332 results about "Model network" patented technology

The invention relates to a system and method that can detect botnets by classifying the communication activities for each client according to destination or based on similarity between the groups of collected traffic. According to certain aspects of the invention, the communication activities for each client can be classified to model network activity by differentiating the protocols of the collected network traffic based on destination and patterning the subgroups for the respective protocols. Those servers that are estimated to be C&C servers can be classified into download and upload, spam servers and command control servers, within a botnet group detected by modeling network activity, i.e. analyzing network-based activity patterns. Also, botnet groups can be detected by way of a group information management function, for generating an activity pattern-based group matrix based on group data, and a mutual similarity analysis, performed on groups suspected to be botnets from the group information.

A method of assessing a network uses a model (450) having nodes (100, 110) to represent parts of the network infrastructure and the application services, and having links to represent how the nodes influence each other. Dependencies or effects of the application services are found by determining paths through the nodes and links of the model (530). Such assessment can be useful for design, test, operations, and diagnosis, and for assessment of which parts of the infrastructure are critical to given services, or which services are dependent on, or could have an effect on a given part of the infrastructure. The dependencies or effects can encompass reachability information. The use of a model having links and nodes can enable more efficient processing, to enable larger or richer models. What changes in the dependencies or effects result from a given change in the network can be determined (830).

In Pavlovian and instrumental conditioning, rewards typically come seconds after reward-triggering actions, creating an explanatory conundrum known as the distal reward problem or the credit assignment problem. How does the brain know what firing patterns of what neurons are responsible for the reward if (1) the firing patterns are no longer there when the reward arrives and (2) most neurons and synapses are active during the waiting period to the reward? A model network and computer simulation of cortical spiking neurons with spike-timing-dependent plasticity (STDP) modulated by dopamine (DA) is disclosed to answer this question. STDP is triggered by nearly-coincident firing patterns of a presynaptic neuron and a postsynaptic neuron on a millisecond time scale, with slow kinetics of subsequent synaptic plasticity being sensitive to changes in the extracellular dopamine DA concentration during the critical period of a few seconds after the nearly-coincident firing patterns. Random neuronal firings during the waiting period leading to the reward do not affect STDP, and hence make the neural network insensitive to this ongoing random firing activity. The importance of precise firing patterns in brain dynamics and the use of a global diffusive reinforcement signal in the form of extracellular dopamine DA can selectively influence the right synapses at the right time.

A method for detecting surveillance activity in a computer communication network comprising automatic detection of malicious probes and scans and adaptive learning. Automatic scan/probe detection in turn comprises modeling network connections, detecting connections that are likely probes originating from malicious sources, and detecting scanning activity by grouping source addresses that are logically close to one another and by recognizing certain combinations of probes. The method is implemented in a scan/probe detector, preferably in combination with a commercial or open-source intrusion detection system and an anomaly detector. Once generated, the model monitors online activity to detect malicious behavior without any requirement for a priori knowledge of system behavior. This is referred to as “behavior-based” or “mining-based detection.” The three main components may be used separately or in combination with each other. The alerts produced by each may be presented to an analyst, used for generating reports (such as trend analysis), or correlated with alerts from other detectors. Through correlation, the invention prioritizes alerts, reduces the number of alerts presented to an analyst, and determines the most important alerts.

A system and method is provided for using network asset management data including attributes such as device type, location, and link speed for modeling network performance and identifying performance issues. The invention provides measured and observed performance data for one or more segments of a network (e.g., one or more hops or end to end) so that identification of diminished performance in the network might be accomplished. Additionally, the invention may provide projections of performance impacts based upon proposed changes to one or more components in the network. Performance metrics, ping times, MIB data, link speeds, propagation delays, device latency, serialization rates, conversational speeds, conversational utilization and other data are used to model a network for determining or isolating problematic issues and to forecast performance impacts of potential changes to the network.

The present invention provides novel techniques for graphically modeling, displaying, and interacting with parametric hybrid models used to optimize and control components of industrial plants and enterprises. In particular, a graphical modeling tool of a control/optimization system for controlling a plant or enterprise is configured to transmit a graphical user interface to a user, wherein the graphical user interface enables a plurality of command inputs relating to a plurality of parametric hybrid models based on a security access level of the user. The parametric hybrid models may be displayed by the graphical user interface as nodes of a network with connections connecting the nodes. The user may graphically manipulate the nodes and connections associated with the parametric hybrids models to either modify optimization constraints of the model network, or actually modify the manner in which the parametric hybrid models function (e.g., inputs, outputs, parameters, and so forth, of the parametric hybrid models), depending on the access level of the user.

Variable accuracy simulation software for modeling an existing or planned wireless network includes a modeling framework containing a of predetermined software objects configured to model a network elements of the existing or planned wireless network. The simulation software enables the generation of at least one process related to at least one of the software objects, wherein the at least one virtual process corresponds to a physical process related to the existing or planned wireless communications network. The simulation software also includes accuracy adjusting means for selectively adjusting the accuracy and execution speed of an output generated upon the interaction of the at least one software object and the at least one virtual process. The simulation software also enables modeling network user behavior and cost-based analysis of the wireless network.

The invention relates to the field of three-dimensional geographic information systems, in particular to a method for rapidly displaying and browsing massive models of a three-dimensional geographic information system in a three-dimensional scene at high efficiency by utilizing SkylineTerraSuite series software which is mainly applied in more than 75,000 three-dimensional models in Shanghai. In the method, a browser/server B/S system architecture consists of at least three servers and a three-dimensional scene browsing client, and massive three-dimensional models are rapidly displayed and browsed in the three-dimensional scene by using technologies in three-dimensional data optimization, model management, model data network transmission and the like. By the method, related technical problems of model format optimization methods, massive model management methods, three-dimensional model network transmission methods and the like are mainly solved. The invention has the advantages that: the scene display speed is increased, the efficiency of model release, loading, query, addition, modification and the like is improved, the displaying and browsing speed of the models is increased, and the reading speed of data and the displaying speed of the models are increased.

The invention relates to a frequency domain full-waveform inversion seismic velocity modeling method. The method comprises the following steps of: 1) acquiring an original seismic shot gather record, focus wavelet information and an initial model used by inversion; 2) analyzing information acquired in the step 1), and determining basic inversion parameters and a full-waveform inversion frame from low frequency to high frequency based on a forward modeling algorithm and an optimization algorithm; 3) calculating to acquire the most appropriate forward and inversion model network for different frequencies; 4) compressing data dimensions which participate in inversion by a principal component analysis method during low-frequency inversion; 5) judging whether projection matrix dimensions corresponding to different frequencies meet the threshold value conversion standard, if the conversion standard is met, performing a next step, and if the conversion standard is not met, returning to the step 4); 6) introducing a focus encoding method, and pressing crosstalk noise by a random phase encoding method; 7) judging whether an iteration stopping condition is met, if the iteration stopping condition is met, performing a next step, and if the iteration stopping condition is not met, returning to the step 6); and 8) if the inversion of all the frequencies is not finished, returning to the step 3) until the inversion of all the frequencies is finished, acquiring the final velocity model, and outputting the velocity model.

The present invention relates to methods for preparing degradable model networks from any monomer functionality with any degradation methodology. It is based on the use of Atom-Transfer Radical Polymerization and CLICK chemistry to form the desired product.

The present invention provides novel techniques for graphically modeling, displaying, and interacting with parametric hybrid models used to optimize and control components of industrial plants and enterprises. In particular, a graphical modeling tool of a control/optimization system for controlling a plant or enterprise is configured to transmit a graphical user interface to a user, wherein the graphical user interface enables a plurality of command inputs relating to a plurality of parametric hybrid models based on a security access level of the user. The parametric hybrid models may be displayed by the graphical user interface as nodes of a network with connections connecting the nodes. The user may graphically manipulate the nodes and connections associated with the parametric hybrids models to either modify optimization constraints of the model network, or actually modify the manner in which the parametric hybrid models function (e.g., inputs, outputs, parameters, and so forth, of the parametric hybrid models), depending on the access level of the user.

The invention aims to solve the problems in the existing traffic signboard recognition method that the recognition target falls into a single group and the speed in doing so is slow. The invention, out of this concept, provides a fast traffic signboard recognition method based on convolution neural network, referred to as FTSR-CNN in abbreviation. This method comprises: using the convolution kernel sliding filter extracted characteristics; obtaining the loss of the network in the forward learning process, and ensuring the accuracy of the network model to the recognition of multiple categories of signboards; optimizing the network performances through the adjustment of the parameters, the activation of the function types and the reduction of dimensions for better accuracy and timeliness eventually; and at the same time, in order to make the samples more diverse, conducting data adding and expanding to the samples in the data set based on affine transformation. The recognition rates of the FTSR-CNN for two data set tests of the German traffic signboard data set GTSRB and the Tsinghua-Tencent 100K are recorded as 95.74% and 96.67% respectively. The results indicate that the recognition speed is increased on the same recognition accuracy level through the modification of a previous model network and the start up of different training strategies by the FTSR-CNN.

The invention discloses a lightweight convolutional neural network pedestrian recognition method, and the method comprises the steps: obtaining original pedestrian data, obtaining a pedestrian image with mark information, and constructing a pedestrian data set; designing an input size of a pedestrian model network, clustering the data set, and selecting an appropriate candidate box; preprocessingthe image, and expanding data; constructing a convolutional neural network, and sending the preprocessed image to the network for training to obtain a network model with a pedestrian recognition function; sending the images with the mark information into a network in batches for training; and checking the loss value and the accuracy on the training set, adjusting the learning rate or increasing the number of iterations and then training again if the result is not ideal, testing the trained network on the verification set if the result is ideal, and adjusting the network again according to theverification result. The identification precision can be improved while the real-time performance of the target detection model is ensured, so that the network model can run on a hardware platform with relatively low configuration.

The present invention provides novel techniques for graphically modeling, displaying, and interacting with parametric hybrid models used to optimize and control components of industrial plants and enterprises. In particular, a graphical modeling tool of a control/optimization system for controlling a plant or enterprise is configured to transmit a graphical user interface to a user, wherein the graphical user interface enables a plurality of command inputs relating to a plurality of parametric hybrid models based on a security access level of the user. The parametric hybrid models may be displayed by the graphical user interface as nodes of a network with connections connecting the nodes. The user may graphically manipulate the nodes and connections associated with the parametric hybrids models to either modify optimization constraints of the model network, or actually modify the manner in which the parametric hybrid models function (e.g., inputs, outputs, parameters, and so forth, of the parametric hybrid models), depending on the access level of the user.

Computer networks include multiple hardware and software devices, each working together to provide a favorable user experience to an operator. Many of these devices are built to standards that have been published by international standards organizations. These standards include functional test criteria that, when executed successfully, assure functionality of the device within a group of devices. The technology disclosed maintains libraries of canned tests based on these published standards. In addition, the technology disclosed can collect, adapt, and execute sets of predefined transactions to a target test network. This will validate that the target test network can scale up to a desired combination of transactions of different types. This is accomplished by manipulating example transaction sets captured from probes or routers that save network accounting records from a model network.

The invention discloses a Chinese comment emotion analysis method based on a GRU neural network, belonging to the field of natural language processing and deep learning. The method comprises the following steps: (1) firstly loading corpus data and segmenting words by a jieba word segmentation tool; (2) removing some useless pause words and dividing the corpus into a training set and a test set proportionally; (3) using word 2vec to train the word vector of the corpus, taking the mean value of all the word vectors of each sentence to generate the vector of the corresponding sentence, and then using word 2vec model network to carry on the back propagation training to calculate and generate the corresponding word vector finally; (4) inputting the word vector with sentence emotion generated byword 2vec into GRU neural network model for training; (5) constructing the test set according to the training set method, and inputting the GRU neural network to classify the emotion. The method utilizes a classification model based on the GRU neural network to classify emotions, and the speed of the model is remarkably improved at the same time of obtaining a good effect.

The invention discloses a method for building adaptive soft sensor. The method comprises the following steps. The input and schedule vectors are constructed, and a novel learning algorithm that uses online subtractive clustering is used to recursively update the structure and parameters of a local model network. Three rules are proposed for updating centers and local model coefficients of existing clusters, for generating new clusters and new models as well as for merging existing clusters and their corresponding models. Once verified, the online inferential model can be created to generate the predicted value of process. Thus, it does not need much memory space to process the method and can be easily applied to any other machine.

The invention discloses a zero-sample image classification method based on a regression variation auto-encoder. The invention relates to a method for identifying graphs by using electronic equipment.A variational auto-encoder is reconstructed by training, aligning and crossing; calculating regression network loss, the global loss L of the whole model network is calculated, the classifier is trained, the classification accuracy is calculated, zero-sample image classification based on the regression variation auto-encoder is completed, and the defects that in the prior art of generalized zero-sample classification, training samples are deficient, generated samples lack semantics, and a generative adversarial network is prone to collapse are overcome.