image region includes a target object; determining a location of the target object in the target image region (i) according to pixel values of pixels in the target image region and a location relationship between the pixels, or (ii) inputting the target image region to a trained segmentation model to obtain the location of the target object in the target image region; and displaying a recognition result of the image, wherein the recognition result indicates the location of the target object in the target image region.

Abstract: Virtualization processing systems, methods, apparatus, and electronic devices are described herein. The systems have a virtualization infrastructure deployed on a virtual machine management and control (M&C) board side and used for constructing a virtualization system, and a management and control virtual machine deployed on a host machine side and used for managing and controlling the user virtual machine by using host machine resources. The system uses a management and control manner based on a virtual node, such that management and control are encapsulated inside a management and control virtual machine, which can be deployed on a host machine side and can also be deployed on a M&C board side. When resources of a management and control board are insufficient, one or more virtual management and control nodes can be dynamically started on the host machine side, and the node can share some management and control tasks using host machine resources.

Abstract: Embodiments of the present disclosure provide methods and systems for processing a speech signal. The method can include: processing the speech signal to generate a plurality of speech frames; generating a first number of acoustic features based on the plurality of speech frames using a frame shift at a given frequency; and generating a second number of posteriori probability vectors based on the first number of acoustic features using an acoustic model, wherein each of the posteriori probability vectors comprises probabilities of the acoustic features corresponding to a plurality of modeling units, respectively.

Abstract: A processing apparatus, an embedded system, a system-on-chip, and a security control method are disclosed. The processing apparatus includes a processor, adapted to execute a program; and a memory, coupled to the processor and adapted to provide a plurality of enclaves isolated from each other. One of the plurality of enclaves is a source enclave, another one of the plurality of enclaves is a target enclave, and the source enclave and the target enclave each are used to provide a storage space required for running a corresponding program. The processing apparatus further comprises a storage access controller, adapted to transmit specified data stored in the source enclave to the target enclave.

Abstract: Embodiments of the present disclosure provides systems and methods for processing video content. The methods include: reconstructing a plurality of luma samples associated with a picture; and reconstructing a chroma block associated with the picture. The reconstructing of the chroma block includes: determining whether the chroma block has non-zero residues; and in response to a determination that the chroma block has one or more non-zero chroma residues, determining an averaged value of the plurality of reconstructed luma samples, and scaling residues of the chroma block based on the averaged value, prior to reconstructing the chroma block.

Abstract: Apparatus and method provided herein are directed to a linear differential directional microphone array (LDDMA), which takes into account the directionality of the array elements. The LDDMA may be designed by generating a steering vector for a linear array (LA) having preselected parameters including parameters ?, p, ?, N, and M, generating a constraint matrix based on the steering vector, reformulating the constraint matrix based on a microphone response matrix and a steering matrix, obtaining a beamformer by applying a minimum norm solution in terms of the constraint matrix, verifying a desired characteristic of the LA by calculating the beamformer for a desired direction, and constructing the LA based on the preselected parameters and the beamformer.

Abstract: Systems and methods for dynamically executing sparse neural networks are provided. In one implementation, a system for providing dynamic sparsity in a neural network may include at least one memory storing instructions and at least one processor configured to execute the instructions to: reduce an input vector and a set of weights of the neural network, execute an input layer of the neural network using the reduced input vector and set of weights to generate a reduced output vector; expand the reduced output vector to a full output vector using first predictable output neurons (PONs); using a PON map, reduce a dimension of the full output vector; execute subsequent layers of the neural network using the reduced full output vector to produce a second reduced output vector; and expand the second reduced output vector to a second full output vector using second PONs.

Abstract: The present disclosure provides methods for convolutional-neural-network (CNN) based filter for video coding. An exemplary method includes: applying motion estimation to a target coding block, to determine a reference block of the target coding block; inputting, to a convolutional neural network (CNN) filter, image data associated with the target coding block and the reference block; and executing the CNN filter to determine a residual associated with the target coding block based on the input image data.

Abstract: A method includes: obtaining video data; receiving a target object; processing the video data to obtain one or more first images of the video data that contain an image of the target object and one or more second images of the video data that do not contain the image of the target object; discarding the one or more second images from the video data; generating a background image based on the one or more second images; synthesizing the one or more first images of the video data with the background image to generate one or more synthesized images; obtaining a target image sequence with the one or more synthesized images based on a time order of the one or more first images in the video data; and displaying each of the one or more synthesized images in the target image sequence in a fade-in and fade-out manner.

Abstract: The example embodiments provide a method, a system, a mobile device, and an acceleration device for processing computing jobs. The method includes: obtaining, by a mobile device, a computing job, wherein a first interface of the mobile device is connected to a second interface, the second interface included in an acceleration device; transmitting, by the mobile device, the computing job from the first interface to the second interface via a write command; receiving, by the acceleration device, the computing job at the second interface; processing, by the acceleration device, the computing job and transmitting a processing result from the second interface to the first interface; and obtaining, by the mobile device, the processing result from the first interface via a read command.

Abstract: A method including searching, on the basis of a voiceprint feature of a speaker, for an identifier of the speaker in a speaker registry, the voiceprint feature of the speaker being a parameter obtained according to a voice signal of the speaker captured by a microphone array; if position information corresponding to the identifier of the speaker in the speaker registry is different from position information of the speaker, updating the speaker registry, the position information of the speaker being a parameter obtained according to the voice signal of the speaker captured by the microphone array; and labeling the voice signal of the speaker with the identifier of the speaker, so as to track the speaker. The present disclosure enables voice tracking of multiple persons.

Abstract: The present disclosure provides systems and methods for processing video content using motion compensation interpolation. The methods include: in response to a target picture and a reference picture having different resolutions, applying a band-pass filter to the reference picture, to perform a motion compensated interpolation with reference down-sampling to generate a reference block; and encoding or decoding a block of the target picture using the reference block.