3109 results about "Neuron" patented technology

The invention relates to the technical field of data protection, and discloses a privacy protection type data joint modeling method based on federated learning, which comprises the following steps: acquiring local data to perform meta-feature extraction, calculating key statistics to characterize data characteristics, collecting meta-features, grouping the meta-features into similar feature clusters through spectral clusters, and carrying out feature clustering on the similar feature clusters; dynamically allocating and calculating resource weights according to the similar characteristic cluster scale and the equipment computing power; distributing a basic privacy budget according to the client type, calculating a local model accuracy rate and an intra-cluster level difference, dynamically adjusting the privacy budget, adding adaptive Gaussian noise based on the privacy budget, and adjusting gradient sensitivity of gradient calculation; verifying gradient compliance through zero knowledge, carrying out safe aggregation on gradients passing verification, optimizing a meta-model through a knowledge distillation loss function, and generating confrontation sample analysis to obtain a leakage risk value to identify knowledge leakage risks; sensitive neurons in the neuron sensitivity positioning element model are analyzed and calculated, directional noise is injected, and initial parameters are adjusted for initialization training.

The invention provides an end-side multi-modal large model accelerated reasoning method and system, and the method comprises the steps: carrying out the two-stage screening and rearrangement of visual tokens based on the CLS attention and text-to-visual attention in a visual encoder and pre-filling stage, and constructing a sparse attention and sparse key value cache; in a decoding stage, an important neuron set is judged according to activation gating or historical statistics, only a corresponding feedforward network weight is pulled and calculated, missed weights are loaded on demand through asynchronous I / O, and hot neurons are maintained in a high-speed memory to utilize model sparsity, so that video memory / memory occupancy and calculation overhead are remarkably reduced on an end side; throughput and time delay performance are improved. According to the method, the internal memory and computing resources required by reasoning of the multi-modal large language model are reduced from two dimensions by utilizing the endogenous sparsity of the end-side large language model in input and the model, so that a higher reasoning speed is achieved by utilizing fewer resources on the premise of keeping the size of the model unchanged, and the performance of the whole system is improved.

The invention discloses a rural power grid user power load prediction method and system based on a liquid neural network. The method comprises the following steps: firstly, collecting rural power grid user power load historical data including multi-dimensional features such as weather and agricultural modes, and carrying out data preprocessing; then constructing liquid neurons based on biological neuron dynamics, and modeling the state of the liquid neurons through a differential equation; thirdly, constructing a liquid neural network based on liquid neurons, improving the characterization capability of multi-scale time sequence data through multi-level time constant setting and time gating residual connection, and supplementing network initial information in combination with a multi-layer perceptron architecture; completing model training by using the time sequence data set; the actual application performance of the model is tested based on the test data and the actual application scene; and finally, deploying the model to practical application, and carrying out power load prediction on rural power grid users. According to the method, the expression capability of the model for the multi-scale time sequence data is improved, and high-precision rural power grid user power load prediction is realized.

The invention discloses a federated learning backdoor defense method based on pruning and fine tuning in the technical field of artificial intelligence and network security, the method realizes defense through two core mechanisms of dynamic pruning and gradient constraint fine tuning, and the method comprises the following steps: firstly, calculating a sensitivity score based on a neuron activation frequency and a weight outlier degree; dynamically identifying and cutting redundant neurons utilized by a backdoor, and blocking an abnormal activation path; secondly, gradient direction consistency detection and amplitude constraint are introduced in the fine tuning stage, and a malicious client is inhibited from reconstructing a back door through an abnormal gradient; the server continuously purifies model parameters and enhances robustness by cyclically executing pruning, fine tuning and aggregation operations; the method does not need to depend on an extra clean data set, strictly follows a federated learning privacy protection principle, reduces communication overhead through lightweight pruning, maintains main task performance in combination with gradient constraint, is suitable for a federated learning scene in which edge equipment participates, and effectively balances a defense effect and model stability.

A system and method for real-time time series forecasting using a compound large codeword model with integrated supervisory neurons. The system processes diverse inputs through adaptive codebook generation and codeword allocation. A projection network fuses different data types, creating unified representations for a latent transformer-based machine learning core. The core contains local neural network regions of interconnected operational neurons, monitored by supervisory neurons. These supervisory neurons receive activation data from operational neurons, perform real-time statistical analysis, determine necessary structural modifications, and initiate their implementation during operation. This architecture enables efficient handling of multi-modal data, capturing complex relationships between different input types. The combination of adaptive codebook generation and the supervisory neuron system ensures responsiveness to evolving data patterns and task requirements. This approach provides more accurate and timely forecasts by leveraging diverse data types in a sophisticated, integrated manner, while continuously adapting its structure to maintain optimal performance.

The invention provides a transcranial stimulation magnetic therapy closed-loop control method and system, and the method comprises the steps: determining a multi-modal physiological signal of a user, the multi-modal physiological signal at least comprising an electroencephalogram signal, an autonomic nerve physiological signal and a subjective feedback signal; constructing an individual dynamic baseline spectrum based on user multi-state data collected before treatment; the method comprises the following steps: preprocessing a multi-modal physiological signal collected in a treatment process, and extracting a real-time feature vector; calculating a comprehensive deviation index of the real-time feature vector and the individual dynamic baseline spectrum, and determining a deviation type based on the comprehensive deviation index; the adjustment amount of transcranial magnetic stimulation parameters is dynamically calculated according to the deviation type, the individual response characteristics and the historical treatment data, stimulation operation is conducted according to the adjustment amount of the transcranial magnetic stimulation parameters, and the transcranial magnetic stimulation parameters at least comprise the stimulation frequency, the intensity, the pulse width and the stimulation duration. Parameters of transcranial magnetic stimulation are adaptively changed according to changes of neuron oscillation of a patient in the treatment process.

The invention discloses a closed-loop noninvasive spinal cord electrical stimulation regulation and control method and system based on motion data, and the system is provided with a flexible electrode array patch which is attached to the skin surface of a to-be-tested human body; the sensing module is used for collecting electromyographic signals or acceleration signals of a user; the signals are processed, and characteristic values related to actions and motions are extracted; the closed-loop feedback module compares the characteristic value with a model set characteristic / threshold value, and judges whether to start stimulation site and stimulation parameter adjustment or not; automatically selecting an optimal electrode unit and a corresponding stimulation parameter based on a judgment result; the dynamic regulation and control module controls the optimal electrode unit to output stimulation current, and regulates the activity of corresponding spinal cord neurons; and continuously performing closed-loop adjustment on the electrode unit and the stimulation parameters according to the motion feedback data. According to the method and system, precise stimulation and closed-loop regulation and control based on the motion response of the patient are achieved, and the problems that stimulation target positioning depends on artificial experience, parameter adjustment is not timely, and individualized and closed-loop feedback capacity is lacked are solved.

The invention is applied to the field of mechanical fault diagnosis signal processing, and particularly provides a pulse neural network mechanical fault diagnosis method based on STFT dimension transformation, and the method comprises the steps: collecting a one-dimensional mechanical vibration signal, carrying out the wavelet decomposition, carrying out the wavelet reconstruction of a low-frequency component and a denoised high-frequency component, and carrying out the wavelet reconstruction of the low-frequency component and the denoised high-frequency component; obtaining a denoised one-dimensional vibration signal; performing short-time Fourier transform, and converting the time-frequency two-dimensional matrix into a time-frequency two-dimensional matrix; inputting the time-frequency two-dimensional matrix into an improved HH threshold neuron model, carrying out Poisson sparse coding on the time-frequency two-dimensional matrix, and only carrying out pulse response on signal significant features; constructing a suprathreshold coding convolutional network with residual connection, inputting a sparse coding matrix, training by adopting an unsupervised learning rule based on STDP, and adaptively adjusting a network synaptic weight; and inputting to a trained above-threshold coding convolutional network, and obtaining pulse emission activity of neurons of an output layer through network forward propagation to determine a fault diagnosis result.

The present disclosure relates to systems, non-transitory computer-readable media, and methods that train and utilize machine learning models to generate perturbation embeddings from phenomic images of cells, including neuronal cell images. Indeed, in one or more implementations, the disclosed systems generate a perturbation embedding using an adapter model or a mixture of experts model. In some implementations, the disclosed systems utilize a mixture of experts model that combines phenomic embeddings from different embedding models to generate a mixture of experts phenomap that contains information from multiple embedding models.

The invention provides an intracranial neuron electrical stimulation system for treating epilepsy and other brain diseases involving epileptic seizure, which can immediately apply intracranial neuron electrical stimulation to the interseizure epilepsy activity in a personalized manner after detecting the interseizure epilepsy pattern each time. And the stimulation is activated by intracranial electroencephalogram (EEG) detection. The intracranial stimulation for epileptic activities in the epileptic seizure period continuously acts along with time, the forming process of an epileptic network can be blocked, the neuronal super-synchronization risk is reduced, and therefore the control effect on epileptic seizure is improved. The invention also includes the use of nerve adjustment biomarkers, thereby dynamically optimizing the detectability and providing more accurate overall assessment for the patient's treatment progress. Finally, according to the scheme, by detecting the intracranial EEG mode and the high-risk state mode (such as the epileptic state and the sudden accidental death risk in epilepsy) in the attack period, an instant alarm is given to the patient / nursing personnel, and it is ensured that safe treatment measures are taken for the patient in time.

The invention provides an agricultural product supply chain traceability data integrity verification method based on machine learning, and belongs to the technical field of agricultural product supply chains. An agricultural product supply chain data acquisition system is established, layering is carried out according to timeliness, a data classification processing framework is constructed, and original data is classified according to complexity; establishing an agricultural product damage prediction model to identify a vulnerable link, and designing a double-layer game optimization model consisting of an upper layer model taking data integrity maximization as a target and a lower layer model taking calculation efficiency optimization as a target, a liquid neural network is adopted to construct a data fusion verification model, liquid neuron state parameters are dynamically adjusted according to the number of supply chain nodes, data timeliness and other parameters, the parameter range of each evaluation matrix is determined through a small-scale verification experiment, and finally integrity verification is executed to calculate and output a comprehensive score result. The technical problems that the agricultural product supply chain traceability data integrity verification accuracy is not high and the calculation efficiency is low are solved.

The disclosure provides technologies that permit detection and / or characterization of brain biomarkers (e.g., disease-associated biomarkers) in non-CNS samples, including by liquid biopsy (e.g., blood, serum, etc.). Among other things, the present disclosure demonstrates that ultrasound opening of the blood brain barrier can achieve detectable increases in brain-derived materials (e.g., brain-derived proteins, neuron-derived extracellular vesicles, and / or cell-free DNA) in readily-sampled systemic liquids.

The invention discloses an electrocardiosignal classification method based on a multi-threshold pulse neural network, which comprises the following steps: acquiring an original electrocardiosignal, and denoising the original electrocardiosignal by using wavelet transform; segmenting the denoised electrocardiosignals into cardiac beat segments according to R wave crest positions, and dividing the cardiac beat segments into a training set and a test set; based on the discharge characteristics of the biological neurons, multi-threshold integral pulse distribution neurons are constructed, and a multi-threshold pulse neural network is constructed; training and testing are carried out, a gradient agent method is used for adjusting parameters of the pulse neural network, an optimal configuration network is obtained, and classification of electrocardiosignals is achieved. According to the method, the extraction capability of the pulse neural network on the electrocardiosignal time sequence characteristics in the training process is improved; and meanwhile, the original sparsity and biological inspiration characteristics are reserved, so that the method not only has excellent performance in the aspect of processing time sequence data, but also can remarkably reduce the calculation energy consumption, and is suitable for portable medical equipment and resource-limited edge equipment.

The invention relates to an edge end spiking neural network compression and deployment method and system, and belongs to the technical field of machine learning, the edge end spiking neural network compression and deployment method performs forward propagation on an initial spiking neural network model based on acquired multi-modal data, obtaining a memory use state in a forward propagation process based on hardware sensing initialization, and performing dynamic sparsification on the initial pulse neural network model based on the memory use state to construct a sparsified pulse neural network; based on the acquired activation statistical information of the sparse spiking neural network, performing dynamic pruning on the sparse spiking neural network by adopting a pruning strategy based on neuron activeness so as to realize compression of a spiking neural network model; the compressed spiking neural network model is optimized according to the edge end configuration, and the optimized spiking neural network model is deployed to the edge end to execute the reasoning task, so that the storage requirement and the calculation complexity are reduced.

The invention belongs to the field of bioinformatics, and relates to an improved integrated deep learning cell communication ligand-receptor interaction prediction method. The method comprises the following steps: firstly, carrying out extraction and dimensionality reduction on biological sequence features of a ligand and a receptor, and constructing multi-modal feature input; secondly, constructing an improved deep neural network branch, introducing a batch normalization layer and a Leaky ReLU activation function, solving the problems of gradient disappearance and neuronal necrosis, and improving regularization strength to prevent overfitting; meanwhile, an enhanced heterogeneous graph auto-encoder branch is constructed, the graph embedding dimension is remarkably expanded to improve the feature capacity, and full convergence of the model is ensured by increasing training rounds; thirdly, fusing the improved deep network with the prediction probability of a heterogeneous graph auto-encoder by adopting a weighted integration strategy; and finally, outputting a potential interaction relationship based on the fusion probability. By optimizing the architecture and the strategy, the prediction accuracy and robustness are remarkably improved, and a reliable tool is provided for analyzing a complex cell communication network.

The invention discloses a retraining-free pruning and recombination method for a sparse expert hybrid large model, and belongs to the technical field of large model compression and optimization. The method aims at solving the problems that due to the fact that an existing sparse expert hybrid (SMoE) model needs to load all expert parameters, memory occupation is too high, and deployment is difficult. According to the method, firstly, redundant experts are identified and pruned based on routing activation statistics; then, decomposing the pruned experts into neuron-level functional fragments, and redistributing the fragments to the reserved experts according to structural similarity; and finally, original fragments and newly distributed fragments are merged in the reserved experts through a weighted clustering algorithm, so that compact experts with fewer parameters and stronger expression ability are reconstructed. According to the method, fine-grained operation is carried out at the neuron level, the inherent representation conflict and dislocation problems among experts are effectively solved, the performance of the compressed model is remarkably improved, and reliable technical support is provided for deploying a large-scale SMoE model.

The invention discloses a pulse neural network image classification method and system based on a space-time adaptive threshold, mainly solving the problems of poor nonlinear expression and limited time sequence and space feature processing ability in the prior art, and the scheme comprises the following steps: obtaining an image data set, and dividing the image data set into a training set and a test set; a spiking neural network main body structure comprising an input layer, a hidden layer and an output layer is selected, an existing neuron model is improved by introducing a space-time joint threshold adjustment mechanism, and improved neurons are placed in each neuron layer in the hidden layer to form a spiking neural network based on a space-time adaptive threshold. The training set is used to carry out iterative training; and inputting the test set into the trained pulse neural network to obtain an image classification result. According to the method, a space-time adaptive threshold mechanism is introduced, the threshold can be dynamically adjusted to adapt to time and space features, the processing capacity of the network on time sequence data and complex features and the classification accuracy of images are remarkably improved, and the method can be widely applied to dynamic visual tasks and event-driven scenes.

The invention relates to a neural network global one-time structured pruning method, system and device and a medium, and the method comprises the steps: carrying out the parallel capturing of the input activation tensors of all target layers in a to-be-pruned neural network through a calibration data set through single-time forward propagation; according to an input activation tensor, synchronously calculating a difference entropy index and amplitude response intensity for an intermediate neuron weight group of each target layer, and performing normalization and fusion to form a static global importance map; determining an importance threshold according to a preset pruning rate, and generating a global to-be-pruned index set of which the mixed importance score is lower than the importance threshold at one time based on the global importance map; and on the basis of the index set, performing one-time physical structured pruning on the weight matrix of each target layer. Therefore, the static global importance map is generated through single forward propagation and parallel capture of the activation tensor, and maskless one-time pruning is completed through physical structured pruning.

The invention, which relates to the technical field of hydropower station dam safety monitoring, discloses a hydropower station dam safety monitoring data acquisition and transmission system comprising a cloud twin brain module and edge neurons. The cloud twin brain module comprises a sequence neural network engine and a reflection kernel generation module, the sequence neural network engine adopts a neural network architecture with parallel and cyclic dual representation, comprises a time mixing module and a channel mixing module, and can learn a normal operation mode of the dam from historical monitoring data; and the reflection nuclear generation module compresses the reference twin model into a lightweight reflection nuclear model and issues the lightweight reflection nuclear model to the edge device. The edge neuron comprises a micro-twinborn prediction module and a hierarchical transmission control module, and the micro-twinborn prediction module predicts a theoretical expected value of a dam state in real time and calculates a reflection deviation with an actual observation value; the hierarchical transmission control module implements a three-level response strategy according to the magnitude of the reflection deviation, transmits abstract information according to an abnormal trend, and uploads an emergency abnormality in time.

The invention provides a storage cluster fault diagnosis method and system based on a spiking neural P calculation model, and the method comprises the steps: collecting operation indexes containing static data and dynamic data from a monitoring system of a storage cluster, defining a neuron for each operation index after preprocessing through the construction and training of the spiking neural P calculation model, and carrying out the fault diagnosis of the storage cluster. Initializing an initial state value of each neuron according to the collected data; and performing state updating and reasoning by using pulse signal propagation and fuzzy reasoning through a trained spiking neural membrane calculation model, judging a fault according to a state value of a neuron, and when the state value of a certain neuron exceeds a set fault threshold value, considering that an index has a fault or a potential fault, and identifying a specific fault type in the storage cluster through state combination of the plurality of neurons. According to the invention, efficient fault diagnosis of the distributed storage system is realized.

The invention provides an artificial cochlea which comprises an implant and an extracorporeal machine, the implant and the extracorporeal machine both adopt a magnet-free design, the extracorporeal machine is configured to receive a sound signal and transmit the sound signal to the implant after first processing, the implant is configured to perform second processing on the sound signal after first processing, and the implant is configured to perform second processing on the sound signal after second processing. The auditory neuron is stimulated by implanting an electrode; the in-vitro machine comprises an electrode array, the electrode array comprises a base and a plurality of electrode assemblies, and the electrode assemblies are inserted into a target scalp area in a semi-intrusive mode so that at least part of the in-vitro machine can be fixed to the target scalp area; wherein the implant is arranged on the inner side of the target scalp area, and the extracorporeal machine is arranged on the outer side of the target scalp area. And due to the non-magnet design, the artificial cochlea can improve the use experience of the user.

The invention relates to the technical field of power system scheduling, and discloses an intra-day look-ahead scheduling rapid solving method considering large-scale new energy cluster power generation volatility. Comprising the following steps of S1, new energy cluster space-time fluctuation scene generation based on a neuron cellular automaton, S2, power grid dynamic security domain definition and simplification based on a physical information neural network, S3, scheduling rapid optimization solution based on model prediction path integration, and S4, scheduling scheme dynamic elasticity and stability evaluation based on a Kupman operator theory. The new energy cluster space-time fluctuation scene generation method based on the neuron cell automaton can effectively generate a space-time scene reflecting large-scale new energy cluster power generation volatility, supports uncertainty analysis, has the advantages of being high in calculation efficiency and scene authenticity, and is suitable for large-scale new energy cluster power generation. The problem that scene generation is inaccurate due to the fact that a traditional statistical model ignores space-time coupling is solved.

The invention relates to the technical field of artificial intelligence model optimization, and discloses a multi-granularity dynamic pruning method and system for a generative AI model. The system comprises a model state acquisition module which acquires output of a middle layer in real time through a probe, and constructs a feature information set of neuron activation distribution, weight matrix norm and connection topology; the sparseness evaluation module outputs a sparseness risk value based on the feature information set, the initial sparseness parameter and the real-time computing resource state; a pruning planning collaborative analysis screening module generates a multi-granularity pruning scheme when the risk value exceeds a threshold value, calculates a collaborative interference value of precision recovery operation and screens an optimal scheme; and the pruning strategy execution feedback module updates model parameters, generates logs and feeds the logs back to the management terminal. The system realizes adaptive structure optimization and resource scheduling of the generative model in the reasoning process.

The invention discloses a poisoning defense method based on characteristic difference analysis and model layer purification. The method comprises the following steps: S1, constructing a poisoning classification model according to an electromagnetic signal sample; adversarial disturbance is introduced, cross entropy loss of disturbed samples is calculated and sorted, and a threshold value is set to distinguish clean samples from poisoned samples; s2, after the poisoning samples are separated out, the poisoning score of each layer of the model is calculated through quantification, and the higher the score is, the stronger the influence of the back door neurons of the layer on the poisoning samples is; s3, generating a pseudo-poisoning sample based on an inversion trigger, analyzing the feature distribution difference between the pseudo-poisoning sample and a clean sample, aligning a feature space and purifying a model layer by optimizing an objective function, and enhancing the distinguishing ability of the model to the sample; s4, repeating the step S3 until the model classification precision converges, and storing the optimal network parameters; and the classification precision of the model on a normal sample and the attack success rate on a poisoning sample before and after defense are evaluated. According to the method, the robustness and the safety of the model are improved, and the method has relatively high universality.

A device for modelling physiological and pathophysiological states of the brain is provided. The device comprises a hydrogel with micropattern channels having a length, width and depth. The micropattern channels comprise cells selected from the group consisting of neurons, astrocytes, microglia, oligodendrocytes, neuronal organoids, cancer cells, cancer spheroids, brain tumoral cells and combinations thereof.

The invention discloses a six-dimensional force / torque sensor decoupling system and method based on a width neural network, and the method comprises the steps: in a data collection and preprocessing unit, carrying out the null drift correction and normalization processing of a collected voltage signal through a weight type calibration platform and a data collection card, and forming a network model training data set; carrying out model training on the established six-dimensional force / torque sensor decoupling model of the width neural network, and solving a mapping matrix from a hidden layer to an output layer by adopting an importance score-based neuron pruning algorithm and a Moore-Pengos generalized inverse to obtain optimal model parameters; and finally, inputting the voltage signal subjected to data preprocessing to obtain a real-time output vector, and performing reverse normalization processing to obtain a decoupled six-dimensional force / torque vector. According to the method, precise decoupling of the six-dimensional force / torque sensor is achieved, and the inter-dimensional coupling error is remarkably reduced.

The invention relates to the technical field of artificial intelligence, and discloses a brain-like multi-mode emotion recognition network, a brain-like multi-mode emotion recognition method and an emotion robot, and the brain-like multi-mode emotion recognition method comprises the steps: converting multi-mode emotion information into a pulse sequence; on the basis of the obtained pulse sequence, constructing a continuous pedigree emotion representation space, and realizing continuous representation of an emotional state; a continuous pedigree emotion representation space is utilized to generate pulse codes of mixed emotions, and a complex emotional state is represented; according to the pulse codes of the mixed emotions, establishing an emotional state conversion probability model, and describing a conversion relation between the emotions; based on an emotional state transition probability model, realizing prediction and processing of an emotional gradual change process, and capturing subtle emotional changes; according to the method, the working principle of human brain neurons can be used for reference, fine representation and accurate prediction of the human emotional state are achieved, and meanwhile high calculation efficiency and biological interpretability are achieved.

The invention provides a large model alignment method and device based on knowledge forgetting, and relates to the technical field of natural language processing. The method comprises the following steps: obtaining each parameter weight in a large language model, calculating an importance score for each neuron, sorting the neuron, and constructing a binary knowledge forgetting mask; determining a knowledge forgetting layer and a forgetting module according to a sorting result; obtaining a harmful instruction and a harmful response, and constructing a harmful knowledge data set; and setting an optimization target, and performing harmful knowledge forgetting training on the knowledge forgetting layer and the forgetting module according to the binary knowledge forgetting mask, the harmful knowledge data set, the optimization target and a limited gradient ascending algorithm to obtain a trained large language model. According to the CKU method provided by the invention, by adopting a constraint optimization technology, a security alignment task is converted into a limited knowledge forgetting task, unnecessary or harmful knowledge in the generative large language model is accurately removed, and meanwhile, the overall performance and efficiency of the model are ensured to be kept.

The invention provides a water chilling unit performance prediction method based on an improved physical information neural network. The method comprises the following steps: step 10, constructing a neural network model; step 20, selecting to obtain a neuron variable, and embedding the neuron variable into the neural network model as a physical neuron to obtain a physical neuron embedded physical information neural network model; step 30, introducing a loss function into the physical neuron embedded physical information neural network model; 40, carrying out data balance processing on the original data set, and screening to obtain a training set; step 50, utilizing the training set to train the physical neuron embedded physical information neural network model to obtain a water chilling unit performance prediction model; and step 60, predicting the performance of the water chilling unit by using the water chilling unit performance prediction model. According to the improved water chilling unit performance prediction method based on the physical information neural network, high-precision prediction of the performance of the water chilling unit is achieved, and the generalization ability and the model interpretability are improved.