92 results about "Network migration" patented technology

A hierarchical communication system, arranged in a spanning tree configuration, is described in which wired and wireless communication networks exhibiting substantially different characteristics are employed in an overall scheme to link portable or mobile computing devices. Copies of data, program code and processing resources are migrated from their source toward requesting destinations based on request frequency, communication link costs and available local storage and / or processing resources. Each appropriately configured network device acts as an active participant in network migration. In addition, portable two-dimensional (2-D) code reading terminals are configured to wirelessly communicate compressed 2-D images toward stationary access servers that identify the code image through decoding and through comparison with a database of images that have previously been decoded and stored.

A method and means for reducing the workload on a system administrator are provided for use in the migration of a storage device or a storage network. A storage network management device is connected to all storage devices and host computers related to the migration via a management network. This configuration allows the storage network management device to acquire configuration information on the network and attribute information on the data extents of the storage devices from the storage devices and the host computers related to the migration. Therefore, the storage network management device can control the migration of configuration management information across different networks by referring to the acquired information.

In embodiments of the present disclosure improved capabilities are described for increasing the bandwidth in a wireless communication network, where centralized optimization servers with publish-subscribe broker services are utilized in conjunction with a queuing service application that provides packet service continuity when a mobile device moves between different access control nodes of the wireless communication network, and wherein the queuing service application is connected to a publish / subscribe broker network to receive service packets matching the service packets directed to the mobile device, wherein the service application makes the matching service packets available to the mobile device to replace service packets that the mobile device did not receive during a time in which the mobile device is in transition between being connected to any of the access control nodes.

The invention belongs to the field of machine learning technology and discloses a neural network migration learning method based on a virtual image dataset. The method comprises the steps that virtual image data is acquired through graphic simulation software, classified marking is performed on a virtual dataset, and an image dataset is obtained; a neural network is utilized to train a target recognition model through preprocessing of the virtual dataset, and a parameter model trained through the virtual dataset is saved; and migration learning is performed, and a new target dataset is utilized to train the saved pre-training model again. According to the method, a needed scene or target is collected through a virtual engine to construct a dataset based on the virtual image dataset, the pre-training model is obtained through training of the virtual dataset, and the migration learning method is utilized to achieve the practical application effect on a small quantity of datasets.

A hierarchical communication system, arranged in a spanning tree configuration, is described in which wired and wireless communication networks exhibiting substantially different characteristics are employed in an overall scheme to link portable or mobile computing devices. Copies of data, program code and processing resources are migrated from their source toward requesting destinations based on request frequency, communication link costs and available local storage and / or processing resources. Each appropriately configured network device acts as an active participant in network migration. In addition, portable two-dimensional (2-D) code reading terminals are configured to wirelessly communicate compressed 2-D images toward stationary access servers that identify the code image through decoding and through comparison with a database of images that have previously been decoded and stored.

A method, correlative device and system for virtual network migration are provided. The method of the embodiments of the present invention includes: locating the source physical node in a region physical network; obtaining the virtual element corresponding to each virtual network in the source physical node and the state information of each physical node in the region physical network; determining, according to the virtual elements and the state information, a physical node that can implement the virtual network migration in the region physical network; reconstructing the mapping of each virtual network and the region physical network respectively in the physical node, comparing the mapping corresponding to each virtual network, and selecting the mapping with the least migration consumption as the mapping that implements the migration; sending, according to the mapping that implements the migration, a migration instruction to the physical node that needs to implement the virtual network migration. Furthermore, the embodiments of the present invention also provide the correlative device and system that implement the method.

The present invention provides a deep migration learning method for a domain-adaptive network. According to the distribution difference corresponding to each task-related layer, classification error rate and mismatch degree, the value of the loss function of the domain-adaptive network is determined, wherein any The distribution difference corresponding to the task-related layer is the distribution difference between the probability distribution of the features in any task-related layer corresponding to the source domain and the target domain respectively; and based on the value of the loss function, the parameters of the domain adaptive network are updated to Adapting the domain-adaptive network to the target domain; thereby taking the distribution difference between the probability distributions of the features in each task-related layer corresponding to the source domain and the target domain respectively as an integral part of the value of the loss function of the domain-adaptive network, Each task-related layer of the deep network is matched in different fields at the same time, and the difference between the marginal distribution and the conditional distribution in different fields is better corrected, which ensures the reliability of transfer learning and finally ensures the effect of domain-adaptive network transfer learning. .

The invention provides a method and device for dynamically migrating VLAN (virtual local area network) configuration. The method comprises the following steps of: storing virtual machine ID and corresponding VLAN configuration information in a network management system in advance; transmitting the VLAN configuration of a virtual machine to an upstream physical switch corresponding to a migrated physical server of the virtual machine when the network management system senses out a network migration event of the virtual machine, wherein the migrated physical server transmits an RARP (reverse address resolution protocol) message so as to update ARP tables of all the upstream physical switches after the migration of the virtual machine is finished; and if the migration event is finished before the network management system finishes transmitting the VLAN configuration, the network management system ages the ARP table on a physical switch by an SNMP (simple network management protocol) or command line so as to update the ARP tables of all the upstream physical switches of the migrated physical server. According to the invention, the VLAN configuration of the virtual machine can be migrated dynamically along with the migration of the virtual machine.

The invention discloses a method and a device for virtual network configuration migration. In the invention, a network management device senses a network migration event of a virtual host, and issues a configuration to a migration uplink physical switch migrated by the virtual host, and updates the access restrictions of the migration uplink physical switch on the virtual host; the network management device also can issue a configuration to an outmigration uplink physical switch of the virtual host, and updates the access restrictions of the outmigration uplink physical switch on the virtual host; the physical switches can flexibly control the access of external users on the virtual host through issuing configurations to the physical switches; as additional protocols are not realized during the process, the realizability is good; meanwhile, the realization process is independent from a detailed physical network, and does not need special supports from the physical network; therefore, the compatibility is very good.

The invention discloses a face attribute recognition method based on multi-instance multi-label depth migration learning, which comprises the following steps of preparing a face image data set, extracting a plurality of neural layer features of a depth convolution neural network migration model for each face image, and combining the neural layer features to form a multi-layer face feature; builinga network model for extracting multi-label relationship features, and training the parameters of the network model by using multi-layer face features as input and multi-face attribute tags as true values; designing a linear binary classifier for each face attribute, migrating the trained network model of multi-label relational features to the multi-face attribute classifier model as a feature extractor, and training each linear binary classifier by using face image data set. The method of the invention selects a transfer learning mode, rapidly and efficiently migrates a highly active transfermodel to a selected data set, builds a multi-tag relation characteristic model with simple training structure, and simultaneously trains a linear binary classifier of multiple human face attributes.

Migration of a circuit-based-switch to a packet-based-soft-switch in a SS7 telecommunications network is accomplished with no disruption to subscriber service and retains and reuses the assigned signaling point code (SPC) of the existing circuit-based-switch. The system and method inserts a new soft-switch between an SS7 signaling transfer point (STP) and the existing circuit-based-switch in a way where traffic is split incrementally off the circuit-based-switch to the soft-switch so that no administrative changes are required at either the STP or the existing circuit-based-switch. The soft-switch is gradually cutover to act as a proxy STP and tandem switch for the existing circuit-based-switch and routes all calls to and from the network on behalf of the existing switch. The system makes use of multiple internal networks of the soft-switch to accomplish the controlled routing of bi-directional traffic. The circuit-based-switch can eventually be removed from service after subscribers are migrated to the new soft-switch.

A method begins by a processing module receiving a retrieval request for a file and determining whether the file is being migrated from a legacy storage system to a dispersed error coding storage system. The method continues with the processing module determining a retrieval option for the file when the file is being migrated from the legacy storage system to the dispersed error coding storage system. The method continues with the processing module retrieving the file, based on the retrieval option, in at least one of a legacy format from the legacy storage system and a plurality of sets of encoded data slices from the dispersed error coding storage system.

The invention discloses a neural network migration method based on shallow learning, and the method comprises the steps: 1 carrying out the classification and division of a target task data set, carrying out the marking of the target task data set, and storing the marking data as the training data x0 of a shallow neural network; 2 inputting x0 to a shallow neural network, training layer by layer to obtain a pre-trained shallow neural network model, and outputting data x2 after x0 passes through the pre-trained neural network model; and 3 taking the obtained output data x2 of the pre-trained shallow neural network model as the input of the deep neural network model of the target task, training the whole deep network by using the marked data of the target task, and carrying out fine tuning on the whole network parameters to complete neural network migration. According to the method, the shallow neural network learning model trained layer by layer is used as a basic model of task migration, so that the migration task is simple and efficient, the expansibility is high, and the problem that the migration effect of the traditional end-to-end deep neural network is uncertain in fluctuation and even appropriate for reversibility is solved.

The invention discloses a video content management method based on semantic hidden indexing, which ensures that a semantic space table (SST) and video data coexist all the time in the network migration and transmission process by defining the SST and performing integrated semantic indexing on the SST and the video data by an information hiding method. For a video application system (such as an intelligent play and download agent system, a video classification management system, a network management and control system and the like), the SST of the video data is extracted to be subjected to comparative calculation with a semantic request table (SRT) at a destination end, so that the calculation result can help the system to decide a specific processing mode of the video data. The semantic information of the hidden indexing is hard to erase or falsify, so that the semantic information loss of the video data in the secondary transmission process is prevented; and the content video data can be effectively unified and correlated, so that the redundant transmission of data is reduced, and the utilization efficiency of the network is greatly improved. Meanwhile, the video search engine and the other video application systems can be helped to better perform the selection, rejection, abandonment and other operations on the video data, so that the video data in the network space can be transmitted orderly, and the video can be found more efficiently and quickly.

Embodiments of the present disclosure disclose a convolution neural network migration method. The method comprises the steps of: improving the last pooling layer of the first convolution neural network to obtain a second convolution neural network so that the resolution of the input image of the second convolution neural network is greater than the resolution of the input image of the first convolution neural network. The method is suitable for input data sets of different sizes, such as high resolution fundus images, and saves the computational resources consumed in the development of specialconvolutional neural networks.

The invention relates to a fast video super-resolution reconstruction method based on a reduced convolutional neural network. Neighborhood information between video frames is utilized and the reconstruction speed is ensured. With consideration of the direct impact on the operation speed of the network by the input size, the provided method enables the pre-interpolation process of the traditional method to be removed; and features are extracted from a plurality of low-resolution input video frames directly and multi-dimensional feature channel fusion is carried out. A parametric linear correction unit is used as an activation function and a network structure is adjusted by using a small filter size to carry out multi-layer mapping, so that a phenomenon that important information of the video is lost because of the zero gradient in the network is avoided. And then a deconvolution layer is added at the end of the network to carry out upsampling, so that a reconstructed video. Meanwhile, with a network migration strategy, a reconstruction model under different scaling factors is realized quickly; and more high-frequency detail information is kept in the reconstructed video image and the reconstruction speed is increased.

The invention discloses an urban noise identification method based on deep network migration features and augmented self-encoding. The method comprises the following steps: 1, preprocessing each typeof collected urban noise signals, including denoising, framing and windowing; 2, converting the processed noise signal into a spectrogram; 3, performing feature extraction on the spectrogram obtainedin the step 2 by using a plurality of pre-trained deep convolutional neural networks; 4, fusing the obtained feature vectors x by using an augmented auto-encoder; 5, on the basis of the fusion features in the step 4, constructing a multilayer one-class classification model; 6, calculating an output weight and a decision threshold value of ML-OCRLS; and 7, carrying out classification prediction onunknown signals. The hidden layer neurons of the augmented auto-encoder provided by the invention can adjust and optimize all features, main information can be extracted based on ML-OCRLS of the augmented auto-encoder, feature redundancy is reduced, and meanwhile, multiple transfer learning features are effectively fused, so that the classification precision of a classifier is improved.

The invention relates to a breast tumor classification algorithm based on a convolutional neural network VGG16. The algorithm comprises the following steps that: data preprocessing: for a dataset which presents a data imbalance state, carrying out imbalance processing and data enhancement processing; the establishment of the convolutional neural network: 1) network pre-training: utilizing the VGG16 to carry out network training on an ImageNet large natural image dataset, and storing trained weight; 2) network key node selection: utilizing different layers of the VGG16 network to carry out feature extraction on a breast tumor DDSM (Digital Database for Screening Mammography) dataset, applying the same SVM (Support Vector Machine) classifier for classification for extracted features, and selecting a layer with highest classification performance as a node constructed by a new network; and 3) connecting two layers of full connection and one layer of softmax to form a new network behind thenode constructed by the selected network; and carrying out migration learning.

The invention provides production method, device and network equipment of a binding table. The method comprises the following steps of monitoring a local MAC (Media Access control) address table; when monitoring that the MAC address table varies, producing a binding table of a variation port in variation table information according to the variation table information and prestored RA message information in the MAC address table. The invention has the technical scheme that network migration is discovered by monitoring the MAC address table, a corresponding binding table is produced according toa varied MAC address entry, production of the binding table does not depend on an address conflict detection message sent by a terminal, the terminal does not need to perceive the network migration, and the traditional binding table can be produced under a network topology structure that the terminal is not directly connected with the network equipment.

The invention discloses a virtual machine entity thermal migration network smooth switching method and device. The virtual machine entity thermal migration network smooth switching method comprises the steps that: a virtual machine entity is thermally migrated to a target physical server from a source physical server and enters a final stage, and before the virtual machine entity is shut off, a network migration auxiliary program is started to cache a datagram sent to the virtual machine entity on the source physical server; and after startup state information of the virtual machine entity on the target physical server is acquired, the cached datagram is forwarded to the virtual machine entity on the target physical server. The virtual machine entity thermal migration network smooth switching method and device provided by the invention solve the problem of datagram loss in the thermal migration process, and can be applicable to any form of Ethernet network.

The invention relates to an unsupervised image restoration method for resisting network migration learning based on mask generation, comprising the following steps: step S101, obtaining an image to berepaired and setting mask parameters;step 102, utilize a mask to generate an antagonistic network for repairing that image to be repaired; The mask generation countermeasure network comprises a self-encoder and a discriminator. The training process of the mask generation countermeasure network comprises adding a mask layer after the output of the self-encoder, and the mask layer is used as the first layer of the discriminator to realize the training of the self-encoder and the discriminator. Compared with the prior art, the invention has the advantages of high accuracy and high efficiency.

The invention discloses a plant disease and insect pest identification method based on sparse network migration, and belongs to the technical field of intelligent agricultural disease and insect pestidentification. The method comprises the following steps: designing a pruning algorithm, iteratively traversing a network, freezing redundant parameters in a source domain network, and generating a retrained optimal sparse sub-network structure; employing deep migration learning, migrating the sparse network to a target domain, proposing a sparse network migration hypothesis, verifying the feasibility of the sparse network, exploring the potential association between a target task and existing knowledge, and initializing the network through the weight of a source domain, and achieving the knowledge migration and reuse on the target domain; finally, finely adjusting the sub-network by using a small number of samples of the target domain data, optimizing the network performance, and finishing the task migration, thereby solving the practical application problem. Plant diseases and insect pests can be recognized, the network detection precision is improved through sparse migration, and meanwhile, the problems that a traditional deep method needs to train a dense network, calculation expenditure is large, the requirement for hardware is high, and popularization is not facilitated are solved.

The invention discloses an automatic configuration migration system and method based on an SDN (Software Defined Network). The system comprises a plurality of virtual machines, a plurality of virtual switches and an SDN controller, wherein the virtual machines are connected with the virtual switches, and the plurality of virtual switches are all connected with the SDN controller. The method comprises the following steps: S1, connecting the SDN controller with the virtual switches successfully; S2, connecting the virtual machines to the virtual switches; S3, owning, by the SDN controller, configuration information of all the virtual switches in the whole network; S4, migrating the virtual machines on line; S5, sensing the migration of the virtual machines and reporting the migration to the SDN controller by new virtual switches; S6, transmitting, by the SDN controller, the corresponding configuration to the new virtual switches; S7, sending, by the SDN controller, a message to delete the configuration of the old virtual switches; and S8, accomplishing automatic configuration migration of the virtual switches. The system and the method have the advantages that the operation is simple and efficient, the reliability is good, the performance is excellent and automatic network migration is realized.

The invention provides an automatic target recognition method based on a load pre-training convolutional network. The automatic target recognition method mainly comprises the steps of network reconstruction, network characteristic synthesis, classifier conversion and network model fine-adjustment, and the process of the method is that module splitting is performed by using a trained convolutionalnetwork framework, reconstruction is performed according to a target recognition task on the premise of weight retaining, and meanwhile the last layer of all original connecting layers is retained toserve as extracted characteristics, is attached with a response label and then is sent to a support vector machine based on a Gaussian kernel for training test. By adopting the method, target form recognition of only small sample databases can be processed, a network migration frame is provided for weight retaining and fine adjustment, and meanwhile the training speed and transferability of targetrecognition tasks are improved.

The invention discloses an Open Flow based linear protection method. The method comprises steps that an Open Flow controller sends a flow table to a data transmitting end and a data receiving end; the data transmitting end receives a multi-protocol label switching (MPLS) message and sends the MPLS message through a communication link to the data receiving end according to the flow table sent from the Open Flow controller; the data receiving end, according to the flow table sent from the Open Flow controller, performs behavior selection on a corresponding communication link to receive the MPLS message. According to the method, 1+1 and 1:1 linear protection can be realized based on the semanteme of Open Flow. The Open Flow based linear protection method has actual application value in promoting network migration to an Open Flow unified management stage.

The invention discloses a deep network migration learning method facing a marked noise apparent age database. The apparent age database is randomly divided into two parts according to the preset proportion, one is training set, and the other is verification set. The proportion of training set is higher than verification set. A small amount of data is randomly extracted from the training set and repeated n times to obtain n sub-training sets. The set of the remaining data in the training set is denoted as data set A. According to the transfer learning method, n classification models are obtained by in-depth learning of n sub-training sets, and then the data set A is identified by using n classification models. The deep network migration learning method facing the marked noise apparent age database enables the apparent age database with high accuracy to be obtained when the marked noise is weakened by the apparent age database, and can effectively weaken the influence of the marked noiseon the experimental result and make the result more credible.

The invention relates to the technical field of knowledge distillation, and discloses a knowledge distillation method and device fusing channel and relation feature learning, and equipment, and the method comprises the steps: constructing an untrained student network and a pre-trained teacher network; respectively inputting the training data into a student network and a teacher network to obtain an output result of the student network and an output result of the teacher network, wherein the training data further comprises corresponding real label data; determining a distillation loss function based on the channel data of the student network and the teacher network, the output result of the student network, the output result of the teacher network, and the relationship between the learning network and the teacher network migration sample; and performing iterative training on the student network based on the distillation loss function. According to the method and device, the student network model can be effectively compressed, and the performance of the student network can be further improved and even exceeds the performance of a large teacher network.

An apparatus and method for mapping a newly assigned Universal Fleet Member Identifier (UFMI) to a previously assigned UFMI is disclosed. In an embodiment, an Integrated Digital Enhanced Network (iDEN) routing request message is received from a dispatch application processor (DAP) associated with a calling party at a serving home location register (HLR) of a called party. The routing request message includes a previously assigned UFMI of the called party. A newly assigned UFMI of the called party is obtained by the HLR by consulting a subscriber table stored in the HLR. An iDEN routing response message is transmitted to the DAP by the HLR, where the routing response message includes the newly assigned UFMI associated with the called party.