12 results about "Synaptic junction" patented technology

The present invention relates to the technical field of neural stimulation, and specifically relates to a non-invasive spinal cord stimulator for upper limb motor function recovery, comprising a cervical spinal cord pulse control unit, an electrode connection unit, a motion monitoring unit, a wireless communication unit, a speech unit, a paradigm decision unit, an electrode array unit, and an artificial intelligence unit. In the present invention, the non-invasive spinal cord stimulator acts on a specific cervical spinal cord segment by means of non-invasive cervical spinal cord stimulation technology, inducing the reconstruction of a spinal cord neural network, strengthening synaptic connection, treating upper limb motor dysfunction of paralyzed individuals, and achieving upper limb motor function recovery.

The application relates to a thymoma epithelial cell subpopulation with neuromuscular characteristics and application, and a thymoma epithelial cell subpopulation with neuromuscular characteristics is obtained through single clone dilution culture screening from a thymoma cell line Thy0517 of a patient with myasthenia gravis (MG) in combination, and a thymoma epithelial cell with neuromuscular characteristics in the cell subpopulation is named as Thymus_NMi; the cell subpopulation has synapse-like structures and neuromuscular adhesion characteristics, and efficiently expresses genes participating in neural cell adhesion and synapse connection, highly integrates neuromuscular double characteristics, and simulates key pathological characteristics of abnormal thymus-induced immune tolerance of MG patients in a molecular phenotype and physiological function, so that a cell model closest to a real clinical state is provided for exploring a myasthenia gravis occurrence mechanism.

This invention discloses a pruning method, apparatus, and electronic device for spiking neural networks. The method includes: initializing an initial vector composed of the weights of each connection in the synaptic connection layer to obtain a weight vector; when pruning the spiking neural network using a backpropagation-based algorithm, calculating the gradient of each loss function value with respect to the hidden parameter vector using a predefined derivative function; updating the gradient of the hidden parameter vector using gradient descent, and calculating a target threshold for subsequent gradient updates using a preset incrementing function; based on the target threshold, mapping the hidden parameter vector back to the weight vector using the soft threshold function; and obtaining a trained spiking neural network model when the number of pruning training rounds reaches a preset number. This invention solves the technical problem of effectively deploying spiking neural networks on neuromorphic computing chips in related technologies.

Disclosed herein are compounds, compositions, and methods for promoting neuronal growth and / or improving neuronal structure with the compounds and compositions disclosed herein. Also described are methods of treating diseases or disorders that are mediated by the loss of synaptic connectivity and / or plasticity, such as neurological diseases and disorders, with fused pyrrolidine psychoplastogens.

The application discloses a synapse connection relationship and weight storage method and storage medium for a brain-like processor, and the method comprises the following steps: S1, storage space division and address mapping; the logical storage is divided into a direct index area and a multi-connection area to eliminate source ID redundancy and realize 0 or 1 direct addressing; S2, data format and minimal Hint coding; a 2-bit Hint code is used to uniformly identify a storage state, distinguish storage types and replace a pointer array; S3, adaptive access control; single / multi-connection access logic is adaptively executed according to the Hint code to realize single-connection zero jump, multi-connection continuous traversal and minimization of memory access times. The storage medium is realized based on the above method. The application has the advantages of high compression ratio, low memory access overhead, high scalability and the like.

The application relates to the technical field of brain science image processing, and discloses a brain tissue neuron morphology three-dimensional reconstruction and connection group analysis method and system. The method comprises the following steps: feature point extraction and matching and optical flow registration are performed on continuous ultrathin section electron microscope images to obtain three-dimensional image body data; a three-dimensional deep convolutional neural network is used for semantic segmentation of cell bodies, dendrites, axons and dendritic spines, and secondary registration is driven through confidence feedback; synapses are detected and classified into excitatory or inhibitory types; topological perception skeletonization reconstruction is performed, and segmentation correction is constrained through topological anomaly feedback; and a synapse connection matrix is constructed, and graph theory analysis is performed.

A method for constructing a clinically-relevant large animal model of traumatic optic neuropathy, employing a goat with orbital anatomy and optic nerve structure similar to humans as a model animal, overing the defects existing in other animals as TON model animals. The target model goat is obtained by screening parameters, such as a maximum transverse diameter of a coronal plane of a body of sphenoid bone and a wall width of a sphenoid bone, thereby facilitating the exposure of an optic nerve in a optic-canal segment under a nasal endoscopic surgery and quantitative injury. With the optic nerve injury animal model, the clinical TON pathogenesis can be explored by studying the optic nerve at the injured optic canal. In the prepared TON animal model, the injured optic nerve is closer to the brain, which is beneficial to rebuilding the synaptic connection.

The application belongs to the field of pulse neural network and image reconstruction, and discloses an image reconstruction method and system based on a pulse memory autoencoder. The pulse memory autoencoder comprises an encoder, a memory network and a decoder. An input image is obtained, the input image is converted into a pulse sequence by using delay coding, and the pulse sequence is input into the encoder to obtain corresponding pulse features. The pulse features are input into the memory network to obtain corresponding memory features. In the recall stage, the pulse features are only transmitted to the perception neuron, the response neuron is activated by the synaptic connection learned by the perception neuron in the memory stage, and the memory features are output to the decoder to generate a corresponding reconstructed image. The application can enhance the maintenance capability of the potential representation and complete high-quality image reconstruction while maintaining a low pulse overhead.

The application discloses a kind of based on the method for training of increment learning neural network of multisynaptic connection and local plasticity modulation, main steps include: constructing neural network based on multisynaptic connection, include fully connected layer and convolution layer: in fully connected layer, introduce multiple synapse connections between adjacent two layers of neurons, in convolution layer, design multiple parallel weight channels for each convolution kernel element;A eligibility trace is associated for each synapse weight, for recording local synapse activity intensity, and network parameters are initialized;Each task is allocated subnetwork by weight mask, eligibility trace is updated according to the output value of neuron, the model is trained using back propagation algorithm, and the allocated weight is frozen;According to the value of synapse modulation factor calculated according to eligibility trace intensity, update weight in combination with modulation factor;The application can effectively improve the accuracy of increment learning, realize zero forgetting, and keep the robustness to task order change.

The application discloses a pulse neural network image recognition method, system, device and storage medium, which is applied to the technical field of image recognition and has the method comprising the following steps: encoding image pixel values into pulse sequence and inputting the pulse sequence into a pulse neural network; updating the input synapse weight of a hidden layer neuron according to the mutual information between the input pulse and the output pulse of the hidden layer neuron; updating the input synapse weight of an output layer neuron according to the difference between the pulse response of the output layer neuron and a preset task target; transmitting the original pulse of the output layer neuron to the hidden layer neuron through a feedback synapse connection; iteratively training the pulse neural network until a preset convergence condition is met; and constructing a multi-layer pulse neural network architecture, wherein each neuron of the hidden layer independently updates the weight according to the mutual information gradient, and through the constraint modulation mechanism of the interlayer feedback signal, the network as a whole can emerge coordinated learning behaviors facing the classification target.

This invention discloses a method for monitoring catering fume emissions based on edge computing, comprising the following steps: collecting catering kitchen fume sensor data, operational behavior data, and environmental parameter data; generating fume emission pulse sequences and operational stimulus pulse sequences; establishing propagation correlations to form a fume emission synaptic network; generating emission behavior neural coding sequences based on the relationship between the operational stimulus pulse sequences and the fume emission pulse sequences; marking synaptic connections as synaptic locked states when stability criteria are met; performing propagation calculations to obtain counterfactual pulse propagation paths and identifying key pollution triggering behaviors; updating synaptic connection weights to obtain a dynamic synaptic structure; and combining key pollution triggering behaviors with the dynamic synaptic structure to generate abnormal fume emission warning information. This invention utilizes counterfactual propagation and dynamic synaptic control methods to achieve causal identification of fume emissions, possessing advantages of high accuracy and robustness.