Abstract: A cell assembly, a battery module, a battery pack and an energy storage system are provided. The cell assembly includes at least two cell units and at least one temperature equalization portion. At least two cell units are arranged at intervals along the Z direction. The temperature equalization portion is provided between two adjacent cell units along the Z direction. Each of the at least one temperature equalization portion is thermally connected to the two adjacent cell units, to equalize the temperatures of the two adjacent cell units.

Abstract: This application provides a battery module, a battery pack, and an electrical device. The battery module includes at least one battery group and multiple buffer members. The battery group includes multiple batteries arranged along a first direction. Each battery includes multiple positive electrode plates and multiple negative electrode plates alternately stacked along the first direction. In each battery, the positive electrode plate and the negative electrode plate respectively include portions that overlap with each other in the first direction to form an overlapping region, while the positive electrode plate and a remaining portion of the negative electrode plate do not overlap in the first direction, forming a non-overlapping region.

Abstract: This invention focuses on the 3D reconstruction of dynamic crowds in large-scene videos and introduces the DyCrowd framework, which reconstructs 3D position, pose, and shape of hundreds of people from a large-scene video. Our approach addresses frequent occlusions and modeling challenges in high-density crowds through a top-down strategy. This includes pre-reconstruction, matching individual movement sequences, and multi-stage iterative optimization. During the optimization process, we introduce a group optimization method with an asynchronous motion consistency loss. This method clusters individuals with similar trajectories, using high-quality and unoccluded movements within the group to guide the recovery of occluded individuals, thereby mitigating long-term occlusion issues. Furthermore, to address the lack of ground-truth human reconstruction labels in current large-scene datasets, we introduce a virtual benchmark dataset called VirtualCrowd for dynamic crowd reconstruction in large-scene videos.

Abstract: This application provides a software package building method. In the method, a plurality of groups of customized files written in a markup language are obtained. The plurality of groups of customized files are used to configure a to-be-built software package, and each group of customized files includes at least one configuration item. Then, a configuration file is determined based on a fusion customization strategy and the plurality of groups of customized files. The configuration file includes at least one configuration item, and the fusion customization strategy is used to integrate a plurality of configuration items in the plurality of groups of customized files, for example, associate the plurality of configuration items and merge the plurality of configuration items. Then, the software package is built based on the configuration file.

Abstract: The present invention discloses an ejection deployment and retrieval mechanism for a tethered satellite with repeatable unlocking and locking and a working method thereof, the present invention includes a separation ejection docking assembly and a sub-satellite; the separation ejection docking assembly includes an ejection sleeve, a locking mechanism, and a sub-satellite connector, the ejection sleeve adopts a two-stage internal-external sliding connection and provides initial kinetic energy to the sub-satellite by compressing a spring; the locking mechanism is mounted inside the ejection sleeve, one end of the sub-satellite connector is connected to the sub-satellite, and another end of the sub-satellite connector is slidably engaged with the locking mechanism to complete the locking/unlocking operations of the sub-satellite.

Abstract: A custodial token platform may implement smart contracts for inbound transactions and flush transactions. The platform may deploy to a blockchain ledger, a plurality of inbound smart contracts, including first inbound smart contract for a first user of a plurality of users of a custodial token platform. The plurality of smart contracts may include a second inbound smart contract for a second user of the plurality of users of the custodial token platform. The platform may deploy a batch smart contract configured to control the plurality of inbound smart contracts. The platform may broadcast a transaction that calls the batch smart contract. The transaction causes a transfer of the first set of one or more crypto tokens of the first inbound smart contract and the second set of one or more crypto tokens of the second inbound smart contract to an outbound address of the custodial token platform.

Abstract: A method is provided. The method includes: inputting, to a first neural network, a user feature and an object feature of an object to be recommended, and acquiring a first recommendation result output by the first neural network for the object to be recommended; determining a feature that needs to be reinforced; inputting, to a second neural network, the first recommendation result and the feature that needs to be reinforced, and acquiring a second recommendation result output by the second neural network for the object to be recommended; and determining, based on at least the second recommendation result, a final recommendation result for the object to be recommended.

Abstract: A method and apparatus for constructing a drug knowledge graph are provided. The method may include: identifying entities in a drug text; replacing medical key entities among the entities with character strings that conform to a preset rule, to obtain a replaced text; restoring the character strings in a word segmentation result, determined based on the replaced text, to the medical key entities replaced by the character strings; forming a linear entity relationship between the entities based on the entities; and generating a drug knowledge graph according to a parsing result obtained by syntactically parsing the linear entity relationship.

Abstract: A custodial token platform may implement smart contracts for inbound transactions and flush transactions. The platform may deploy to a blockchain ledger, a plurality of inbound smart contracts, including first inbound smart contract for a first user of a plurality of users of a custodial token platform. The plurality of smart contracts may include a second inbound smart contract for a second user of the plurality of users of the custodial token platform. The platform may deploy a batch smart contract configured to control the plurality of inbound smart contracts. The platform may broadcast a transaction that calls the batch smart contract. The transaction causes a transfer of the first set of one or more crypto tokens of the first inbound smart contract and the second set of one or more crypto tokens of the second inbound smart contract to an outbound address of the custodial token platform.

Abstract: A heating assembly, applicable to an aerosol-generating device, includes: a heating rod; a covering film covering part of the heating rod; and a fixing base fitted on the heating rod, a fitting hole being formed on the fixing base. The heating rod includes a functional part with the covering film and a fitting part without the covering film. The fitting part is accommodated in the fitting hole.

Abstract: A heating assembly, used in an aerosol-forming apparatus, includes: a heating rod; a base film at least partially wrapping the heating rod; a heating film arranged between the base film and the heating rod; and two electrode leads. Each electrode lead of the two electrode leads has a conduction portion arranged between the base film and the heating rod and connected to the heating film.

Abstract: A method and apparatus for constructing a drug knowledge graph are provided. The method may include: identifying entities in a drug text; replacing medical key entities among the entities with character strings that conform to a preset rule, to obtain a replaced text; restoring the character strings in a word segmentation result, determined based on the replaced text, to the medical key entities replaced by the character strings; forming a linear entity relationship between the entities based on the entities; and generating a drug knowledge graph according to a parsing result obtained by syntactically parsing the linear entity relationship.

Abstract: A method is provided. The method includes: inputting, to a first neural network, a user feature and an object feature of an object to be recommended, and acquiring a first recommendation result output by the first neural network for the object to be recommended; determining a feature that needs to be reinforced; inputting, to a second neural network, the first recommendation result and the feature that needs to be reinforced, and acquiring a second recommendation result output by the second neural network for the object to be recommended; and determining, based on at least the second recommendation result, a final recommendation result for the object to be recommended.

Abstract: The disclosed computer-implemented method for performing live migrations of software containers may include (i) identifying a request to migrate a software container from a source computing system to a target computing system while a process executes within the software container, (ii) creating a checkpoint of the process in execution (iii) transferring the checkpoint to the target computing system, (iv) updating the checkpoint recurrently by recurrently creating an incremental checkpoint of the process and merging the incremental checkpoint into the checkpoint, (v) predicting, before updating the checkpoint with an iteration of the incremental checkpoint and based on a size of the iteration of the incremental checkpoint, that finalizing a migration of the software container to the target computing system would meet a predetermined time objective, and (vi) finalizing the migration of the software container to the target computing system.

Abstract: The disclosed computer-implemented method for performing live migrations of software containers may include (i) identifying a request to migrate a software container from a source computing system to a target computing system while a process executes within the software container, (ii) creating a checkpoint of the process in execution (iii) transferring the checkpoint to the target computing system, (iv) updating the checkpoint recurrently by recurrently creating an incremental checkpoint of the process and merging the incremental checkpoint into the checkpoint, (v) predicting, before updating the checkpoint with an iteration of the incremental checkpoint and based on a size of the iteration of the incremental checkpoint, that finalizing a migration of the software container to the target computing system would meet a predetermined time objective, and (vi) finalizing the migration of the software container to the target computing system.

Abstract: The disclosed computer-implemented method for performing live migrations of software containers may include (i) identifying a request to migrate a software container from a source computing system to a target computing system while a process executes within the software container, (ii) creating a checkpoint of the process in execution (iii) transferring the checkpoint to the target computing system, (iv) updating the checkpoint recurrently by recurrently creating an incremental checkpoint of the process and merging the incremental checkpoint into the checkpoint, (v) predicting, before updating the checkpoint with an iteration of the incremental checkpoint and based on a size of the iteration of the incremental checkpoint, that finalizing a migration of the software container to the target computing system would meet a predetermined time objective, and (vi) finalizing the migration of the software container to the target computing system.

Abstract: The disclosed computer-implemented method for performing live migrations of software containers may include (i) identifying a request to migrate a software container from a source computing system to a target computing system while a process executes within the software container, (ii) creating a checkpoint of the process in execution (iii) transferring the checkpoint to the target computing system, (iv) updating the checkpoint recurrently by recurrently creating an incremental checkpoint of the process and merging the incremental checkpoint into the checkpoint, (v) predicting, before updating the checkpoint with an iteration of the incremental checkpoint and based on a size of the iteration of the incremental checkpoint, that finalizing a migration of the software container to the target computing system would meet a predetermined time objective, and (vi) finalizing the migration of the software container to the target computing system.

Abstract: The disclosed computer-implemented method for performing live migrations of software containers may include (i) identifying a request to migrate a software container from a source computing system to a target computing system while a process executes within the software container, (ii) creating a checkpoint of the process in execution (iii) transferring the checkpoint to the target computing system, (iv) updating the checkpoint recurrently by recurrently creating an incremental checkpoint of the process and merging the incremental checkpoint into the checkpoint, (v) predicting, before updating the checkpoint with an iteration of the incremental checkpoint and based on a size of the iteration of the incremental checkpoint, that finalizing a migration of the software container to the target computing system would meet a predetermined time objective, and (vi) finalizing the migration of the software container to the target computing system.

Abstract: The disclosed computer-implemented method for performing live migrations of software containers may include (i) identifying a request to migrate a software container from a source computing system to a target computing system while a process executes within the software container, (ii) creating a checkpoint of the process in execution (iii) transferring the checkpoint to the target computing system, (iv) updating the checkpoint recurrently by recurrently creating an incremental checkpoint of the process and merging the incremental checkpoint into the checkpoint, (v) predicting, before updating the checkpoint with an iteration of the incremental checkpoint and based on a size of the iteration of the incremental checkpoint, that finalizing a migration of the software container to the target computing system would meet a predetermined time objective, and (vi) finalizing the migration of the software container to the target computing system.

Abstract: The present invention relates to a foldable table mechanism that is convenient to transport. The foldable table mechanism includes table tops and a foldable mechanism. The foldable mechanism includes at least one shelve unit and a plurality of support legs. The support legs include a plurality of movable units. The movable units are pivotably connected. The shelve unit is disposed between the table tops. The shelve unit is movably connected to the movable units. The support legs support and secure the table tops at an open position and are stored between the table tops at a folded position. The present invention is simple in structure, foldable and flexible to use, convenient to carry and transport, thus suitable to use in small space, such as in household setting, small food service areas, and also suitable to use for outdoor activities because it is easy to transport.