179results about How to "Efficient training process" patented technology

The invention belongs to the technical field of image classification neural networks, and discloses a lightweight image classification neural network architecture system based on knowledge distillation. The invention provides a lightweight image classification neural network architecture system based on knowledge distillation. The lightweight image classification neural network architecture system comprises a data set preprocessing module, a teacher model preprocessing module, a subnet model construction module, a personalized training module, a real loss calculation module, a distillation loss calculation module, a total loss calculation module and a student model construction module. The invention provides a lightweight neural network architecture system based on knowledge distillation, and designs a brand new training mode by using the superiority of knowledge distillation for teacher model weight extraction: the knowledge distillation of teacher and student models is combined with a deep compression method; the system for constructing the lightweight network can be used for training various complex image classification neural networks so as to improve the network performance.

The invention discloses an explosion-proof junction box fault diagnosis method and system based on a sensor, which are used in the field of fault diagnosis, and the method comprises the following steps: obtaining historical multi-source sensor data reflecting the state of an explosion-proof junction box; extracting a time sequence sample of the preprocessed historical multi-source sensor data by using a sliding time window method; multi-domain features reflecting fault types are extracted from the time sequence samples, and a fault diagnosis model is constructed based on the feature combination with the optimal diagnosis performance; performing feature extraction on the multi-source sensor data acquired in real time, and inputting the extracted features into a fault diagnosis model for fault diagnosis; and the operation state and the diagnosis result of the explosion-proof junction box are displayed in real time. According to the method, the subtle change of the fault signal is captured from multiple angles, the Pearson correlation coefficient is used for screening the features significantly related to the fault type, and the dimension reduction processing is carried out through principal component analysis, so that the operation efficiency and accuracy of the model are improved.

The invention relates to the technical field of AI computing, in particular to a knowledge reinforcement learning method and system based on multiple agents. Modifying parameters of the large language model by using a supervised fine-tuning training method for the pre-trained large language model; constructing a knowledge base of a large language model, and vectorizing, storing and retrieving texts in various file formats; a multi-agent workflow optimization information processing and decision-making process is developed, a real-time problem is decomposed, a logic chain for solving the problem is constructed, and internet texts are obtained, filtered and summarized. Compared with the prior art, the method has the advantages that the knowledge quantity of the pre-trained large language model in the specific field is increased, the knowledge base is high in flexibility, the obtained text is high in real-time performance, certain interpretability is achieved and the like, and then the ability of the large language model to solve problems in the specific field is improved.

The invention discloses a federal learning data processing method based on privacy protection, and the method comprises the following steps: S1, a client uses local data to train a generator of a generative adversarial network, and generates virtual data similar to the local data in distribution; s2, setting three privacy protection levels according to the privacy sensitivity of the local data; s3, performing model updating operation by using a homomorphic encryption technology; s4, utilizing a dynamic weight optimization mechanism based on a multi-arm bandit algorithm to obtain a priority processing data set; s5, preliminarily updating the global model, analyzing generated data through a reverse inference detection mechanism, and judging whether a privacy disclosure risk exists or not; s6, the server feeds back the updated global model to the client, and the client continues personalized model training in combination with local data; and S7, generating new virtual data, and carrying out the next round of global model training. According to the method, the generative adversarial network, differential privacy and homomorphic encryption are combined, so that efficient federated learning under privacy protection is realized.

The invention provides a deep learning large model training method and system for heterogeneous equipment, belongs to the technical field of model training, and aims to solve the problem that a heterogeneous GPU cluster cannot be effectively utilized when a traditional scheme is used for training a deep learning large model. Different network layers of a deep learning large model to be trained are divided into a plurality of stages, forward propagation and back propagation calculation of all network layers in each stage is executed by independent virtual equipment, and meanwhile, utilization of GPU resources of different structures is coordinated in combination with a proposed hybrid parallel training strategy, so that efficient model training is realized.

The invention discloses a lightweight small target detection method based on improved YOLOv8. The method comprises the following specific steps: constructing a small target data set; building an Improved-YOLO small target detection model on the basis of a YOLOv8 framework; training an Improved-YOLO small target detection model by using the data set, and performing model quantization operation in the training process to obtain a lightweight detection model; reasoning acceleration of the lightweight detection model is carried out, and the model is embedded into an edge computing terminal for deployment; a camera is used for collecting video data, and the collected data are input into an edge computing terminal for processing; and outputting the processed video data, and displaying the processed video data on a client in real time. According to the method, efficient training and quantification of the model are realized, the model is successfully deployed in the edge computing terminal more innovatively, and the problems of poor real-time performance, limited computing resources and the like of a traditional target detection method in an edge environment are solved.

The invention belongs to the technical field of data asset management, and discloses an enterprise data asset auditing method and system based on big data, and the method comprises the steps: collecting financial information data, business operation data, management decision data and external industry data of an enterprise; preprocessing the collected financial information data, business operation data, management decision data and external industry data to obtain a financial feature data set, a business operation feature data set, a management decision feature data set and an external industry feature data set; performing feature extraction and weighted fusion on the financial feature data set, the business operation feature data set, the management decision feature data set and the external industry feature data set to obtain a comprehensive feature data set; constructing an audit risk prediction model, inputting the comprehensive feature data set into the audit risk prediction model, and predicting to obtain an audit risk coefficient of the enterprise data assets; and a more comprehensive view angle is provided, and the data asset risk of the enterprise can be evaluated more accurately.

The invention discloses an automatic model tuning system based on an AI training intelligent workbench, and relates to the technical field of artificial intelligence, and the system comprises the steps: setting a data collection instrument and an environment monitoring point, collecting model real-time data in a model training process, and carrying out the preprocessing; performing model performance analysis according to the model real-time data and the model original data to obtain final parameter configuration; and performing model training by utilizing the final parameter configuration, dividing the model training process into different stages, dynamically adjusting an intelligent control strategy based on model performance requirements, and performing optimization on the different stages of the model. According to the method, through multi-dimensional performance evaluation, parameter configuration is optimized, and the robustness and generalization ability of the model are improved. Through dynamic adjustment of an intelligent control strategy, fine management of a training process is realized, efficiency and resource utilization are improved, model convergence is accelerated, and finally an efficient, stable and high-adaptability model training effect is achieved.

The invention provides a time sequence data missing value filling method based on a diffusion model. The method comprises the following steps: constructing a filling model, training the filling model, and performing data missing value filling by utilizing the trained filling model. The filling model comprises a denoising diffusion module and an evaluation filling module; the de-noising diffusion module comprises an input data processing module, a residual connection block and an output integration module; the evaluation filling module comprises an incremental filling module and a sampling result integration module; according to the method, based on the current most advanced diffusion model architecture DiffWave, a self-supervised training mode is adopted, a filling target is generated through data of a Mask non-missing part, and meanwhile, an observable part information is used for assisting a diffusion model to predict noise, so that efficient and accurate training of the diffusion model is realized; according to the method, the time sequence missing value filling effect can be effectively improved, the problem of data missing in the time sequence is solved, and the data quality and value are improved.

The invention relates to the technical field of artificial intelligence, and discloses a multi-modal model training fine-tuning optimization method based on artificial intelligence, which comprises the following specific steps: preparing and preprocessing data, collecting a large-scale multi-modal data set, constructing an image text data set containing one million high-quality images and text description accurately corresponding to the images, and performing fine-tuning optimization on the image text data set. According to the method, data are subjected to comprehensive and meticulous cleaning operation, noise data and mistakenly labeled data in the data are strictly removed, so that the accuracy and reliability of the data are ensured, and the effective utilization rate of the data can be increased by at least 20% through data cleaning; wherein the image data is screened, blurred, low-resolution and unclear-content images are removed, it is ensured that each image can clearly convey specific visual information, and text description, grammar error checking, spelling error and semantic ambiguity parts are corrected or deleted.

The invention relates to the technical field of city management, in particular to a city management illegal behavior recognition system and method based on a large model, and the system comprises a front-end sensing module which is used for collecting city site information; the edge calculation module is used for carrying out real-time analysis on the collected city site information, and comprises a model construction sub-module for constructing a Transform model for identifying city management illegal behaviors; using a CLIP algorithm to carry out joint training on the image and the text, and mapping the image and the text to a shared embedding space; reinforcement learning is carried out by using an RLHF algorithm, and the learning effect of the intelligent agent is optimized by combining knowledge and experience of experts; the data management module is used for labeling and managing the data; and the cloud management storage module is used for storing the data of the front-end sensing module and managing and providing a remote access interface. According to the technical scheme, urban management illegal behaviors can be efficiently and accurately identified and analyzed in real time, so that the urban management efficiency is improved.

The invention provides a low-resource language-oriented large language model training method and device, equipment and a medium, and relates to the technical field of artificial intelligence. The method comprises the following steps: acquiring low-resource language data and preprocessing the low-resource language data to obtain low-resource language unsupervised training data; generating a low-resource language vocabulary and performing vocabulary expansion on the existing model; performing unsupervised pre-training on the model after word list expansion according to the low-resource language unsupervised training data in combination with a first low-rank adaptive module; performing instruction fine tuning on the pre-trained model according to the low-resource language instruction fine tuning data set in combination with a second low-rank adaptive module; and respectively inputting a high-resource language supervised corpus into the model subjected to instruction fine tuning and the existing mature model to respectively obtain negative example data and positive example data so as to generate a reinforcement learning low-resource language data set based on model feedback, and carrying out reinforcement learning training on the model subjected to instruction fine tuning in combination with a third low-rank adaptive module.

The invention provides a single-channel speech enhancement method based on an improved convolutional recurrent network, and the method comprises the steps: 1, downloading a data set for model training and testing, carrying out the preprocessing of the speech of the data set through employing a VoiceBank + DEMAND data set, and extracting the amplitude spectrum of a speech signal through short-time Fourier transform (STFT); 2, inputting the amplitude spectrum into an encoder, and extracting high-dimensional features from the amplitude spectrum; step 3, using the Channel-S4D block as a recursion module of the model, inputting a result output by the Channel-S4D block into a decoder, and recovering the result to an original dimension; 4, constructing a joint loss function; and 5, reconstructing and evaluating the enhanced voice signal. According to the invention, the Channel-S4D module constructed based on the diagonalization state space model efficiently processes a voice sequence, captures channel-level voice key information, better models the time dependence relationship of voice signals, and improves the adaptability of the model to a complex noise environment while reducing the parameter quantity and the calculation quantity of the model.

The invention discloses an infrared target detection method and model based on convolution attention. The model comprises a backbone network, a neck network and a detection layer. Generating feature maps of multiple scales through repeated feature extraction and attention adjustment of the backbone network; performing multi-scale feature fusion on the multi-layer feature map through a neck network module to generate a final feature map suitable for detection; and converting the final feature map into a final target detection result through a detection layer. According to the method, small targets in the image can be more effectively concerned by introducing an attention mechanism, so that the detection precision is improved, detail information and context information of the image can be captured at the same time by utilizing a multi-scale feature fusion technology, the feature expression ability is enhanced, and the content of the image can be more carefully understood.

The invention discloses a short-term power load prediction method, system and device based on a VAE-BiLSTM hybrid neural network and Bayesian optimization, and a medium. The method comprises the following steps: preprocessing obtained load data related to a power system and feature data influencing a load; performing feature dimension reduction on the processed data by using kernel principal component analysis; performing hyper-parameter optimization on the network structures of the variational auto-encoder and the bidirectional long-short-term memory network by using a Bayesian optimization method; training the dimensionality-reduced data containing the data principal component by using a variational auto-encoder with an optimal hyper-parameter; inputting a coding layer containing data potential representation in the trained variational auto-encoder network structure into the bidirectional long-short-term memory network model with the optimal hyper-parameter for load prediction to obtain a short-term power load prediction result; performing accuracy evaluation on the prediction result; according to the method, reasonable selection of time sequence correlation and hyper-parameters can be fully considered in power load prediction, and the accuracy of power load prediction is improved.

The invention discloses a flight service system based on a knowledge graph and a large model. Comprising an ontology construction module used for predefining related information of a flight service system based on user demands; the data acquisition module is used for acquiring open source data related to the flight service system from multiple channels and performing deduplication and normalization processing on the open source data to obtain standard open source data; the knowledge graph construction module is used for extracting triads from standard open source data based on the related information to obtain a triad set of the knowledge graph; the prompt generation module is used for generating a corresponding target prompt based on a user demand; the large model fine tuning module is used for training a pre-constructed large model based on the target prompt and the triple set to obtain a trained large model; the trained large model is used for generating a result matched with the target task according to an input user prompt and the triple. According to the invention, diversified services can be provided for users.

The invention discloses a thromboembolism occurrence risk assessment method and assessment system. The method comprises the following steps: acquiring a blood vessel image map of a to-be-assessed patient; inputting the blood vessel image map of the patient to be assessed into a pre-trained risk assessment model; wherein the pre-training of the risk assessment model is based on a ResNet-50 model, and self-supervised contrast learning iteration is carried out by inputting positive and negative samples so as to update model parameters; extracting a feature embedding vector in the blood vessel image; according to the feature embedding vector, calculating probability distribution of embolism occurrence risk levels through a model classification head; and according to the probability distribution, selecting a risk level label and outputting the risk level label as an evaluation result. According to the method based on the time axis symmetry, the image data of the patient is collected before and after the embolism occurrence time point, the positive and negative sample pairs are generated, the problem that the embolism sample is difficult to obtain is solved, and meanwhile the sample collection cost is reduced through self-supervised learning.

According to the saliency reference and decoupling esophageal tumor CT image diffusion segmentation method provided by the invention, the accuracy of tumor segmentation is improved by combining a saliency pseudo tag and a diffusion denoising network. The method comprises the following steps: firstly, preprocessing an esophageal tumor CT image, generating a significant pseudo tag, and enhancing the attention of a network to a tumor region; and then, constructing a saliency reference segmentation network and a diffusion denoising network, and gradually carrying out noise adding and denoising to restore the image by adopting a conditional diffusion model. A multi-task learning framework for decoupling segmentation and de-noising tasks is designed, and collaborative optimization between tasks is ensured through consistency constraint. Through joint optimization of cross entropy loss, Dice loss, mean square error and consistency loss, the segmentation performance is improved. Experimental results show that the method can effectively improve the segmentation precision of the esophageal tumor, and has high clinical application potential.

The invention relates to the technical field of image recognition, in particular to an intelligent safety helmet construction environment detection method and system based on image recognition, and the method comprises the steps: obtaining the image data of a construction environment; preprocessing the acquired image data, including denoising the construction image and enhancing the denoised image data; the preprocessed image is used as input to construct a prediction model, and an attention mechanism is introduced for the prediction model; the method comprises the steps of constructing a prediction model based on a convolutional neural network, a recurrent neural network and a generative adversarial network, performing model training according to the constructed prediction model, performing feature extraction based on the trained prediction model, and performing judgment of a danger threshold according to output image features, and can fully utilize the advantages of different networks by constructing the prediction model based on the convolutional neural network, the recurrent neural network and the generative adversarial network. Therefore, the judgment on dangerous conditions is improved.

The invention provides a big language model training method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence. The training method of the large language model comprises the following steps: acquiring an instruction response data pair; wherein the instruction response data pair is composed of an instruction and a response generated by utilizing a large language model according to the instruction, and the response comprises a text of a preset field; extracting constraint information contained in the response from the response contained in the instruction response data pair; generating a training data set according to the response, constraint information contained in the response and an instruction corresponding to the response; and using the training data set to continuously train the pre-trained large language model to obtain a target large language model, and using the target large language model to generate a response according to the instruction. According to the method, the large language model capable of accurately executing the instruction with the complex constraint can be efficiently trained, so that the utilization rate of computing resources is effectively improved.

The invention discloses an underwater vehicle hydrodynamic noise rapid forecasting method and system based on a BP neural network optimized by a genetic algorithm, belongs to the field of ship underwater radiation noise forecasting, and designs a BP neural network noise forecasting model optimized by the genetic algorithm. Key variables including shell thickness, molded line parameters, surface roughness, material attributes, sailing speed, fluid medium density, viscosity coefficient and underwater vehicle diving depth are selected as characteristic conditions. The initial weight and threshold value of the optimal BP network are obtained through the global optimization capacity of the genetic algorithm, the optimal initial weight and threshold value obtained through the optimization algorithm serve as the initial weight and threshold value of the BP neural network, and training is carried out to avoid falling into the local minimum value. And performing iterative training on the optimization model by using the training set to obtain a high-precision noise forecasting model, thereby realizing forecasting of the hydrodynamic noise of the underwater vehicle. The method is suitable for forecasting the hydrodynamic noise of the underwater vehicle, and has the advantages of wide applicability, high accuracy and the like.

The invention discloses a physical fitness training cloud data acquisition system and method, and relates to the technical field of fitness data acquisition, and the method comprises the following steps: collecting physical fitness training data from multiple data sources; dividing the participants with similar characteristics into the same group, calculating the variation of each body part data of the participants in each group before and after training, and determining the body part influenced by each physical fitness training type; screening out a group having similar characteristics with the user from the sub-data set, and generating an initial training scheme in combination with the difference between the training target of the user and the current physique level; the total training load of each body part is calculated, the initial training scheme is adjusted, and a final training scheme is generated; according to the body part corresponding to each physical fitness training in the final training scheme, acquiring physical fitness training cloud data; the personalized implementation of the training scheme is ensured, and the accuracy and effectiveness of the data are also improved.

The invention relates to the technical field of medical image analysis, in particular to a training method and system for a cataract recognition model and a storage medium. The method comprises the following steps: acquiring cataract eye image samples and non-cataract eye image samples in different stages; lesion region segmentation and image synchronous simulation enhancement are carried out on the cataract eye image samples and the non-cataract eye image samples in different stages to obtain cataract lesion enhanced image samples and non-cataract enhanced image samples; extracting local detail features of the lesion and extracting structural features of a mask pair, and constructing a multi-task convolutional neural network learning framework to train a cataract recognition model and optimize focus weighted loss; and obtaining a cataract related key feature set and carrying out online incremental learning updating to generate a cataract recognition online incremental updating model. According to the invention, the accuracy and robustness of cataract recognition model training can be improved.

The invention relates to the technical field of radar weak target detection, in particular to a radar weak target detection method and device based on deep learning. According to the technical scheme, the method comprises the following steps: data preprocessing and enhancement: receiving an original echo signal of a radar, generating three-dimensional time frequency-wavelet coefficient spatial data by combining time frequency analysis with wavelet transform, and expanding a data set in multiple modes; and feature extraction network construction: constructing a 3D-CNN and LSTM fused network, wherein the 3D-CNN has a convolution operation with a variable convolution kernel size. According to the method, the data are richer and more comprehensive through time-frequency analysis in combination with wavelet transform and data enhancement, and multi-dimensional features can be accurately captured through 3D-CNN and LSTM fusion and an adaptive convolution kernel; and the training effect is ensured through the adaptive learning rate and the early stop method, and the accuracy and adaptability of weak target detection are jointly improved.

The invention relates to a continuous identity authentication method based on a multi-level memory enhanced auto-encoder, which belongs to the technical field of continuous authentication and comprises the following steps: S1, collecting user behavior data; s2, performing normalization processing on the user behavior data; s3, constructing a VAE-based data enhancement model, and performing training by using normalized data; s4, constructing a data reconstruction model based on MulMemAE, and performing training by using normalized data and VAE enhanced data; s5, inputting the user behavior data collected and normalized in real time into the trained data reconstruction model to reconstruct the user behavior data; s6, comparing the reconstructed user behavior data with the input user behavior data, and calculating a reconstruction error; s7, the reconstruction error is compared with a preset threshold value, and if the error is lower than the preset threshold value, the user is a legal user; on the contrary, if the user is a famous person, the user is required to log in again

The invention provides a multi-source software-defined science and education resource recommendation method and system based on a knowledge graph, and the method comprises the steps: constructing an artificial intelligence knowledge graph based on the relation between artificial intelligence knowledge entities; obtaining science and education resource data of a science and education institution, extracting knowledge entities in the science and education resource data, and connecting the knowledge entities to the artificial intelligence knowledge graph to obtain a science and education resource knowledge graph; extracting local resource features / user features based on a graph neural network to obtain local resource features and user features of each client; constructing a local knowledge graph based on the science and education resource knowledge graph, and performing model training by taking the local resource features of each client as initial features until the model converges; and inputting the users and the resources into the graph neural network for prediction, and recommending the resources with high scores to each user. According to the scheme, the relationship between knowledge entities in the field of artificial intelligence is effectively extracted, a structured knowledge graph is formed, and the recommendation precision is improved.

The invention belongs to the technical field of federal learning, and discloses an intelligent model training method and system for distributed equipment collaborative optimization and a storage medium, and the algorithm comprises the steps that a client receives an original model sent by a server, and completes an initialization task of the model; the client carries out local training on the received initialization model based on the local data set to obtain a local training model, all basic parameters of local training are obtained through calculation, and a memory-free SR1 matrix is generated; the server randomly extracts part of clients, and the selected clients upload local model training parameters of the clients; updating an ADMM direction, and updating an internal problem; solving a variable-scale near-end mapping problem to obtain a near-end direction, and solving an external problem; and updating the global model according to the obtained near-end direction until the model training is completed. Under the condition that statistical heterogeneity exists, compared with an existing first-order composite optimization algorithm, the algorithm has the advantages that the convergence speed is higher, the convergence condition of the algorithm can be met more quickly, and progressive convergence is achieved.

The invention discloses a Gaussian sputtering general rendering enhancement method based on space fusion. The method comprises the following steps: obtaining a real image to be processed; constructing a degradation simulator of a Gaussian sputtering style, adding a gating mechanism into the degradation simulator, selecting a degradation type in an image degradation process, and generating a training data set; inputting a real image to be processed into the degradation simulator to generate a training image; and a Gaussian intensifier is constructed, the Gaussian intensifier utilizes a spatial information fusion network to carry out feature fusion and enhancement rendering processing on the training image through a view fusion module and a depth modulation module, and a final enhancement result is generated. According to the method, highly-correlated and high-quality training images can be mixed, and details are rendered by using the depth information of the target image, so that the output rendering quality of the image is improved.

The invention discloses a large model training resource evaluation and strategy recommendation method, which comprises the following steps of: 1) constructing a large model database, and storing model state data of different types and sizes; the model database supports expansion and is used for synchronously updating the latest open source large model; 2) constructing a calculation overhead evaluation database of different distributed strategies; defining the video memory efficiency as an index for measuring the degree of storing redundant information on each video card; the calculation efficiency is defined as the proportion of the time for measuring the effective calculation of the graphics card, namely the time for updating the model state and the time for communication; 3) scanning existing computing power resources, traversing the strategy priority queue, and selecting a first hit strategy as a recommendation strategy; aiming at the current available computing resources, recommending a proper distributed training strategy, and giving a super parameter of training recommendation; and according to the selected large model type, estimating the training overhead, scanning node computing resources, giving a training strategy, automatically constructing a configuration file of a training framework, and cooperating with a training platform to realize efficient training of the large model.

The invention discloses a distributed photovoltaic power prediction method and system based on time sequence data reconstruction. The method comprises the following steps: collecting a plurality of single-class distributed photovoltaic power curves; adopting improved photovoltaic power data empirical mode decomposition to decompose the plurality of single-class distributed photovoltaic power curves into sub-power curves with different frequencies; the sub-power curves of different frequencies are reconstructed; a neural network is adopted to predict the reconstructed sub-power curves, and then superposition and summation are carried out on predicted values of the sub-power curves to obtain distributed photovoltaic power data of a certain type; according to the method, photovoltaic power is classified into single-class curves of different scenes according to different seasons and weathers, and sub-power curve division is performed by using improved empirical mode decomposition, so that the randomness and volatility of original data can be reduced to the greatest extent before prediction, a neural network can be conveniently and efficiently trained, and the prediction efficiency is improved. And outputting accurate photovoltaic power time sequence prediction data.