19 results about "Neuron membrane" patented technology

The invention discloses a pulse neural network lightweight method based on nonlinear pruning, and relates to the technical field of artificial intelligence and neural networks. The method comprises the following steps: S1, constructing an NDI-LIF spiking neural network model, and introducing a bilinear product term into an input current in a dynamic updating process of a neuron membrane potential; s2, constructing a nonlinear synaptic pruning mechanism, representing a connection weight as a re-parameterization function composed of a re-parameterization weight and a conversion gain coefficient, and pruning the connection weight based on the re-parameterization function; and S3, training the NSPDI-SNN model to obtain a lightweight pulse neural network. By introducing a nonlinear dendritic integration mechanism and a state-adjustable synaptic pruning mechanism, balance between network sparsity and high performance is realized while the expression ability of the model is enhanced, and a lightweight spiking neural network model with high spatial-temporal expression ability, highly sparse structure and reasonable biological mechanism is constructed. And the generalization ability and the efficient training performance under various tasks are ensured.

The invention relates to an acupuncture brain machine interface system based on brain-like recursive network model decoding. The system comprises an electroencephalogram acquisition module; an electroencephalogram preprocessing module; the needling parameter fitting module is used for accurately fitting needle body motion parameters through a polynomial regression method; the state space model is used for accurately simulating the nonlinear change characteristic of the neuron membrane potential according to the needle body motion parameters output by the acupuncture parameter fitting module; the brain-like recursive network model is used for depicting time dependence of neural activities of a single brain region under the acupuncture action and correlation of neural activities among different brain regions; a Kalman filtering generator; and the neural decoder is used for performing behavior decoding on the optimal internal state output by the brain-like recursive network model updated by the Kalman filtering generator. The system integrates electroencephalogram signal acquisition, acupuncture parameter acquisition, brain nerve activity modeling and acupuncture manipulation decoding, can perform real-time online operation, and is high in time resolution and high in response speed.

The invention discloses a narrow space automatic driving path planning method based on artificial intelligence, and the method comprises the following steps: S1, collecting laser radar point cloud and image data, executing semantic segmentation and clustering, and building a topological graph; s2, mapping topological graph nodes into pulse coupling neural network neurons, and constructing a double-domain coupling mechanism comprising a structure coupling item and a state coupling item; s3, generating a path direction vector, constructing a path intention gating factor, and dynamically adjusting a coupling input value; s4, periodically updating the neuron membrane potential, and adaptively adjusting the discharge threshold; s5, recording a first discharge node number and time, and generating a path activation sequence according to a discharge sequence; s6, performing connectivity check and curvature smoothing on the path activation sequence to generate a continuous track; and S7, converting the trajectory into a control instruction, and driving the vehicle to execute path tracking. According to the invention, high-precision, high-connectivity and high-control-stability output of path planning in a narrow space is realized.

The invention provides a photovoltaic cleaning robot path planning method and system based on the pollution degree, and relates to the field of robot path planning. The method comprises the following steps: collecting a reflective rate change event flow on the surface of a photovoltaic module, and generating an asynchronous data packet containing event coordinates, timestamps and polarities; and converting the asynchronous data packet into a dynamic pollution level diagram, wherein the pollution level numerical value is in positive correlation with the event occurrence frequency in unit time. A distributed dynamic pollution map is constructed by using a spiking neural network, and pollution diffusion and cleaning recovery processes are simulated through neuronal membrane potential evolution. And establishing an action selection competition mechanism in the pulse neural network, and generating a cleaning path instruction according to the spatial and temporal distribution characteristics of the dynamic pollution map. And executing a cleaning action and feeding back an environment change event, and dynamically updating a dynamic pollution map and an action selection competition mechanism. The technical problems of high energy consumption, slow response to sudden pollution and dynamic environment, poor adaptability and weak anti-interference capability are solved.

The invention relates to the field of artificial intelligence and the field of integrated circuits, in particular to a chip for in-memory calculation of a spiking neural network, which comprises a synaptic weight storage array for storing synaptic weights; the threshold storage array is used for storing a neuron threshold voltage; the membrane potential storage array is used for storing a neuron membrane potential, a bit line of the membrane potential storage array is connected with a bit line of the synaptic weight storage array, and logic calculation results of the synaptic weight and the neuron membrane potential are directly generated on the connected bit lines; the membrane potential accumulation unit obtains a membrane potential accumulation result through processing of an addition logic circuit; the comparison unit is used for opening read sub-lines of the membrane potential storage array and the threshold voltage storage array and multiplexing an addition logic circuit to realize subtraction calculation of a membrane potential accumulation result and a threshold voltage; and the transmitting unit is used for judging whether the neurons output pulses or not and resetting the membrane potential. The method can achieve the push-training integration of in-memory calculation, remarkably improves the SNN operation energy efficiency, and reduces the power consumption.

This invention provides a method, system, electronic device, and storage medium for analyzing the energy consumption of neural networks based on ion channel regulation. It constructs a neural network dynamics model containing excitatory and inhibitory neurons. The neuronal membrane potential dynamics equation is defined by a set of differential equations for various voltage-gated ion currents (fast sodium ion current, continuous sodium ion current, slow potassium ion current, etc.) and leakage current. By adjusting the unit area conductivity parameter and / or inactivation time constant parameter of the voltage-gated ion channels in the model, the external intervention effect is simulated. After obtaining membrane potential sequence data through simulation, the total energy consumption, synaptic energy consumption ratio, average discharge rate, and average synchronization rate between neurons of the model are calculated. Finally, a quantitative correlation model between parameter changes and the above indicators is established, clarifying the coupling effect of the target ion channel dynamics characteristics on the neural network discharge behavior pattern and energy metabolism.

This invention provides a multimodal human-machine interaction chip based on an adaptive threshold spiking neural network, comprising: a sensor module for real-time acquisition of input data from three modalities: visual signals, pressure signals, and surface electromyography (sEMG) signals; an acquisition module for analog-to-digital conversion and preprocessing of pressure signals and sEMG signals to generate feature tensors; and a recognition module comprising three adaptive threshold spiking neural networks (CNNs) that process the feature tensors of the three modalities respectively, perform cross-modal feature fusion, and output the robot's motion intention. The adaptive threshold spiking neural network dynamically adjusts the membrane potential threshold of the spiking neurons, reducing power consumption while maintaining computational accuracy. This invention, by combining the input data from the three modalities with the adaptive threshold spiking neural network algorithm for feature fusion, can comprehensively capture the operator's motion intention and the environmental interaction state, significantly improving the accuracy and adaptability of robot motion generation and enhancing the naturalness and reliability of human-machine collaboration.

The application discloses a kind of neuron membrane potential data downsampling methods, applied to pulse neuron simulation, comprising: the original acquisition membrane potential data is divided into bucket;Filtering local strict extreme point in bucket;If there is no local strict extreme point in bucket, using the first and last membrane potential data in bucket represent all the membrane potential data in bucket;Otherwise, using the first membrane potential data, the last membrane potential data and all or part of local strict extreme points represent all the membrane potential data in bucket;Collect representative data in all buckets to form the final downsampling data.The application realizes the premise of effectively reducing the total amount of membrane potential data to the greatest extent preserving the variation characteristics of membrane potential data.

The invention discloses a pulse Jaccard attention and twin network-based electrocardiogram classification method, which comprises the following steps of: acquiring an original electrocardiogram signal, performing cascade filtering on the original electrocardiogram signal, mapping the filtered signal to a dynamic range matched with an LIF neuron membrane potential threshold value by utilizing a self-adaptive threshold value normalization algorithm, obtaining a normalized signal, inputting the normalized signal into a pulse encoder, and outputting the normalized signal into a twin network; a pulse sequence is obtained, and the pulse sequence is input into the twinborn double-branch pulse neural network with the shared weight; based on the output of the twinborn double-branch pulse neural network, calculating the total loss by utilizing a joint loss function, updating network parameters through back propagation, obtaining the trained twinborn double-branch pulse neural network, performing anomaly classification on the real-time electrocardiogram signal, and outputting an anomaly result. According to the method, the event-driven characteristic of the spiking neural network, the high discrimination ability of the Jaccard attention mechanism and the small sample learning ability of the twin network are utilized, so that the robustness of the model in a small sample scene is enhanced.

The invention provides a dynamic object detection method and device based on a pulse neural network model, equipment and a medium, and the method comprises the following steps: 1, capturing an object of a target scene through a dynamic visual sensor, constructing a neuromorphic data set, and determining the neuromorphic data of the target scene based on the neuromorphic data set; according to the method, parameterized leakage integration distribution neurons are introduced into a spiking neural network model, synaptic and membrane related parameters are synchronously optimized in a training process, membrane time constant differences of spiking neurons in different brain regions are fully considered, neuron heterogeneity is enhanced, and network expressive force is improved; meanwhile, by means of a multi-scale attention feature fusion module, time, channel and space attention mechanisms are organically combined in a layered aggregation mode, multi-dimensional and multi-scale features are aggregated, the limitation that attention features are processed separately in the prior art is changed, the model focuses on key information of a dynamic object better, and high-precision real-time detection of the dynamic object in a complex environment is achieved.

The application provides a kind of biomimetic neuron memristor and preparation method thereof, it is related to biomimetic neuron technical field, including substrate, parallel electrode and dielectric layer;The parallel electrode and the dielectric layer are all arranged on the substrate;The parallel electrode includes positive electrode layer and negative electrode layer, and the dielectric layer is arranged between the positive electrode layer and the negative electrode layer;The dielectric layer uses semiconductor material containing low activation energy ions in lattice;There are two kinds of low activation energy ions in the lattice in the dielectric layer, and under the driving of electric field, two kinds of low activation energy ions are regulated by electric field, and move to positive and negative two poles respectively, for simulating the membrane potential change process generated by the change of sodium and potassium ion transport in and out of membrane when neuron is stimulated.The application can realize the simulation of neuron membrane potential change, neuron cumulative emission phenomenon and neuron refractory period behavior without building peripheral circuit.

This invention relates to a low-power neuron circuit supporting multiple encoding methods. The neuron circuit includes a two-stage sensitive amplifier structure and corresponding logic control circuitry. Specifically, the input pulse signal generates an enable signal for the first-stage sensitive amplifier through delay control. The first-stage sensitive amplifier compares the neuron membrane capacitance potential Vmem with the first-stage threshold voltage Vrefs to determine whether to generate an enable signal for the second-stage sensitive amplifier. The output of the second-stage sensitive amplifier generates an output pulse signal through an SR latch. Further, the output pulse signal controls the resetting of the neuron based on the current input configuration signal. Compared with existing technologies, this invention has the advantages of low power consumption and support for multiple encoding methods.

The application discloses a kind of narrow space automatic driving path planning method based on artificial intelligence, comprising the following steps: S1, laser radar point cloud and image data are collected, semantic segmentation and clustering are executed, and topological graph is established;S2, topological graph node is mapped as impulse coupling neural network neuron, and the double-domain coupling mechanism including structure coupling term and state coupling term is constructed;S3, path direction vector is generated, path intention gating factor is structured, and coupling input value is dynamically adjusted;S4, neuron membrane potential is periodically updated, and discharge threshold is adaptively adjusted;S5, the first discharge node number and time are recorded, and path activation sequence is generated according to discharge order;S6, connectivity check and curvature smoothing are performed on path activation sequence, and continuous trajectory is generated;S7, trajectory is converted into control command, and vehicle is driven to execute path tracking.The application realizes the high-precision, high-connectivity and high-control stability output of path planning in narrow space.

The invention discloses a closed-loop transcranial stimulation system and a closed-loop transcranial stimulation method. 0.3 mg of scopolamine and 3 mg of midazolam are intramuscularly injected before an operation; after a patient enters a room, sufentanil is subjected to intravenous injection induced by right subclavian vein and radial artery puncture general anesthesia, propofol is subjected to target-control infusion, the plasma target concentration is 3.0 mu g / mL, vecuronium bromide is 0.15 mg / kg, tracheal intubation is performed after 5 min, general anesthesia maintains target-control infusion of propofol, and after the patient enters the room, the target-control infusion of propofol is performed. A TMS stimulation coil electricity-magnetism-heat-stress model, a single-phase and double-phase pulse discharge circuit model and a neuron dynamic response model under the action of a TMS are established, and the influence of TMS system parameters on space-time distribution characteristics of an intracranial induced electric field and neuron membrane potential can be accurately expressed. Behavior perception and system auditory stimulation are combined to be applied to nursing intervention of severe craniocerebral injury coma patients, self-repairing and consciousness awakening of the brain functions of the patients can be effectively promoted, and meanwhile recovery of limb functions and daily life ability of the patients is facilitated.

The invention discloses a POS machine data remote management method and system, and relates to the technical field of remote risk control, and the method comprises the steps: a plurality of POS terminals collect original transaction data streams in real time, and transmit the original transaction data streams to a cloud server through a remote network; inputting the ciphertext data packet into a gravitation lens big data analysis engine, and outputting a risk probability value by calculating the trajectory offset of the ciphertext data packet in a multi-dimensional probability field formed by a gravitation source; and when the risk probability value exceeds a preset warning threshold value, the cloud server sends an interception instruction to the POS terminal initiating the current transaction and the associated management terminal, terminates the current transaction and activates the alarm. According to the method, a structured transaction data stream is input into a neuromorphic stream processor, the transaction amount change rate and the neuron membrane potential are subjected to nonlinear coupling, and millisecond-level feature compression is triggered when it is detected that the continuous amount change rate exceeds the standard.

The invention discloses a convolutional spiking neural network recognition system and method for high-dynamic single-photon imaging, and the system comprises a photon event compression module which carries out the sparse processing and event flow reconstruction of a continuous photon trigger sequence in a high-dynamic range, so as to reduce the data bandwidth and highlight the effective photon trigger information; the counting pulse time conversion module is used for mapping the compressed photon event sequence to a time domain feature so as to realize unified expression of high photon flow and low photon flow conditions; the common-mode noise suppression module is used for filtering noise and intercepting effective data in a time domain; the pipelined convolution pulse operation module is used for performing efficient feature extraction on the preprocessed photon event data; and the storage module is used for storing the convolution weight and the neuron membrane potential required in the convolution operation module. According to the invention, real-time feature extraction with low resource consumption can be realized under a high-dynamic photon event condition, and SPAD image processing and intelligent identification performance can be improved.

To provide a method capable of non-invasively visualizing and recording membrane potentials of neuronal populations and individual neurons in zebrafish in real time and over a long period of time.SOLUTION: A method involves observing embryos obtained by crossing zebrafish individuals which express ArcLight at high brightness selected from an ArcLight gene-introduced zebrafish population. These embryos are reared at 23°C under either a 14-hour light / 10-hour dark cycle or continuous darkness for 24 hours, and 30 minutes prior to sample preparation on the day of recording, the rearing temperature is changed to 28.5°C. The embryos are treated with a muscle relaxant to prevent movement during observation, mounted in low-melting-point agarose, and membrane potentials are observed and recorded using a fluorescence microscope.SELECTED DRAWING: Figure 6

The invention belongs to the technical field of biological medicine, and discloses a preparation method and application of a composition for improving memory. The composition is prepared from 10 to 50 parts of sheep brain peptide, 20 to 60 parts of egg yolk phospholipoprotein, 1 to 10 parts of N-acetylneuraminic acid, 5 to 25 parts of tea leaf theanine and 2 to 15 parts of rhizoma curcumae longae. By constructing a peptide-phospholipid composite delivery system, the molecular delivery efficiency, neuronal membrane structure support, synaptic plasticity and neuroinflammation regulation are synergistically enhanced. The invention provides the memory improving composition with clear action mechanism, efficient synergistic effect and good safety, and the preparation method is controllable in process, stable in quality and suitable for large-scale industrial production, can be applied to preparation of functional food for improving learning and memory ability, formula food for special medical purposes or dietary supplements, and can be used for preparing functional food for improving learning and memory ability. And the method has obvious social value and industrialization prospect.