Abstract: Some embodiments provide a method of implementing service rules for a container cluster that is configured by a first SDN controller cluster. The method registers for event notification from an application programming interface (API) server to receive notification regarding events associated with resources deployed in the container cluster. The method forwards to a second SDN controller cluster resource identifiers collected through the registration for resources of the container cluster. The second SDN controller cluster defines service policies that are not defined by the first SDN controller cluster. The method receives, from the second SDN controller cluster, service policies defined by the second SDN controller cluster based on the resource identifiers. The method distributes service rules defined based on the service policies to network elements in the container cluster to enforce on data messages associated with machines deployed in the container cluster configured by the first SDN controller cluster.

Abstract: There are provided method, apparatus, device, and medium for determining update gradient for contrastive learning model. In the method, a gradient factor of a first type for the contrastive learning model is determined based on a first group of training data and a second group of training data for training the contrastive learning model. The gradient factor of the first type is not used for backpropagation during a training process. In a first stage of the training process, a gradient factor of a second type associated with the first group of training data is determined based on the contrastive learning model. The gradient factor of the second type is used for backpropagation during the training process. Gradient is obtained for updating the contrastive learning model based on the gradient factor of the first type and the gradient factor of the second type associated with the first group of training data.

Abstract: Provided are an indoor distribution system and a signal transmission method. The system includes an access unit, an extended unit, a first coupling unit, and at least one remote cascaded chain. Each remote cascaded chain includes at least one second coupling unit and at least one remote unit. The at least one second coupling unit in each remote cascaded chain is cascaded, and each remote unit is respectively connected to a corresponding second coupling unit. The first coupling unit transmits a radio-frequency signal transmitted by the extended unit to the second coupling unit. The second coupling unit transmits a radio-frequency signal of input power to the remote unit connected to the second coupling unit according to a coupling value. The remote unit adjusts the power of the radio-frequency signal from the input power to the output target power and output the adjusted radio-frequency signal.

Abstract: The present disclosure provides a method comprising obtaining gas data of a gas user continuously, and storing the gas data into a storage unit; predicting a probability of full storage of the storage unit within a preset future time period based on a historical data increment of the storage unit; in response to the probability of full storage meeting a preset probability condition: obtaining the gas data of the gas user within the preset time period from the storage unit; determining an upload demand degree of the gas data based on the gas data; and in response to the upload demand degree meeting a preset condition, transmitting the gas data within the preset time period to a smart gas management platform based on a smart gas object platform at a target upload time point of a smart gas meter of the gas user.

Abstract: A method of training and applying contrastive learning model. The method includes obtaining a sample set and label information for training contrastive learning model, the sample set including a plurality of first samples of a first modality and a plurality of second samples of a second modality, the label information indicating a correlation between samples of the plurality of first samples and samples of the plurality of second samples; determining whether sample mixing is to be performed on the first modality or the second modality; in accordance with a determination that sample mixing is to be performed on the first modality, generating at least one first mixed sample of the first modality by mixing at least one pair of first samples among the plurality of first samples; and training the contrastive learning model at least based on the at least one first mixed sample and first mixed label information.

Abstract: A method of contrastive learning comprises: determining, based on a model construction criterion, a first encoder for a first modality and a second encoder for a second modality; constructing a first contrastive learning model, the first contrastive learning model comprising the first encoder and a third encoder for the second modality, and a model capacity of the third encoder being greater than a model capacity of the second encoder; performing pre-training of the first contrastive learning model based on a first training dataset for the first modality and the second modality; and providing the pre-trained first encoder in the pre-trained first contrastive learning model for a downstream task. Because only the model capacity of one encoder is increased in the pre-training stage, model performance may be improved without increasing model training overhead during downstream task fine-tuning and model running overhead during model application.

Abstract: Methods, apparatuses, a device, and a medium for training a contrastive learning model are provided. In a method, a plurality of sample sets for training the contrastive learning model are obtained, and the plurality of sample sets comprises a first sample set and a second sample set. A first target sample set is selected from the first sample set and the second sample set according to a predetermined rule. A first set of samples are determined based on the first target sample set according to a predefined batch size. The contrastive learning model is trained using the first set of samples. In this way, on the one hand, performance degradation of the contrastive learning model due to sample set bias may be avoided; on the other hand, a forgetting problem in the training process may be alleviated.

Abstract: Methods, systems, and computer-readable storage media for receiving user input representative of user selection of a set of decision elements to be included in a workflow task UI and a data object, each decision element being associated with a decision identifier, the workflow task UI including a container component and metadata, the container component storing parameters of the application at run-time, the metadata including logic that is executable within a view of the container component during run-time, during run-time use of the application: displaying the workflow task UI, and receiving user input to a decision element within the view of the container component, and in response, passing a decision identifier associated with the decision element from the container component to the metadata, and executing at least a portion of the logic of the metadata to update the decision identifier with a workflow backend to complete the task.

Abstract: Provided are a data processing system and method based on dynamic redundancy heterogeneous encoding, and a device. The method comprises: respectively performing error correction encoding on information to be processed and a processing rule, so as to form encoded information to be processed and an encoded processing rule; processing, by using the encoded processing rule, the encoded information to be processed, so as to obtain response data; and then performing error correction decoding on N pieces of response data, so as to obtain processing result information of the information to be processed.

Abstract: The present disclosure provides a method of processing data based on QUIC protocol stack, the method including: obtaining a connection identifier of a data packet; determining a server program used for processing the data packet, wherein the server program comprises a plurality of worker threads, wherein QUIC protocol stacks run in the plurality of worker threads; determining a corresponding worker thread from the multiple worker threads based on the connection identifier; and distributing the data packet to the corresponding worker thread to make the QUIC protocol stack on the corresponding worker thread process the data packet. The disclosure further provides a system, a computing device and a computer-readable storage medium of processing data based on QUIC protocol stack.

Abstract: The present disclosure provides a method for acquiring a network topology and an indoor distribution system. The method comprises: a host unit receives position information transmitted by each extended unit connected to the host unit, and determines a network topology structure according to the position information. In this way, the labor cost can be reduced. In addition, the network topology structure obtained by the method provided by the present disclosure has higher accuracy, so that position information of each node can be learnt at a host unit, the fault can be quickly located, and the system operation and maintenance cost is reduced.

Abstract: The present disclosure relates to a distributed antenna system and a signal transmission method. A first remote cascaded chain corresponding to a first channel set and a second remote cascaded chain corresponding to a second channel set are obtained by means of division, such that when a distributed antenna system needs to support multi-channel MIMO, remote units in the first remote cascaded chain and remote units in the second remote cascaded chain only need to be configured with communication link modules, the number of which is less than the number of channels, and the multi-channel MIMO is jointly supported by using the remote units in the first remote cascaded chain and the remote units in the second remote cascaded chain, such that the multi-channel MIMO capability of an overlapping region can be realized.

Abstract: An orbital sander (10) comprises a housing (14) and a drive unit (50) in the housing (14). The drive unit (50) includes an electric motor (58) and a drive shaft (68). The electric motor (58) has a motor output shaft defining a rotational axis (62). The drive shaft (68) is coupled for co-rotation with the motor output shaft about the rotational axis (62). The drive shaft (68) has an eccentric portion (70) configured to orbit about the rotational axis (62). A sanding pad (100) is coupled to the eccentric portion (70). A handle (16) extends from the housing (14) at an oblique angle greater than 90 degrees relative to the rotational axis (62). A battery pack (26) is coupled to the handle (16) for providing power to the electric motor (58). The orbital sander (10) is convenient for users to grasp.

Abstract: Methods, systems, and computer-readable storage media for receiving, by an application studio, user input indicating two or more artifacts of an enterprise application developed using the application studio, generating, by an enterprise application artifact composer of the application studio, an enterprise application artifact as a universal tag that provides an abstract representation of the enterprise application, and during deployment of the enterprise application: retrieving the enterprise application artifact, determining a target runtime that the enterprise application is to be deployed to, determining a set of dependencies for each artifact of the enterprise application, and for any dependency in the set of dependencies that is absent from the target runtime, retrieving binary code of the dependency from a library to deploy the dependency to the target runtime.

Abstract: Methods, systems, and computer-readable storage media for receiving data as input to a workflow task user interface (UI) provided by a workflow task application, the data being representative of a task of a workflow, and determining that a decision action has been executed in response to user input to a UI element representative of the decision action, and in response, providing, from the workflow task application through an entry application within which the workflow task application executes, an update entity request, inserting, by the entry application, a set of properties to a header of the update entity request, and transmitting the update entity request to a data service to update a data object at least partially based on the data, a workflow service updating an instance of the workflow based on the set of properties.

Abstract: An online system for determining advertisements for retrieval is disclosed. The online system may assign a unique category path associated with a hierarchy of advertisement categories to a plurality of advertisements accessible by the online system to determine advertisement information features. The online system may determine a plurality of types of user engagements associated with prior user interactions by a user with one or more of the advertisements or types of advertisements associated with the advertisements to determine a set of user engagement features. The online system may select user engagement features, from the set, based on determined user engagement results to obtain user related features. The online system may associate the user related features with an assigned unique category path, associated with the ad information features, to generate inverted index features. The online system may apply the inverted index features as inputs to a neural network.

Abstract: A semiconductor laser beam combining device includes at least two modular laser input equipments, a second diffraction grating, and an output coupler. The modular laser input equipment includes a semiconductor laser, a beam shaping component, a transformation lens, and a first diffraction grating arranged along an optical path in sequence. The semiconductor laser generates a beam. The semiconductor laser is located at a front focal point of the transformation lens. The first diffraction grating is located in front of a back focal point of the transformation lens. Each beam is gathered by the transformation lens and diffracted by the first diffraction grating to the second diffraction grating. Each beam is combined at an identical position and an identical diffraction angle on the second diffraction grating to generate a combined beam. The combined beam from the second diffraction grating enters the output coupler vertically and is outputted by the output coupler.

Abstract: An optical-path calibration module is provided. The optical-path calibration module defines a first light inlet, a second light inlet, and a first light outlet, and a first light-splitting device is disposed in the optical-path calibration module. The first light inlet is configured to receive a calibration beam of a calibration light-source or be closed. The second light inlet is configured to receive a target-light-source beam or the calibration beam from the calibration light-source. The first light outlet is configured to emit a detection beam to a photoelectric sensor. An angle of 45° is defined between the first light-splitting device and each of the first light inlet and the second light inlet.