Abstract: A hierarchical audio-visual feature fusing method for audio-visual question answering and a product relate to the field of audio-visual question answering. By fusing audio embedding in an input video clip with a baseline model as well as video embedding and question embedding respectively at an early stage, a middle stage and a late stage in a hierarchical feature fusing process, a first answer probability distribution, a second answer probability distribution and a third answer probability distribution are obtained, and the answer probability distributions are added based on preset weights, and then averaged for hierarchical integration to generate a final answer.

Abstract: A modularized parametric visual program induction algorithm, a device, a medium and a product are provided. An optimized modularized parametric model is obtained by constructing a modularized parametric model and training with the cooperation of a hierarchical Monto-Carlo Tree-Search algorithm, data to be processed is input into the optimized modularized parametric model and processed in conjunction with the hierarchical Monto-Carlo Tree-Search algorithm, and a result program expression is output. The modularized parametric visual program induction algorithm is applicable to complex visual scenes; and model training data of the modularized parametric model is augmented in a training stage in conjunction with the hierarchical Monto-Carlo Tree-Search algorithm.

Abstract: An architecture search method and an architecture search apparatus for a large-scale graph, and a device and a storage medium are provided. The method includes: obtaining a subgraph of a large-scale graph by performing local sampling on the large-scale graph; sampling a plurality of neural network architectures in a pre-constructed super network according to a pre-customized importance sampling strategy; obtaining a plurality of trained neural network architectures by training, according to a peer learning method, the plurality of neural network architectures with the subgraph; obtaining a trained super network by iteratively executing the subgraph sampling, the architecture sampling, and the architecture training; and selecting an optimal architecture in the trained super network to process the large-scale graph to obtain a graph processing result.

Abstract: Methods and apparatus for pillar bending improvement by cut tiers pattern implementation. The method uses a cut tier pattern in a staircase region of a 3D memory structure to reduce pillar bending in a pillar array region. The pillar array region includes a plurality of memory tiers comprising wordline layers interposed between isolation layers, where a memory tier comprises a two-dimensional (2D) array of memory cells. A plurality of vertical structures comprising pillars pass through memory cells in the wordline layers and pass through the isolation layers. The staircase structure is disposed adjacent to the pillar array region and includes vertical wordline drivers coupled to the wordline layers. A cut tier pattern comprising vertical trenches is formed in the staircase structure toward a side of the staircase structure adjacent to the pillar array region. The cut tier pattern includes one or more breaks used for routing circuitry in the wordlines.

Abstract: A method and apparatus for learning graph representations for out-of-distribution generalization.

Abstract: A disentangled commodity recommendation method includes: receiving information of commodities to be recommended and historical behavioral information of a user, the historical behavioral information including a clicked commodity sequence, an unclicked commodity sequence, a disliked commodity sequence and behavioral time information of the user; filtering the clicked commodity sequence and the unclicked commodity sequence according to the disliked commodity sequence to obtain representations of interested commodities of the user; filtering the representations of the interested commodities according to the behavioral time information of the user and the information of the commodities to be recommended to obtain representations of historical interested commodities of the user; clustering and aggregating the representations of the historical interested commodities to obtain a plurality of disentangled representations of the user; and determining whether the commodities to be recommended are the interested commodi

Abstract: A network convergence method and device, an electronic apparatus, a storage medium are provided. The method includes: taking online information obtained at least from network data as first nodes, and constructing a first network with the first nodes based on an association among the online information, the online information including at least one account and at least one piece of network media information; taking offline information obtained at least from mobile communication data as second nodes, and constructing a second network with the second nodes based on an association among the offline information, the offline information including at least one terminal identifier and location information of a terminal corresponding to the at least one terminal identifier; and converging the first network and the second network by using an association between the account in the online information and the terminal identifier in the offline information, to construct a heterogeneous network.

Abstract: A network convergence method and device, an electronic apparatus, a storage medium are provided. The method includes: taking online information obtained at least from network data as first nodes, and constructing a first network with the first nodes based on an association among the online information, the online information including at least one account and at least one piece of network media information; taking offline information obtained at least from mobile communication data as second nodes, and constructing a second network with the second nodes based on an association among the offline information, the offline information including at least one terminal identifier and location information of a terminal corresponding to the at least one terminal identifier; and converging the first network and the second network by using an association between the account in the online information and the terminal identifier in the offline information, to construct a heterogeneous network.

Abstract: Techniques for configuring and operating a multimedia aware cloud, particularly configured for mobile device computing, are described herein. In some instances, clusters of servers are organized for general computing, graphic computing and data storage. A load balancing server may be configured to: identify multimedia types currently being processed within the multimedia edge cloud; determine desired quality of service levels for each identified multimedia type; evaluate individual abilities of devices communicating with the multimedia edge cloud; and assess bandwidth of each network over which the multimedia edge cloud communicates with a mobile device. With that information, multimedia data may be adapted accordingly, to result in an acceptable quality of service level when delivered to a specific mobile device. In one example of the techniques, graphic computing server clusters may be configured to process workload using a configuration that includes elements of both parallel and serial computing.

Abstract: Techniques for configuring and operating a multimedia aware cloud, particularly configured for mobile device computing, are described herein. In some instances, clusters of servers are organized for general computing, graphic computing and data storage. A load balancing server may be configured to: identify multimedia types currently being processed within the multimedia edge cloud; determine desired quality of service levels for each identified multimedia type; evaluate individual abilities of devices communicating with the multimedia edge cloud; and assess bandwidth of each network over which the multimedia edge cloud communicates with a mobile device. With that information, multimedia data may be adapted accordingly, to result in an acceptable quality of service level when delivered to a specific mobile device. In one example of the techniques, graphic computing server clusters may be configured to process workload using a configuration that includes elements of both parallel and serial computing.

Abstract: Techniques for configuring and operating a multimedia aware cloud, particularly configured for mobile device computing, are described herein. In some instances, clusters of servers are organized for general computing, graphic computing and data storage. A load balancing server may be configured to: identify multimedia types currently being processed within the multimedia edge cloud; determine desired quality of service levels for each identified multimedia type; evaluate individual abilities of devices communicating with the multimedia edge cloud; and assess bandwidth of each network over which the multimedia edge cloud communicates with a mobile device. With that information, multimedia data may be adapted accordingly, to result in an acceptable quality of service level when delivered to a specific mobile device. In one example of the techniques, graphic computing server clusters may be configured to process workload using a configuration that includes elements of both parallel and serial computing.

Abstract: Predictive, multi-layer caching architectures may be used to predict which elements a user is most likely to navigate to within a collection of elements associated with a predefined layout and, in response, to increase the accessibility of these elements to a client device of the user. For instance, the techniques may utilize a predictive, multi-layer caching architecture for storing these predicted elements to decrease the latency to render these images if the user navigates within the collection of elements in the predicted manner. The collection of elements may comprise images (e.g., a 3D model, a map, etc.), video files, audio files, text files, or any other type of file that is consumable on a client device.

Abstract: A depth construction module is described that receives depth images provided by two or more depth capture units. Each depth capture unit generates its depth image using a structured light technique, that is, by projecting a pattern onto an object and receiving a captured image in response thereto. The depth construction module then identifies at least one deficient portion in at least one depth image that has been received, which may be attributed to overlapping projected patterns that impinge the object. The depth construction module then uses a multi-view reconstruction technique, such as a plane sweeping technique, to supply depth information for the deficient portion. In another mode, a multi-view reconstruction technique can be used to produce an entire depth scene based on captured images received from the depth capture units, that is, without first identifying deficient portions in the depth images.

Abstract: Techniques for configuring and operating a multimedia aware cloud, particularly configured for mobile device computing, are described herein. In some instances, clusters of servers are organized for general computing, graphic computing and data storage. A load balancing server may be configured to: identify multimedia types currently being processed within the multimedia edge cloud; determine desired quality of service levels for each identified multimedia type; evaluate individual abilities of devices communicating with the multimedia edge cloud; and assess bandwidth of each network over which the multimedia edge cloud communicates with a mobile device. With that information, multimedia data may be adapted accordingly, to result in an acceptable quality of service level when delivered to a specific mobile device. In one example of the techniques, graphic computing server clusters may be configured to process workload using a configuration that includes elements of both parallel and serial computing.

Abstract: Techniques for configuring and operating a multimedia aware cloud, particularly configured for mobile device computing, are described herein. In some instances, clusters of servers are organized for general computing, graphic computing and data storage. A load balancing server may be configured to: identify multimedia types currently being processed within the multimedia edge cloud; determine desired quality of service levels for each identified multimedia type; evaluate individual abilities of devices communicating with the multimedia edge cloud; and assess bandwidth of each network over which the multimedia edge cloud communicates with a mobile device. With that information, multimedia data may be adapted accordingly, to result in an acceptable quality of service level when delivered to a specific mobile device. In one example of the techniques, graphic computing server clusters may be configured to process workload using a configuration that includes elements of both parallel and serial computing.

Abstract: Predictive, multi-layer caching architectures may be used to predict which elements a user is most likely to navigate to within a collection of elements associated with a predefined layout and, in response, to increase the accessibility of these elements to a client device of the user. For instance, the techniques may utilize a predictive, multi-layer caching architecture for storing these predicted elements to decrease the latency to render these images if the user navigates within the collection of elements in the predicted manner. The collection of elements may comprise images (e.g., a 3D model, a map, etc.), video files, audio files, text files, or any other type of file that is consumable on a client device.

Abstract: This document describes techniques for creating three-dimensional models of image collections using a network-accessible (e.g., cloud-based) computation service, rather than a client computing device. In some instances described below, the computation service performs at least a portion of the operations for creating these models in parallel.

Abstract: Predictive, multi-layer caching architectures may be used to predict which elements a user is most likely to navigate to within a collection of elements associated with a predefined layout and, in response, to increase the accessibility of these elements to a client device of the user. For instance, the techniques may utilize a predictive, multi-layer caching architecture for storing these predicted elements to decrease the latency to render these images if the user navigates within the collection of elements in the predicted manner. The collection of elements may comprise images (e.g., a 3D model, a map, etc.), video files, audio files, text files, or any other type of file that is consumable on a client device.

Abstract: A method for recommending multimedia information includes obtaining an instant user-multimedia connection set; determining relevance indexes between the users and the multimedia information according to the obtained instant user-multimedia connection set; recommending the multimedia information to the users related to the multimedia information according to the determined relevance index. The invention improves an accuracy of multimedia information recommendation. The invention further provides a system for recommending multimedia information.

Abstract: Embodiments of quasi-orthogonal space-time encoder, decoder and method for space-time encoding and decoding orthogonal frequency division multiplexed (OFDM) signals in a multiple-input multiple-output (MDMO) system are generally described herein. Other embodiments may be described and claimed. In some embodiments, a method of decoding received multicarrier signals comprises deriving objective functions from a received signal matrix, minimizing the objective functions to generate decoded hard bits, and solving a set of linear equations of the objective functions to generate decoded soft bits.