Abstract: An online trimming device and method for a micro-shell resonator gyroscope is provided. A micro-shell resonator gyroscope fixing fixture and a mode test circuit in the device are placed in a vacuum test cavity provided with a circuit interface. The mode test circuit and a host computer are connected through a circuit interface on the vacuum test cavity. The gyroscope fixing fixture is provided with a signal interface, and the electrodes on the gyroscope substrate are connected to the signal interface. The signal interface on the fixture is connected to the mode test circuit. The laser etching module is located at the top of the device. An opening is formed in the gyroscope fixing fixture. The vacuum test cavity is provided with a transparent trimming window. The laser acts on the edge of the resonant structure of the gyroscope through the trimming window and the through hole of the fixture.

Abstract: The disclosure relates to a method and a system for hand pose recognition, a device and a storage medium are disclosed in embodiments of the disclosure. The method includes: capturing a RGB image of a hand from a RGB camera and capturing a depth image of the hand from an active depth camera, so as to obtain a hand pose data set according to the RGB image and the depth image; processing the hand pose data set to obtain a 3D joint position, and taking the 3D joint position as a data set for training a software model; extracting the RGB image by a feature extractor based on a depth neural network to obtain a feature map of a hand pose; and processing the feature map according to an attention mechanism to obtain a global feature map of the hand pose.

Abstract: A training method for a robust neural network based on feature matching is provided in this disclosure, which includes following steps. Step A, a first stage model is initialized. The first stage model includes a backbone network, a feature matching module and a fullple loss function. Step B, the first stage model is trained by using original training data to obtain a second stage model. Step C, the second stage model is attacked so as to generate PGD adversarial samples of the original training data, and the second stage model is trained again with the generated adversarial samples and the original training data. Step D, training parameters are adjusted and the second stage model is trained again, and parameters for which the model has highest accuracy on an original test set are saved.

Abstract: A dynamic spectrum sharing method based on user online learning and low-overhead cooperation is provided. In the method, a secondary user (SU) obtains spectrum sharing scene information and requirement parameters; then initializes state parameters and spectrum sensing parameters of a channel. At the beginning of each slot, the SU determines whether a direct transmission condition is met according to the transmission condition of the previous slot, and directly continues transmission on the channel of the previous slot or obtains an idle channel according to the channel state and spectrum sensing for transmission, and updates state parameter and spectrum sensing parameter. Finally, an SU base station performs conflict coordination according to the channel occupancy condition for a fixed number of slots.

Abstract: Disclosed is a method for distinguishing sunny-rainy weather based on time division long-term evolution network, including the following steps: acquiring sunny-rainy feature by extracting communication measurement statistics of time division long-term evolution network base stations in a certain area; establishing a training set according to that observation result of multiple statistical periods, multiple base stations and multiple rain gauges in the region; establishing a sunny-rainy discrimination model combined with machine learning binary classification algorithm, so as to realize the identification of rainfall events covered by a single base station; calculating the reliability of rainfall events at specific locations based on the comprehensive judgment results of multiple base stations.

Abstract: The disclosure relates to a method and device for identifying unknown traffic data based on a dynamic network environment. The method includes following steps. The known traffic in the network data is classified by using the known network traffic classification model, then the preliminary determination is performed according to a classification prediction result, network data preliminarily determined as the unknown traffic data is classified by using the adaptive clustering method, and then respective classes are identified by using a similarity coefficient estimation method so as to identify the classes of the malicious traffic and the normal traffic, that is, to further identify and learn the unknown traffic data, and transform it into known traffic data, and then the known network traffic classification model is trained and updated again with the new known traffic data.

Abstract: The present disclosure provides a system and method for controlling a dynamically controllable ultrawide-amplitude and high-response ion source, including: resolving dwell time of ion beam machining during iterative machining; selecting an appropriate velocity V of a movable shaft of a machine tool according to a calculation result of the dwell time; and dynamically calculating process parameters of an ion source according to an initial surface error of an optical component and the velocity V of the movable shaft, and generating a corresponding numerical control (NC) program to machine the optical component. The present disclosure can control the removal function of the ion beam polishing in real time, improve the precision and efficiency of the ion beam polishing, and further reduce the requirement on a movement system of the machine tool and the depth of a damaged layer.

Abstract: The present invention provides a dynamic spectrum sharing method based on user online learning and low-overhead cooperation, and relates to the technical field of electromagnetic spectrums. The method comprises: first, a secondary user obtaining spectrum sharing scene information and requirement parameters; then initializing a state parameter and a spectrum sensing parameter of a channel; then, at the beginning of each time slot, the secondary user determining whether a direct transmission condition is met or not according to the transmission condition of the previous time slot, selecting to directly continue transmission on the channel of the previous time slot or obtain idle channel according to the channel state and spectrum sensing for transmission, and updating the state parameter and spectrum sensing parameter of the channel; and finally, a base station of the secondary user performing conflict coordination according to the channel occupancy condition every a fixed number of time slots.

Abstract: A gear-based mechanical metamaterial with continuously adjustable elastic parameters in a large range is provided. The gear-based mechanical metamaterial includes a gear array, a frame and connecting shafts. The gear array is formed by periodically extending mechanical metamaterial cells along an x direction and a y direction. Each of the mechanical metamaterial cells is formed by arranging a multiple gears. Adjacent gears of the multiple gears are meshed with each other. Each of the multiple gears includes a center hole and two centrosymmetric irregularly-shaped holes. A thickness of an elastic arm between the each of two centrosymmetric irregularly-shaped holes and an outer wall of a corresponding one of the multiple gears is uniformly increased or decreased. Each of the connecting shafts is arranged in a center hole of a corresponding one of the multiple gears.

Abstract: A device for isolating vibrations includes an ion trap, a cryocooler, a primary chamber, a secondary chamber, a vacuum ion pump, a heat exchanger, a sample chamber, a support part, a connector, a heat conduction part, a first platform, a second platform, and a flexible connecting part. The primary chamber, the secondary chamber, and the vacuum ion pump are fixedly disposed on the first platform. The connector is a hollow structure disposed between the primary chamber and the secondary chamber. The primary chamber communicates with the secondary chamber via the hollow structure thereby forming an airtight chamber. The vacuum ion pump is connected to the primary chamber via a five-way flange. The support part is fixed on the second platform. The cryocooler is fixed on the support part. The cryocooler includes a cold head and a machine head. The cold head is suspended in the primary chamber.

Abstract: A harmless low-consumption on-orbit continuous launch system includes a satellite platform, a launch apparatus and a plurality of CubeSats. The satellite platform carries the launch apparatus and dozens or hundreds of CubeSats, and is launched from a ground into an orbit for on-orbit operation. The launch apparatus is configured to store the plurality of CubeSats and provide power for on-orbit launching of each of the CubeSats. A solid working medium in the launch apparatus is activated by heating to undergo a phase change, and the activated solid working medium expands instantly and is converted into a high-pressure gaseous working medium. The high-pressure gaseous working medium does work to eject the CubeSats, such that the CubeSats obtain a speed increment. The CubeSats enter a transfer orbit towards different target spacecraft through the speed increment applied by the launch apparatus to perform a plurality of different on-orbit serving missions.

Abstract: A cross-site scripting (XSS) risk analysis method based on a Bayesian network and a STRIDE model includes: constructing an XSS attack-related STRIDE threat model of a network information release system; obtaining a network structure of a Bayesian network-based XSS attack risk analysis model based on the STRIDE model; obtaining prior probabilities of all nodes in the network structure of the Bayesian network-based XSS attack risk analysis model based on expert experience and a node ranking algorithm; obtaining a training dataset through simulation by using a rejection sampling algorithm or a direct sampling method; performing network training on the network structure of the Bayesian network-based XSS attack risk analysis model by using the training dataset to obtain the Bayesian network-based XSS attack risk analysis model; and reasoning a Bayesian network to obtain a quantitative analysis result of an XSS attack risk of the network information release system.

Abstract: The invention provides a converged network interface card, a message coding method and a message transmission method thereof. The converged network interface card comprises a PCIE host interface processing module, a high speed network card core logic, a crossbar switch XBAR, an Ethernet network card core logic, an Ethernet message dicing/slicing module, a physical layer, a high speed network/Ethernet message conversion module EoH, and a high speed network/Ethernet configurable network port. The invention supports customized high speed interconnection interface and a standard Ethernet interface on a set of network hardware, and supports three working modes on a set of physical hardware (high speed network mode, Ethernet mode and EoH mode transmitting Ethernet messages over the high speed network), implements seamless compatibility between the high speed network/Ethernet, and flexibly supports multimode applications such as scientific computing and cloud computing.

Abstract: The present disclosure discloses a Single-Event Transient (SET) pulse measuring circuit capable of eliminating impact thereof, and an integrated circuit chip. The SET pulse measuring circuit capable of eliminating impact thereof includes four parts: a SET pulse test chain, a latch circuit, a flip-flop test circuit, a latching self-trigger circuit. The integrated circuit chip is provided with a test chain module and two sets of SET pulse measuring circuits capable of eliminating impact thereof, and inputs of the two sets of SET pulse measuring circuits capable of eliminating impact thereof are the same and each are connected to an output terminal of the test chain module.

Abstract: A session-based recommendation method and device according to one or more embodiments of this disclosure are provided, which use a pre-trained recommendation model to perform item recommend. The method includes following contents: a directed session graph is constructed according to a session to be predicted; the directed session graph is then input into a gated graph neural network which outputs the item embedding vector; a user's dynamic preference is determined according to a user's current preference and a first long-term preference, the current preference is an item embedding vector of a last item in the session and the first long-term preference is determined according to the item embedding vector and an importance score of the item; a prediction score of a respective item is determined according to the dynamic preference and the item embedding vector; and a recommended item is output according to the prediction score of the respective item.

Abstract: The invention discloses a data structure-aware prefetching method and device on a graphics processing unit. The method comprises the steps of acquiring information for a memory access request in which a monitoring processor checks a graph data structure and read data, using a data structure access mode defined by a breadth first search and graph data structure information to generate four corresponding vector prefetching requests and store into a prefetching request queue. The device comprises a data prefetching unit distributed into each processing unit, each data prefetching unit is respectively connected with an memory access monitor, a response FIFO and a primary cache of a load/store unit, and comprises an address space classifier, a runtime information table, prefetching request generation units and the prefetching request queue.

Abstract: A fact validation method including the following steps: a statement to be validated is inputted and a searching is made for the statement to obtain an evidence set of the statement; a hierarchical heterogeneous graph consisting of an entity node, a sentence node and a context node is constructed based on the evidence set; the statement and the evidence set are spliced and a node is initialized to obtain feature representation of the node; the feature representation of the node is updated based on inference according to a propagation direction of a neural network of the node in the hierarchical heterogeneous graph; and an inference path for the updated feature representation of the node is built and a prediction result of the statement is output according to the inference path.

Abstract: A harmless low-consumption on-orbit continuous launch system includes a satellite platform, a launch apparatus and a plurality of CubeSats. The satellite platform carries the launch apparatus and dozens or hundreds of CubeSats, and is launched from a ground into an orbit for on-orbit operation. The launch apparatus is configured to store the plurality of CubeSats and provide power for on-orbit launching of each of the CubeSats. A solid working medium in the launch apparatus is activated by heating to undergo a phase change, and the activated solid working medium expands instantly and is converted into a high-pressure gaseous working medium. The high-pressure gaseous working medium does work to eject the CubeSats, such that the CubeSats obtain a speed increment. The CubeSats enter a transfer orbit towards different target spacecraft through the speed increment applied by the launch apparatus to perform a plurality of different on-orbit serving missions.

Abstract: Disclosed are a network traffic analysis method and a device based on multi-source network traffic data. The method includes: deploying a pre-training classifier pool in a network stream data source; receiving multi-source network stream data at a current moment for each data source, classifying the multi-source network stream data through an online classifier, performing feature processing and transformation on data collected by each network stream data source at each preset time interval, and transmitting processed traffic data features and a feature transformation matrix to a traffic drift detection module. The traffic drift detection module contains historical concept data to detect a concept drift according to the traffic data features, the feature transformation matrix and the historical concept data; if the concept drift is detected, the online classifier deployed by multiple sources is reset. This method is used for continuous real-time and accurate analysis of the multi-source network traffic data.

Abstract: A method for updating a stable cuckoo filter used for membership testing of data streams, executed by a processor, is described. The method includes the steps of: performing a first hash on a first element to be inserted into the stable cuckoo filter to determine a first candidate bucket; performing a second hash on a fingerprint of the first element to determine a second candidate bucket; selecting a target candidate bucket from a group consisting of the first candidate bucket and the second candidate bucket; inserting the first element into the target candidate bucket; updating the stable cuckoo filter according to one or more of a random update strategy, an insertion failure update strategy, a scanning strategy, a skip scanning update strategy, a blocked design strategy and a local time-sensitive update strategy; and obtaining an updated stable cuckoo filter.