67 results about "Connectome" patented technology

The invention provides an intelligent analysis system and method for magnetic flux leakage detection data in a pipeline. The process comprises the following steps: adopting time domain-based sparse sampling and KNN-softmax data complete set construction module, obtaining a complete magnetic flux leakage data set; a pipeline connection assembly discovery method based on combination of selective search and a convolutional neural network is adopted in the discovery module to obtain the accurate position of the weld joint; adopting an abnormal candidate region searching and identifying method based on a Lagrangian multiplication frame and multi-source magnetic flux leakage data fusion in the discovery model to find out defective magnetic flux leakage signals; Adopting a defect quantification method based on a random forest in a quantification module to obtain a defect size; a pipeline solution improved based on the ASME B31G standard is adopted in the solution module, and an evaluation result is output. An analysis method is provided from the overall perspective, and pretreatment, connection assembly detection and abnormity detection, defect size inversion and final maintenance decision are achieved.

The invention concerns agents with anti-bacterial activity and methods and intermediates for their production. The present invention further concerns the use of such agents for the treatment of bacterial infections in animals, including man. The agents are derivatives of vancomycin-type antibiotics, of structure: V-L-W-X; wherein V is a glycopeptide moiety which inhibits peptidoglycan biosynthesis in bacteria; L is a linking group; W is a peptidic membrane-associating element such as an element based on naturally-occurring animal or bacterial peptide antibiotics; and X is hydrogen or a membrane-insertive element.

A method is proposed for segmenting a brain image into a CSF region, a WM region and a GM region. An upper limit for the intensity values of a CSF region in the image is estimated such that the points of the image having an intensity less than this upper limit include a subset of the points which form a spatially connected group and which have a peaked intensity distribution. In other words, the invention exploits both the expected spatial distribution and expected intensity distribution of the CSF region. This makes it possible for the method to provide reliable discrimination of the CSF region even in CT images with poor image quality. Various methods are proposed for using the upper limit, and for improving the segmentation accuracy.

Existing methods for analyzing person-specific ‘connectomes’ are not computationally equipped for scalable, flexible, and integrated processing across multiple network resolutions and drawing from disparate data modalities—a major roadblock to utilizing ensembles and hierarchies of connectomes to solve person-specific machine-learning problems. The processes implemented in software described herein consists of an end-to-end pipeline for deploying ensembles and hierarchies of network-generating workflows that can utilize multimodal, person-specific data to sample networks, extracted from that data, across a grid of network-defining hyperparameters. In essence, this pipeline enables users to perform ensemble sampling of connectomes for given individual(s) based on any input phenotypic datatype, constructed from any data modality or hierarchy of modalities at any scale, and based on any set of network-defining hyperparameters.

The invention discloses a multi-scale feature fusion block in combination with context information. The multi-scale feature fusion block comprises a forward hierarchical connection group, a backward hierarchical connection group and a channel multi-scale selection module, wherein the forward hierarchical connection group is used for progressive inter-scale information fusion; the backward hierarchical connection group is used for information fusion between spanning scales; and the channel multi-scale selection module is used for carrying out scale feature channel selection on the backward hierarchical connection group. The invention further discloses a convolutional neural network comprising the multi-scale feature fusion block in combination with the context information, so that effectivefusion of multi-scale features is realized. The invention further discloses a pedestrian re-identification method, device and equipment based on the convolutional neural network and a storage medium,and the pedestrian re-identification accuracy can be improved.

The invention relates to a desktop upper limb rehabilitation robot and using method thereof, which relate to the technical field of the rehabilitation robot. The desktop upper-limb rehabilitation robot is composed of a bench rack, an omnidirectional wheel and drive system (1), a lower computer control system (2), a contact force detection system (3), a position detection system (4), an electromyography measurement system (5), an EEG measurement system (6) and a host computer control system (7), a Bluetooth system (8), and power supply which are connected by electrical signals. The desktop upper limb rehabilitation robot has the characteristics of small volume, strong household applicability, high measurement precision, high cost performance and wide application range. The EEG and EMG signals are used for training rehabilitation and evaluation for patients. The evaluation results are more convincing and accurate. The virtual animation display is realized through the host computer, whichcan improve the enthusiasm of the patient, increases the interest in a training process, and prevents the training from being too boring. The remote interaction function is realized by the host computer, so that the patient can also enjoy professional guidance and monitoring at home.

The invention discloses a multi-mode multi-layer fusion deep neural network for face anti-spoofing, and relates to image abnormal sample detection. The deep neural network comprises an image feature extraction front end and a neural network classifier. The network comprises a staggered neural network layer, a multi-modal weight adaptive module and a full connection layer classification unit; wherein the neural network front end comprises a plurality of different modal data processing branches used for respectively processing image data of various different modals, and each branch is formed byconnecting a plurality of residual neural network layers; feature fusion is performed on the image features output by each residual neural network layer of each branch through a multi-modal weight adaptive module; the multi-modal weight self-adaptive module comprises an upper branch and a lower branch, and the upper branch is used for fusing the features of the multi-modal information through an image convolution operation to obtain a fused feature; the lower branch comprises an image convolution operation unit, a global pooling layer, a softmax unit, a ReLU activation unit and a full connection layer.

Methods of making a white matter fibrogram representing the connectome of the brain of a subject, comprising: (a) performing a multispectral multislice magnetic resonance scan on the brain of a subject, (b) storing image data indicative of a plurality of magnetic resonance weightings of each of a plurality of slices of the brain of the subject to provide directly acquired images, (c) processing the directly acquired images to generate a plurality of quantitative maps of the brain indicative of a plurality of qMRI parameters of the subject, (d) constructing a plurality of magnetic resonance images indicative of white matter structure from the quantitative maps, and (e) rendering a white matter fibrogram of the brain of the subject from the plurality of magnetic resonance images.

Existing methods for analyzing person-specific ‘connectomes’ are not computationally equipped for scalable, flexible, and integrated processing across multiple network resolutions and drawing from disparate data modalities-a major roadblock to utilizing ensembles and hierarchies of connectomes to solve person-specific machine-learning problems. The processes implemented in software described herein consists of an end-to-end pipeline for deploying ensembles and hierarchies of network-generating workflows that can utilize multimodal, person-specific data to sample networks, extracted from that data, across a grid of network-defining hyperparameters. In essence, this pipeline enables users to perform ensemble sampling of connectomes for given individual(s) based on any input phenotypic datatype, constructed from any data modality or hierarchy of modalities at any scale, and based on any set of network-defining hyperparameters.

The invention discloses a visual generation method and equipment for a machine learning model. The method comprises the steps: displaying at least one node assembly on a display device, and enabling the node assemblies to be in one-to-one correspondence with operators forming the machine learning model; responding to a dragging instruction of a user for the node component, moving the node component into a view window of the display device, and generating a selected node component; responding to a selection operation of a user, judging a connection mode selected by the user, judging whether theselected node component corresponds to other selected node components or not according to the connection mode, and if so, generating a connection component and connecting at least two corresponding selected node components through the connection component to obtain a process view; and generating a machine learning model according to the process view. By applying the technical scheme of the invention, a user can realize full-process and end-to-end model construction without deeply understanding algorithm principles and technical details, so that the process of code definition modeling is omitted, and the period from model development to application is greatly shortened.

A method for creating a dynamic neural function library that relates to Artificial Intelligence systems and devices is provided. Within a dynamic neural network (artificial intelligent device), a plurality of control values are autonomously generated during a learning process and thus stored in synaptic registers of the artificial intelligent device that represent a training model of a task or a function learned by the artificial intelligent device. Control Values include, but are not limited to, values that indicate the neurotransmitter level that is present in the synapse, the neurotransmitter type, the connectome, the neuromodulator sensitivity, and other synaptic, dendric delay and axonal delay parameters. These values form collectively a training model. Training models are stored in the dynamic neural function library of the artificial intelligent device. The artificial intelligent device copies the function library to an electronic data processing device memory that is reusable to train another artificial intelligent device.

A method of forming a neural network includes specifying layers of neural network neurons. A parameter genome is defined with numerical parameters characterizing connections between neural network neurons in the layers of neural network neurons, where the connections are defined from a neuron in a current layer to neurons in a set of adjacent layers, and where the parameter genome has a unique representation characterized by kilobytes of numerical parameters. Parameter genomes are combined into a connectome characterizing all connections between all neural network neurons in the connectome, where the connectome has in excess of millions of neural network neurons and billions of connections between the neural network neurons.

A system and method for identifying a patient-specific neurosurgery target location is provided. The system receives brain imaging data for a patient that includes tracts and networks in the patient brain, accesses a quantitative connectome atlas comprising population-based, disease-specific structural and functional connectivity maps comprising a pattern of tracts and networks associated with an optimal target area (OTA) identified from a population of patients, and defines the patient-specific neurosurgery target location based on a comparison between a pattern of the tracts and networks from the brain imaging data for the patient and the pattern of tracts and networks associated with the OTA identified from the population of patients in the quantitative connectome atlas. The quantitative connectome atlas comprises a disease-specific, population-based quantitative connectome atlas that identifies an optimal target location for treatment associated with a maximal clinical improvement for each disease in the population of patients.

The invention belongs to the technical field of water environment remote sensing data analysis, and particularly relates to a water environment remote sensing data modeling method based on a multilayer convolutional neural network, and a data model is formed by sequentially connecting an input layer, a training layer and an output layer; the input of the input layer is preprocessed remote sensingimage data; the training layer comprises a convolution layer, a pooling layer and a full connection layer; each of the convolution layer, the pooling layer and the full connection layer is composed ofa plurality of hidden neurons with mutually independent matrixes; the output layer is used for outputting results, the training layer learns and inputs high-level features through layer-by-layer feature extraction of a remote sensing spectral feature curve of remote sensing image data acquired by a preprocessed satellite, and inputs the high-level features into the full connection layer to identify a result; and the large-scale water environment online remote sensing water quality accurate identification and diagnosis system aims to realize large-scale water environment online remote sensingwater quality accurate identification and diagnosis of the three gorges reservoir area so as to provide a reliable and easy-to-use large-scale water environment monitoring and auxiliary decision-making tool.

A method and system are provided for the creation and use of artificial connectomes for the purpose of invoking the sensing of various inputs, processing the input paradigms and causing motor output that can be used to develop contextual evidence of the paradigm being sensed. Encoded sensory data is passed to an artificial connectome, where such connectome is based on the guiding principles of how animal nervous systems are wired and modulated, with the resulting processed data terminating in physical movement or virtual motor output that can be expressed as output and / or provide feedback input into the sensory input data feed. The entire system comprises an emulation of animal nervous systems from sensory input to motor output that can be used for various general intelligence purposes.

The invention discloses a coal-fired unit water wall temperature prediction neural network model, and the model is characterized in that the model is formed by successive connection of a plurality ofneural networks, and each neural network is composed of an input layer, a hidden layer and an output layer; the input layer is divided into a predicted heating surface wall temperature variable input,an upstream heating surface wall temperature variable input and other key variable inputs, key factors influencing the wall temperature and the upstream heating surface wall temperature change condition are considered, and meanwhile the influence of predicted heating surface wall temperature historical data on the input layer is considered; an input variable structure is determined, an input parameter delay coefficient, the number of hidden layers and an activation function are corrected, the model training and generalization precision is improved, the change trend of the predicted wall temperature at different moments is obtained through successive wall temperature prediction, and the better water wall temperature prediction precision is achieved.

The invention relates to the technical field of medical rehabilitation equipment, and particularly discloses a language training device for autistic children based on artificial intelligence. The device comprises a robot body and a movable base, the robot body comprises a lower body section, an upper body section, a robot head and robot arms, the lower body section, the upper body section and the robot head are sequentially connected from bottom to top, and the two robot arms are arranged on the two sides of the upper body section in a mirror symmetry mode. The language training device for the autistic children based on the artificial intelligence can prevent the autistic children from making contact with obstacles in the driving process, and is novel in structural design; and meanwhile, the device has a good interest cultivation function while performing language training on the autistic children, has a good treatment effect on language training of the autistic children, and is diversified in function, high in intelligent degree and excellent in use effect.

The invention discloses a fast relation extraction method based on a convolutional neural network and improved cascade labeling. The method comprises the following steps: firstly, encoding an initial text by a text encoder based on a deep neural network formed by connecting expansion convolution, a gating unit and residual errors to obtain a text encoding representation with rich context semantics; then, according to the obtained text codes, marking spans of all head entities and corresponding entity types by adopting improved cascade marking and a head entity marking device; then, through text coding representation and feature representation of the head entities, a tail entity annotator annotates all tail entities corresponding to each head entity; and finally, verifying through a relation extraction task in the real world. The method has the advantages of rapid training and prediction, and can meet the requirements of relation extraction scenes oriented to massive texts.

The invention discloses a visual tracking failure detection system based on a neural network and a training method thereof, and belongs to the technical field of visual tracking. The method comprisesthe steps: establishing the visual tracking failure detection system based on the neural network, wherein the system is formed by connecting a related filtering module and a tracking anomaly sensing module in series; according to the visual tracking failure detection system, judging whether target tracking fails or not according to a result graph generated by a related filter by utilizing the strong visual perception capability of a deep neural network; and enabling the correlation filtering module to perform model parameter updating according to a result of the tracking exception sensing module. In view of the fact that the neural network method has good classification precision but needs a large number of samples for training, the training needs a large number of samples including positive samples and negative samples, a corresponding large-scale training sample generation method is designed, and the method is mainly used for training of a deep neural network model. And testing is carried out on the public data set. The method can support training of the deep neural network.

The invention discloses an automatic modulation classification method based on an improved stacked hourglass neural network. The method comprises the following steps: obtaining a modulation signal as original data, and carrying out normalization processing on the original data; the method comprises the following steps of: acquiring feature information of a modulation signal by adopting two convolution kernels with different shapes, and connecting the two acquired convolution features in a channel dimension to form multi-local feature information; receiving multi-local feature information and adopting an initial convolution module to increase the number of feature channels; the multi-local feature information with the increased number of feature channels is subjected to end-to-end separation in sequence by adopting four-stage hourglass module stacking; wherein each hourglass module takes the bottleneck layer as a basic unit, the channel dimension change is carried out in the bottleneck layer, and each hourglass module filters the channel by adopting a channel attention mechanism in the down-sampling stage and the up-sampling stage. The method is improved on the basis of the baseline network of the stacked hourglass neural network, and the modulation recognition accuracy can be remarkably improved.

The invention relates to an automatic simulation method for an ethylene glycol regeneration and recovery system. According to the method, an intelligent agent module is set to achieve the real-time input, control and result feedback of an ethylene glycol regeneration and recovery process; particularly, a plurality of intelligent agent modules can be mutually connected and networked to form a distributed control network to realize standardized transmission between data, so that full-dimensional simulation of the ethylene glycol regeneration and recovery process and an automatic control system thereof is realized; and in combination with a digital twin platform, a user can discuss and analyze different control schemes and process conditions to the maximum degree of freedom, so that time is saved, and a large amount of funds and operation cost are also saved.

The invention relates to the field of data modeling, and provides a big data task draggable modeling method and system, a storage medium and terminal device. The method comprises the steps of determining the task type of a target task, and uploading source data of the target task; determining a processing flow of the target task based on dragging tracks of the functional components and a line connection sequence among the functional components, and determining an analyzer of the target task according to the task type; and triggering the analyzer to analyze a processing flow of the target task, and processing the source data according to the processing flow to obtain a processing result of the target task. The modeling process is realized by dragging and connecting the components, the big data modeling task is simplified, and the invention is simple to operate and easy to master.

The invention discloses an electroencephalogram activity simulation method and system based on a coupled neuron group. The method comprises the following steps that electroencephalogram signals are preprocessed; calculating Pearson's correlation coefficients among the channel signals to represent synchronous connection degrees among the channel signals, and forming a matrix by the calculated Pearson's correlation coefficients; constructing a coupled neuron group model, and mutually connecting excitability, slow-speed inhibition and fast-speed inhibition intermediate neuron sub-clusters with different parameters to form the neuron group model; by removing different neuron groups to simulate results of necrosis of different regions of the brain and comparing electroencephalogram signals output by the remaining neuron groups and high-frequency energy value changes of all channels, the influence degree of necrosis of different regions of the brain on high-frequency electroencephalogram activity is judged, and therefore the region with the maximum influence on the high-frequency electroencephalogram activity is determined. According to the invention, more stereoscopic and deeper electrophysiological information of the brain can be explored, and the time resolution is higher.

The invention discloses a convolutional neural network training method and device, an image reconstruction method and device and a storage medium. The convolutional neural network comprises a featureextraction module, a nonlinear mapping module and an image reconstruction module, and the nonlinear mapping module comprises a plurality of connection groups and a first weighted channel cascade unit;the first weighted channel cascade unit carries out weighted integration on the information output by each connection group and then outputs the information, each connection group comprises a plurality of connection units and a second weighted channel cascade unit, and the second weighted channel cascade unit carries out weighted integration on the information input into the first connection unitand the information output by the last connection unit and then outputs the information. According to the method, all jump connections in a network level are in a weighted channel cascading mode, different weights are given to feature map channels, the difference between the channels is increased, integration of local features of a latter convolutional layer is realized, and the representation capability of the convolutional neural network is improved. The method is widely applied to the technical field of image processing.

A method of forming a neural network includes specifying layers of neural network neurons. A parameter genome is defined with numerical parameters characterizing connections between neural network neurons in the layers of neural network neurons, where the connections are defined from a neuron in a current layer to neurons in a set of adjacent layers, and where the parameter genome has a unique representation characterized by kilobytes of numerical parameters. Parameter genomes are combined into a connectome characterizing all connections between all neural network neurons in the connectome, where the connectome has in excess of millions of neural network neurons and billions of connections between the neural network neurons.

The invention relates to an enterprise factor group representation method based on a human brain neural network model and a system thereof. The method comprises steps: cerebral cortex partition, connection group and neuron features are built; the cerebral cortex partition, connection group and neuron features are adopted to represent refined factors of an enterprise factor group; the relationshipbetween the connection group and the cerebral cortex partition and the neuron in the enterprise factor group is analyzed; and the refined factors of the enterprise factor group represented by the cerebral cortex partition, connection group and neuron features and the relationship are integrated to form an enterprise factor group expression. With a neuron representing an enterprise decision factor,enterprise data are decomposed in a finest grain mode, the enterprise information data acquisition has atomicity, the data redundancy is reduced, the data have higher real-time performance, relevanceand easy scalability, the team capability, the resource capability and the service capability of traditional or emerging enterprise teams are flexibly evaluated, and the evaluation on the enterpriseby an enterprise service mechanism or organization is more comprehensive and rigorous.

The invention discloses a virtual reality scene, an interaction method thereof and terminal equipment, wherein the virtual reality scene comprises a virtual environment intelligent member, a virtual role intelligent member and a semantic path processing unit. The semantic path processing unit is used for constructing a semantic path which is a track that is composed of directed connections betweennodes on a geometric pattern of the virtual environment intelligent member, the node information at least comprises node position information, node behavior semantic information and node environmentsemantic information. The virtual rule intelligent member is used for acquiring the semantic path and moving and executing a target action according to a self task and the semantic path. The virtual environment intelligent member is used for obtaining action result information according to the information of the target action of the node of the virtual role intelligent member in the semantic path,and indicating the semantic path processing unit to update the node environment semantic information of the semantic path. The virtual reality scene can effectively realize interaction in the virtualreality scene.