627 results about "Training phase" patented technology

The invention relates to the technical field of spatial-temporal trajectory data anomaly detection, in particular to a trajectory anomaly route detection method and system based on self-supervised trajectory representation learning. The method comprises the following steps: preprocessing acquired trajectory data; track double-view comparison representation learning is carried out based on a double-view-angle synchronous mask strategy; coding time dynamic based on a space-time fusion mechanism and fusing the time dynamic with spatial features; learning essential representation from the fused spatio-temporal features to perform trajectory reconstruction; and error checking is carried out based on the reconstructed trajectory, and an abnormal trajectory is judged through the reconstructed error. The invention provides a brand new trajectory anomaly detection model. According to the model, a GPS track and a grid-based track feature are fused, so that track representation is enriched; and meanwhile, a double-view synchronous mask mechanism is designed, so that the model can sense local disturbance of space and time dimensions at the same time in a training stage, and thus the sensitivity to local anomaly is improved.

The invention discloses a three-dimensional scene reconstruction method and device based on large model geometric prior, and a medium, and aims to solve the problems that a conventional 3DGS is liable to have artifacts and detail loss in geometric discontinuity, data redundancy and illumination variation scenes, and predicts a dense depth map and a normal map from a monocular image by using a pre-trained large model. The position and form of the Gaussian kernel are constrained as additional geometric priori; a primitive adjustment strategy based on kernel density estimation is introduced in the training stage, small Gaussian primitives with similar structures and adjacent spaces are combined into a large Gaussian primitive, the rendering quality is kept, redundancy is reduced, and the volume of the model is reduced; an exposure coefficient is adaptively estimated for each input image, an exposure compensation image loss function is constructed, and floating artifacts caused by illumination differences at shooting moments are eliminated. Experiments show that compared with the prior art, the method improves the three-dimensional reconstruction precision and real-time rendering quality of complex illumination and less-texture areas in a public data set and an unmanned aerial vehicle aerial photography scene.

The invention relates to the technical field of computer-generated holographic three-dimensional display and deep learning, in particular to an unsupervised deep learning method for realizing three-dimensional holographic display. The method comprises the steps of generating a depth map corresponding to a two-dimensional image; the depth image and the two-dimensional image are spliced in the channel dimension to serve as input of a hologram encoder, and a double-U-Net cascade neural network architecture serves as the hologram encoder; angular spectrum diffraction back propagation is carried out through the generated pure phase hologram to realize three-dimensional scene discretization reconstruction, a reconstructed image with a specified depth is obtained, a depth map is uniformly quantized to obtain a plurality of binary masks, loss calculation is carried out on the reconstructed image and a target image superposed with the corresponding depth binary masks, network parameter optimization is carried out, and a target image with the depth corresponding to the target image is obtained. And when the training of the double U-Net cascade neural network architecture is converged, the training stage is ended. According to the method, high-quality three-dimensional hologram reconstruction is realized through layered angular spectrum propagation, and the method has relatively high precision and detail reduction capability.

The embodiment of the invention discloses a flight path data analysis method based on a depth auto-encoder and a generative adversarial network, relates to the technical field of air traffic management, and adopts a cubic spline interpolation method to enable each flight path to have the same length; a deep auto-encoder (DAE) and a generative adversarial network (GAN) are used as a basic framework to establish a flight trajectory data analysis framework, and two tasks of trajectory anomaly detection and flow pattern recognition are synchronously executed; in the pre-training stage, on the basis of DAE, a discriminator is integrated to establish an abnormal trajectory detection task, and an abnormal score function is formulated in combination with reconstruction loss and discriminator loss; in the fine tuning stage, a Gaussian mixture model (GMM) is adopted for low-dimensional representation of the trajectory in a potential space output by an encoder to establish a flow pattern recognition task, and in order to enhance clustering, a loss function is constructed in combination with clustering distribution enhancement. The method is suitable for civil aviation air traffic management.

The invention relates to the technical field of computer vision, in particular to a reloading pedestrian re-identification method based on a visual language pre-training model. The method comprises the following steps: a training stage: inputting an image to obtain a clothing mask image, and generating clothing irrelevant / relevant prompts; text encoder parameters are fixed, prompt weights are optimized, text prompts are input into an encoder to obtain features, and a classifier is constructed to achieve image-text alignment through cross entropy loss; using a visual encoder to extract mask pattern features, and constraining a class center Euclidean distance to realize image-image alignment; stripping clothes characteristics: extracting clothes area characteristics and corresponding text characteristics, optimizing by a classifier, and introducing orthogonal loss to decouple clothes correlation; in the reasoning stage, a query image is input into a trained image encoder to extract features, and cosine similarity ranking and result returning are calculated according to the features of the image library. According to the technical scheme, the recognition accuracy of the pedestrian re-recognition method under the condition of pedestrian clothing change can be improved.

The invention relates to the field of intelligent automatic driving, in particular to a multi-vehicle cooperative controllable confrontation test method based on a diffusion model. Comprising the following steps: S1, a diffusion model training stage: training a diffusion model based on real driving data, and learning vehicle behavior distribution through forward noise addition and reverse denoising to obtain fixed model parameters; s2, scene and multi-vehicle space-time modeling: fusing the road map, the lane topology and the vehicle state information, and constructing a space-time scene representation of multi-vehicle interaction as input of diffusion generation; s3, a diffusion generation mechanism based on adversarial guidance; s4, performing partial diffusion control and diversified confrontation generation; and S5, evaluating the authenticity and controllability of the multi-agent confrontation scene. Compared with the prior art, the method has the advantage that the authenticity, controllability and closed-loop consistency of multi-vehicle cooperative confrontation scene generation are obviously improved.

The invention provides a snakelike robot control method based on hierarchical reinforcement learning in a highly limited environment, and the method specifically comprises the steps: building a highly limited complex terrain simulation scene, and determining a task to be executed; a snake-shaped robot simulation model is established, and modeling is conducted on the structure, joints and kinetic parameters of the snake-shaped robot; a hierarchical control structure comprising a high-layer strategy network and a low-layer strategy network is designed based on a multi-layer perceptron and is respectively used for generating a global navigation sub-target and controlling specific actions; defining a state space, an action space and a reward function; in the training stage, a PPO algorithm is adopted to optimize the high-layer strategy and the low-layer strategy respectively, repeated tests are conducted through a simulation environment, strategy convergence is guided through a reward function, and finally the robot completes path planning and motion control only by depending on the motion state of the robot and obtaining the target position. Autonomous navigation and obstacle avoidance of the snakelike robot are realized, and the environmental adaptability of the snakelike robot is enhanced.

The invention provides an unsupervised industrial anomaly detection method based on an improved full-convolution cross-scale flow network. The method comprises the steps of performing data preprocessing on an industrial image data set; industrial image data is used as input, a pre-trained visual backbone network is used for extracting multi-scale features, a resolution-perceived channel attention module is used for enhancing expression of the multi-scale features, and an enhanced multi-scale feature map is obtained; performing multilayer reversible transformation by taking the enhanced multi-scale feature map as input and taking an ICSF-Net model based on hierarchical attention and expansion convolution as a cross-scale normalized flow network, and modeling multi-scale feature distribution; in the training stage, optimizing and updating ICSF-Net model parameters by maximizing the log likelihood of a normal sample under potential Gaussian distribution based on multi-scale feature distribution and combining an LSGR mechanism; in the reasoning stage, probability density estimation is carried out based on multi-scale feature distribution, an abnormal score graph is generated, and defect detection and positioning are achieved. According to the invention, the accuracy, robustness and pixel-level positioning precision of industrial product defect detection are improved.

The invention provides a storage battery health state assessment method fusing ultrasonic data and hybrid machine learning, and belongs to the technical field of storage battery health state assessment. The method focuses on performance comparison of an ANN model and a GWO-ANN model in storage battery SOH regression prediction, and finds that the GWO-ANN model accurately adjusts and optimizes ANN parameters by means of a grey wolf optimization algorithm, so that the storage battery health state assessment accuracy is improved. Compared with the prior art, the method has the advantages that higher prediction precision and generalization ability are shown, the R2 value is close to perfect, the error is extremely low, and the ANN model is remarkably surpassed, the HHT-GWO-ANN model is innovatively introduced, the model integrates the advantages of HHT and GWO, the calculation process is accelerated, the robustness and prediction precision of the model are enhanced, efficient and stable model construction is achieved, and the method is suitable for large-scale popularization and application. And the test performance of the method is consistent with that of the training stage, so that the practical potential of the method in storage battery SOH nondestructive testing is verified. The ultrasonic nondestructive testing method proposed by the invention shows unique advantages in SOH recognition of the storage battery, and overcomes the limitation of a traditional method.

The invention provides a large language model instruction optimization method based on constraint perception self-learning reasoning, which can be applied to the technical field of natural language processing and artificial intelligence. The method comprises the steps that in the cold start training stage, a thinking chain example generated by a pre-training language model is utilized to assist a target large language model in recognizing constraints in a training instruction set, and the constraints in the training instruction set comprise hard constraints and soft constraints; performing supervision and fine tuning on the target large language model by utilizing the thinking chain example and the training annotation data so as to enable a thinking chain generation process of the target large language model to meet a hard constraint and a soft constraint; in the self-learning training stage, a constraint satisfaction evaluation mechanism is constructed by using the hard constraint and the soft constraint, and a thinking chain generation strategy of the target large language model is optimized through a reinforcement learning algorithm based on the constraint satisfaction evaluation mechanism.

The invention discloses a zero-sample industrial defect detection method and device based on visual enhancement and a medium, and the method comprises the steps: collecting an industrial product image and a text semantic tag corresponding to the industrial product image, and constructing a training set and a test set; building a zero-sample industrial defect detection model based on visual enhancement, and training by using the training set; and inputting the test set into the trained model for reasoning and evaluation. The model is composed of a text encoder and an image encoder, and deep interaction and optimization of text and image features are realized by introducing a bidirectional cross attention mechanism, a multi-task fusion module and an image feature refining module. In the training stage, the model learns uniform defect features from part of industrial product categories; in the test stage, defect identification and positioning are carried out on the types of industrial products which are not contacted. According to the method, the dependence on manual labeling is effectively reduced, the adaptability and detection precision of the model in a complex industrial scene are improved, and the method has good generalizability and industrial deployment value.

The invention provides a VAE-ALSTM-based bridge structure state anomaly detection method and system, and the method comprises the steps: carrying out the standardization and sliding window segmentation of vibration monitoring data, extracting potential features through a variational auto-encoder, and carrying out the time sequence prediction and reconstruction in combination with an attention-enhanced long-short-term memory network, thereby achieving the detection of the abnormal state of a bridge structure. Fusing the prediction error and the reconstruction error to generate an abnormal score; a threshold value is automatically set through quartile distance statistics, and self-adaptive judgment under different bridge types is achieved; when the score crosses the boundary, real-time alarm is triggered; the system correspondingly comprises a data preprocessing and sample construction module, a VAE-ALSTM model construction module, a training stage prediction and reconstruction module, an error calculation and anomaly scoring module, a threshold setting module and an anomaly judgment and alarm module. The method and the system do not need manual threshold parameter adjustment, are high in precision and good in real-time performance, and can be widely applied to the fields of bridge health monitoring, operation and maintenance early warning and the like.

The invention discloses an infrared super-resolution imaging method and an infrared super-resolution imaging system based on nerve compression self-coding, in a training stage, a physical imaging process is modeled into an optical degradation encoder according to a compression sensing theory, and a fuzzy operation is combined by introducing a learnable binarization mask, so that a super-resolution image is obtained. Simulating a sparse coding process of compressed sensing to obtain sparse representation of the input image; restoring the sparse representation into a high-resolution image by using a CNN-based super-resolution decoder; in the reasoning stage, the optical degradation encoder is replaced by a real sampling process, and the optimized binarization mask is used as a mask set in the spatial light modulator and is finally projected to an image collected by a sensor; and the trained CNN model is also stored and is used for carrying out a super-resolution task on a really captured coded image. According to the invention, an efficient and reliable solution is provided for super-resolution reconstruction of the low-resolution infrared image.

The invention discloses a three-dimensional point cloud geometric information compression method based on implicit neural representation, and the method comprises the steps: constructing a trunk structure of an implicit neural network through a plurality of sine representation network layers which are connected in series, and introducing a variable-scale position coding mechanism on this basis, the method enables a network to obtain higher geometric reduction precision while keeping a compression ratio, and comprises the following steps: (1) inputting space coordinates of divided voxels into a position coding module with adjustable scale parameters; (2) feeding a coding result into an implicit neural network constructed by a network layer based on sine representation, and outputting the occupancy probability of the voxel through an activation function; (3) in a training stage, the model continuously optimizes parameters, so that the output probability distribution is highly consistent with a real occupied label; (4) after model training is completed, a method of combining an AdaRound second-order quantization optimization strategy and quantization perception training is introduced, network weight is finely adjusted, and quantization errors are reduced; (5) in a reasoning stage, judging whether the voxel is occupied or not according to a preset threshold value, and when the prediction probability exceeds the threshold value, regarding the voxel as occupied; and (6) all voxels judged to be occupied are aggregated, and reconstruction of the geometric structure of the point cloud is completed.

A system for training and deploying a sanitizer model of a large language model (LLM) to sanitize the harmful language text. The system generates paired data sets for training using LLM techniques. Training includes a first training phase using a paired dataset to predict the benign language text in the paired dataset based on loss of information computed between the benign language text in the paired dataset and the sanitized language text. Training also includes a second training phase using a harm model that computes a harm score of the LLM prediction by adjusting the trained sanitizing step to reduce the harm score for subsequent LLM responses.

The invention relates to the field of artificial intelligence deep learning model optimization, and discloses an ultra-low parameter efficient fine tuning method based on sparse frequency domain projection. Comprising the steps of setting an upper projection matrix which is frozen after random initialization in an attention mechanism layer or a feedforward mapping layer of a pre-training model for mapping a low-dimensional reconstruction result to an original weight dimension; constructing a sparse spectrum matrix in a frequency domain, generating a non-zero index set according to a frequency priority and a random sparse mixed strategy, and keeping the whole training process fixed; performing inverse transformation on the sparse spectrum matrix to obtain a time domain reconstruction matrix, and generating a weight increment through linear mapping of an upper projection matrix; only updating the non-zero spectrum coefficient under the limited parameter budget, and keeping the pre-training weight and the upper projection matrix frozen; optionally, spectrum regularization, energy constraint and gradient clipping are applied in the training stage, and quantification and distillation joint optimization is performed in the reasoning stage.

A system generates an MLM comprising a plurality of layers. The system assigns a first encryption scheme for a first subset of layers in the plurality of layers. During a training phase of the MLM, the system determines whether a first input training vector comprises private data, in response to determining that the first input training vector does not comprise the private data, the system train the MLM such that, during backpropagation, an optimization algorithm is used to update any necessary weights in the plurality of layers; and in response to determining that the first input training vector comprises the private data, the system trains the MLM such that during the backpropagation, the optimization algorithm is used to update weights solely in the first subset of layers. The system executes the trained MLM on a user input vector to generate a user output value.

Systems and methods for object tracking are described. One or more aspects of the systems and methods include receiving a video depicting an object; generating object tracking information for the object using a student network, wherein the student network is trained in a second training phase based on a teacher network using an object tracking training set and a knowledge distillation loss that is based on an output of the student network and the teacher network, and wherein the teacher network is trained in a first training phase using an object detection training set that is augmented with object tracking supervision data; and transmitting the object tracking information in response to receiving the video.

The invention discloses an unmanned cluster dynamic task game confrontation method based on MAPPO deep reinforcement learning of Actor-Critic. An MAPPO deep reinforcement learning strategy of Actor centralized training and Critic distributed execution is adopted as a training algorithm of unmanned cluster dynamic task game confrontation. According to the method, centralized training is carried out by using global information in a training stage, and distributed execution is carried out by only using local observation information and historical information in an execution stage, so that the task can be completed through deep cooperation among multiple agents; the problem that the task completion efficiency is low due to the fact that the flight path drifts and jitters and is prone to falling into a local optimal solution during training of a traditional algorithm is solved, multi-aircraft cooperative path planning can be effectively achieved, and the learning efficiency and stability are improved in a complex multi-agent scene.

The invention provides a fan fault diagnosis method for recognizing a vibration map based on a convolutional neural network, and relates to the technical field of neural networks, and the method comprises the steps: obtaining a multi-dimensional vibration signal in the operation process of a fan, and generating a two-dimensional vibration map through time-frequency transformation; and a convolutional neural network is utilized to automatically extract multi-layer time-frequency features and realize fault category discrimination. In the training stage, parameter optimization is carried out based on known fault samples, in the reasoning stage, real-time signals are input into a trained model to obtain fault type probability distribution, the fault type is determined according to the maximum probability, a fault evolution result is generated in combination with the historical operation trend, and therefore automatic, intelligent and rapid diagnosis of fan faults is achieved.

The invention discloses a class increment image classification method and system based on a multi-modal pre-training model, and the method comprises the steps: obtaining a class increment learning data set, and for each task, model training comprises two stages: an intra-task training stage and a cross-task fine tuning stage; in-task training stage: for the data of the current task, performing fine adjustment on the pre-trained visual language model by adopting a task-specific adapter to realize classification among categories in the task; a cross-task fine tuning stage: introducing a mapping module for image feature expression, mapping features of a specific feature space of a task to a feature space shared by the task, and realizing cross-task category separability; during reasoning, a reasoning strategy based on prediction uncertainty is adopted for image classification. According to the method, the problem of category confusion existing across tasks can be solved, and the accuracy of selecting the output features is improved.

The invention discloses a sparse view angle 3D-DSA reconstruction method based on a three-dimensional Poisson generative model. The method comprises the following steps: obtaining pairing data of a sparse view angle 2D-DSA projection drawing and a 3D-DSA reconstruction image; extracting features of a projection image by using a projection domain encoder, then converting features of two-dimensional projection into a three-dimensional image domain according to a geometrical relationship of the cone beam CT, and then obtaining a prior image by using an image decoder; the method comprises the following steps: constructing a three-dimensional Poisson generation model, adding noise to a three-dimensional image patch in a training stage to obtain a disturbance image, outputting a network reconstruction image by taking a prior image as a condition and the three-dimensional image patch before noise addition as a target image, and calculating loss of the output image and the target image to update network parameters; the mean square error loss and the mean square error loss of the maximum intensity projection images of the three orthogonal planes are used during loss calculation; in the sampling stage, random noise is used as input, a prior image is used as a condition, noise of a noise image is continuously denoised within a limited step length, and finally a reconstructed 3D-DSA image is obtained.

The invention discloses a distributed damage identification method based on self-supervised learning, and the method comprises the steps: collecting multi-dimensional acceleration data through a sensor, and carrying out the preprocessing; constructing a self-supervised damage identification neural network, inputting the preprocessed data for self-supervised learning, and mining low-dimensional compression features, spatial features and time features in the data; training a self-supervised damage identification neural network to realize compression reconstruction, sensor label prediction and single-step time prediction; identifying whether damage occurs or not according to the sensor label prediction result, and if the sensor prediction label is consistent with the serial number of the sensor, indicating that no damage occurs; identifying the damage degree according to the average value of the sensor reconstruction error, the label prediction error and the actual prediction error, wherein the larger the average value is, the more serious the damage is. According to the method, the spatial correlation relation is learned in the training stage, so that data exchange in the test or use stage is avoided, and the accuracy and robustness of distributed damage identification can be effectively improved.

The invention relates to an injection molding product quality prediction method based on a random forest algorithm. The method comprises a model training stage and a prediction stage. The model training stage comprises the following steps: acquiring a historical data set, wherein the historical data set comprises a process parameter set and a corresponding product quality label; performing principal component analysis on the historical data to obtain a principal component set and a first transformation matrix; screening the principal component set through a gradient lifting decision tree algorithm; training the random forest prediction model by using the sample data set; the prediction stage comprises the following steps: collecting a process parameter set in a production process of a to-be-predicted product; performing transformation and screening; and predicting the product quality by using the trained random forest prediction model. According to the invention, a prediction model having significant advantages in the aspects of data dimension reduction, feature selection and nonlinear data processing can be provided, the accuracy of quality prediction of the injection product is improved, and the processing quality of the injection product is improved in the aspect of improving real-time process parameters.

The invention discloses an open vocabulary object detection method based on reasoning score adjustment, and belongs to the technical field of object detection in machine vision. The method aims at solving the problem that in existing open vocabulary object detection, a model is poor in novel category detection capacity. Comprising the steps of obtaining an open vocabulary object detection pre-training model in a training stage; in the reasoning stage, an open vocabulary object detection pre-training model is adopted to obtain mapping features of a detection image, and cosine similarity calculation is carried out on the mapping features and coded category labels to obtain an original cosine similarity score; obtaining prototype visual representation of a novel category, replacing the corresponding mapping features with the prototype visual representation of the novel category according to a calculation result related to the novel category in cosine similarity calculation, and performing cosine similarity calculation with the encoded category label by adopting the prototype visual representation of the novel category to obtain an adjusted cosine similarity score; and combining the adjusted cosine similarity score with other original cosine similarity scores to obtain an object detection result. The method is used for open vocabulary object detection.

The invention provides an image deblurring method based on amplitude, phase and time domain channel fusion, a network adopts a U-shaped architecture, an adaptive amplitude and phase compensation strategy is introduced, and information weights of different frequency bands are dynamically adjusted on a channel level; the method comprises two stages and five steps: a training stage: S1: in the training stage, obtaining an image deblurring data set and preprocessing an image; s2, constructing an image deblurring model based on amplitude phase and time domain channel fusion; s3, optimizing a loss function; s4, training and constructing an image deblurring type based on amplitude phase and time domain channel fusion; and S5, performing performance evaluation on the training model in the test stage. According to the method, the time domain and frequency domain feature enhancement strategies are combined, the image definition is recovered, meanwhile, the influence of low resolution and blurring is effectively reduced, and therefore the overall performance of a deblurring task is improved.

The invention discloses an aircraft cluster multi-task scheduling system based on a cooperative game and a working method thereof. The system comprises a simulation environment module, a hierarchical strategy network module, a centralized value network module and a training and execution module. The simulation environment module is used for constructing a multi-agent air combat confrontation environment and generating states, rewards and interaction data required by training; the hierarchical strategy network comprises a shared space-time representation encoder f theta (.), a high-layer strategy network pi H (aHz) and a low-layer strategy network pi L (aLz; aH); the centralized value network is used for receiving global state information in a training stage, estimating the overall return of our agent cluster, calculating a dominant function, and realizing the optimization of the network by minimizing the value loss; and the training and execution module optimizes a centralized value network parameter and hierarchical strategy network parameters theta H and theta L by using a global state St in a centralized training stage, and outputs an air combat decision ai according to local observation independent decisions of each agent in a distributed execution stage.

The invention provides a power distribution network voltage control method based on multi-agent deep reinforcement learning, and relates to the technical field of power distribution networks, and the method comprises the steps: dividing a power distribution network into a plurality of regions, and carrying out the centralized management and optimization of the voltage of each region; a cooperative voltage control problem between regions is converted into a Markov game so as to realize effective cooperation between intelligent agents; a centralized training and decentralized execution framework is adopted, so that the intelligent agent fully utilizes global information in a training stage and independently makes a decision according to local observation in an execution stage; a self-attention mechanism is integrated, so that the agents are more focused on information related to self rewards, a multi-agent reinforcement learning technology can be combined with the self-attention mechanism, the problem of voltage fluctuation in a power distribution network is effectively solved, stable control and optimal management of voltage are realized, and the power distribution network quality is improved. The voltage regulation precision and response speed of the power distribution network are improved, and the stability and reliability of a power system are ensured.

There is provided a system and method of runtime defect detection in a semiconductor specimen. The method includes obtaining a plurality of runtime images acquired for a plurality of dies on the specimen, feeding the plurality of runtime images to a plurality of input channels of a neural network (NN) in an input order, wherein the NN is previously trained in a training phase, and processing, by the NN, the plurality of runtime images simultaneously, to obtain a plurality of defect maps, each corresponding to a respective runtime image and indicating probabilities of defect candidate presence thereof. Each given runtime image is processed as a target image using remaining images in the plurality of runtime images as reference images of the target image, and the defect map of the target image remains invariant, irrespective of changes to the input order.

The invention provides a model deployment method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: loading weight data of a target model; the data precision type of the weight data stored in the current computing equipment is an integer type; performing inverse quantization on the weight data based on the first floating-point number type to obtain weight data after inverse quantization; based on the inversely quantized weight data and the input data, performing reasoning calculation of the target model in the current calculation equipment; the first floating-point number type is a data precision type of model reasoning executed by the current computing device; the data precision type of the weight data of the target model in the model training stage is a second floating-point number type; the data precision type of the weight data is converted from a second floating-point number type to an integer type prior to storing the weight data to the current computing device. According to the method and the device provided by the invention, the model is deployed in the computing equipment which does not support the low-level floating-point number type.