Abstract: Methods and apparatuses for automatic speech recognition are provided. The method includes: generating a weight matrix for a layer of a plurality of layers in a neural network; dividing the weight matrix into a plurality of blocks, each block including a plurality of weights; selecting a set of blocks from the plurality of blocks for block-wise pruning by minimizing a cost function subject to a pre-determined block-wise constraint; and generating a block-wise pruned weight matrix by setting one or more weights in the set of blocks to zero. The weight matrix includes a set of weights associated with the layer, the plurality of layers includes a first layer receiving a first input associated with one or more audio feature sequences, and the plurality of layers are executed on one or more processors.

Abstract: Example methods and apparatus for managing a storage system are described. The storage system includes a plurality of control nodes. One example method includes receiving a first operation request of a client by a first control node in the plurality of control nodes. The first control node creates a first object on a first storage node based on the first operation request.

Abstract: The present application provides a composite coating agent for a positive electrode material, including a first coating agent, a second coating agent and a third coating agent. The first coating agent is a hydroxide, an oxide, a sulfide, a nitrate or a carbonate of a first coating element, the second coating agent is a hydroxide, an oxide, a sulfide, a nitrate or a carbonate of a second coating element, and the third coating agent is a hydroxide, an oxide, a sulfide, a nitrate or a carbonate of a third coating element. In the composite coating agent disclosed by the present application, different elements are compounded on the surface of the positive electrode material so as to reduce excessive precipitation of Li in the positive electrode material, lower the formation of residual alkali, mitigate the generation of micro-cracks on the surface of the positive electrode material.

Abstract: A high-nickel ternary positive electrode material, has a structural formula of LixNiaMbCocNzO2. A content of a cobalt element within a range of 1.5 ?m from a surface of the positive electrode material accounts for 48%-85% of a content of the cobalt element in the positive electrode material. A high-nickel ternary positive electrode material is designed in the present application, which has a specific structure with cobalt element distributed gradiently from inside to outside, and thus a high-nickel low-cobalt positive electrode material product with good stability can be directly obtained. The content of cobalt in the near-surface region and surface region of the positive electrode material is high, and high-content cobalt may react with Li+that cannot be re-inserted to generate lithium cobaltate. This increases part of the capacity and avoids the generation of residual alkali, thereby improving the capacity and cycle performance of the high-nickel ternary positive electrode material.

Abstract: A construction inspection robotic system and method thereof, it may autonomously implement the inspection for construction finish quality, generate a visual report, which is time-saving, labor-saving, and greatly saving the inspection cost, implement a comprehensive inspection for the construction surface problems, and ensure 100% coverage for the construction; it conforms to the same standards for the construction inspection and ensures the consistency of inspection; it instantaneously records and uploads the original inspection data, instantaneously stores and uploads the original inspection data, thus providing original data reference for disputes possibly caused by inspection results.

Abstract: A method and an apparatus for automatic speech recognition are provided. The method includes: generating a weight matrix for a layer of a plurality of layers in a neural network; dividing the weight matrix into a plurality of blocks, each block including a plurality of weights; selecting a pre-determined percentage of weights from at least one block for block-wise pruning; and generating a block-wise pruned weight matrix by setting the pre-determined percentage of weights selected from the at least one block to zero. The weight matrix includes a set of weights associated with the layer, the plurality of layers includes a first layer receiving a first input associated with one or more audio feature sequences, and the plurality of layers are executed on one or more processors. The method efficiently accelerates model inference using irregular pruning.

Abstract: Example methods and apparatus for managing a storage system are described. The storage system includes a plurality of control nodes. One example method includes receiving a first operation request of a client by a first control node in the plurality of control nodes. The first control node creates a first object on a first storage node based on the first operation request.

Abstract: Disclosed herein are uses for a compound of Formula (I), below, or a pharmaceutically acceptable salt thereof: wherein n is an integer selected from the group of 0, 1, 2, 3, and 4, or a pharmaceutically acceptable salt thereof, for enhancing the effect of an anti-cancer agent, such as an inhibitor of CREB-mediated gene transcription.

Abstract: Methods and apparatuses for automatic speech recognition are provided. The method includes: generating a weight matrix for a layer of a plurality of layers in a neural network; dividing the weight matrix into a plurality of blocks, each block including a plurality of weights; selecting a set of blocks from the plurality of blocks for block-wise pruning by minimizing a cost function subject to a pre-determined block-wise constraint; and generating a block-wise pruned weight matrix by setting one or more weights in the set of blocks to zero. The weight matrix includes a set of weights associated with the layer, the plurality of layers includes a first layer receiving a first input associated with one or more audio feature sequences, and the plurality of layers are executed on one or more processors.

Abstract: A method and an apparatus for automatic speech recognition are provided. The method includes: generating a weight matrix for a layer of a plurality of layers in a neural network; dividing the weight matrix into a plurality of blocks, each block including a plurality of weights; selecting a pre-determined percentage of weights from at least one block for block-wise pruning; and generating a block-wise pruned weight matrix by setting the pre-determined percentage of weights selected from the at least one block to zero. The weight matrix includes a set of weights associated with the layer, the plurality of layers includes a first layer receiving a first input associated with one or more audio feature sequences, and the plurality of layers are executed on one or more processors.

Abstract: A construction inspection robotic system and method thereof, it may autonomously implement the inspection for construction finish quality, generate a visual report, which is time-saving, labor-saving, and greatly saving the inspection cost, implement a comprehensive inspection for the construction surface problems, and ensure 100% coverage for the construction.; it conforms to the same standards for the construction inspection and ensures the consistency of inspection; it instantaneously records and uploads the original inspection data, instantaneously stores and uploads the original inspection data, thus providing original data reference for disputes possibly caused by inspection results.

Abstract: A simulation method for flow field of multi-stage fracturing on horizontal well in tight oil reservoir is provided.

Abstract: A prepolymer, which is prepared by subjecting an unsaturated bond-containing compound and a bis(vinylphenyl) compound monomer or a polymer thereof to a prepolymerization reaction, is provided. The bis(vinylphenyl) compound contains between 80% and 99% of para-para vinyl groups and has a monomer content of between 80% and 100%. Moreover, a resin composition comprising the prepolymer and an additive, a method of preparing the prepolymer, and an article made from the resin composition are also provided.

Abstract: The present disclosure discloses a method for safely preparing pimavanserin and tartrate thereof by using triphosgene. The synthesis route includes: Raw materials used in the method of the present disclosure are safe and inexpensive, thereby effectively reducing the production costs. The method of the present disclosure has mild reaction conditions, and avoids using phosgene, which is highly toxic and difficult to handle is avoided. Thus, the method is readily implemented industrially.

Abstract: A simulation method for flow field of multi-stage fracturing on horizontal well in tight oil reservoir is provided.

Abstract: The present disclosure discloses a method for safely preparing pimavanserin and tartrate thereof by using triphosgene. The synthesis route includes: Raw materials used in the method of the present disclosure are safe and inexpensive, thereby effectively reducing the production costs. The method of the present disclosure has mild reaction conditions, and avoids using phosgene, which is highly toxic and difficult to handle is avoided. Thus, the method is readily implemented industrially.