403 results about "Sample classification" patented technology

The invention discloses a semi-supervised identification method and system for a small target of a container, and belongs to the technical field of artificial intelligence and computer vision, and the method comprises the steps: carrying out the target detection of a container image through a pre-trained target detection model, intercepting a sub-image, and inputting the sub-image into an initial classification model, and obtaining a classification confidence coefficient; the uncertainty of the model on a sample classification result is quantified through a Monte Carlo Dropout method; a feature space distance filtering and dynamic threshold adjusting mechanism is combined, and samples with high confidence, low uncertainty and consistent feature space are screened out to serve as pseudo label data; pseudo label data and initial synthesis data are mixed, and the generalization ability of the model is gradually improved through semi-supervised iterative training. According to the method, the dependence on manual annotation can be remarkably reduced, meanwhile, the distribution difference between synthetic data and real scene data is gradually reduced, and finally, high-precision recognition and strong generalization ability of a classification model in a real scene are achieved.

Methods and apparatus to determine machine learning (ML) configurations for classifying malware are disclosed. An example server comprises interface circuitry, machine readable instructions, and programmable circuitry to at least one of instantiate or execute the machine readable instructions to determine a computing parameter associated with a computing device, the computing device communicatively coupled to the server, select a ML model to deploy on the computing device based on the computing parameter, determine a configuration of the ML model based on the computing parameter, deploy the ML model to the computing device, and cause the deployed ML model to classify a sample as clean or malicious, the sample received at the computing device.

The invention discloses an OCR (Optical Character Recognition)-based document automatic recognition intelligent management system, and relates to the technical field of intelligent document OCR processing. A preprocessing module adopts adaptive filtering and a GAN (Generic Area Network) to repair low-quality documents and separate seals from characters; the recognition engine realizes end-to-end recognition of inclined characters and mixed characters through a multi-scale feature pyramid network, and the semantic module constructs a dynamic knowledge graph and supports event evolution reasoning and cross-year correlation analysis; the classification module realizes small sample classification based on a prototype network, the security module realizes fine-grained authority control and anomaly detection through a block chain and federated learning, the text edge definition is improved after system processing, and real-time reasoning is supported. The intelligent level of document processing is improved through cooperation of multiple modules, the preprocessing and recognition module solves the complex scene problem, and deep analysis and correlation analysis are achieved through semantic understanding and a knowledge graph; the security module guarantees data security; and the efficiency and decision scientificity are comprehensively improved.

The invention relates to a Han-Cross industry text classification method and system based on prompt learning and adaptive loss weighting, and belongs to the technical field of natural language processing. The method comprises the following steps: designing and constructing a universal prompt template; the method comprises the following steps: recombining a Han-Cross cross-border industry text classification data set, namely converting an original single sample into paired samples; in a few-sample and multi-language scene, related vocabularies are adopted as external knowledge resources, and vocabularies most related to the mapping labels are retrieved from the related vocabularies; expanding the vocabulary mapper by introducing synonyms and associated vocabularies; adopting a dynamic mixed loss function and applying the dynamic mixed loss function to a pre-training language model to optimize a few-sample classification task; and classifying Chinese and Vietnamese cross-border industry texts by using the optimized pre-training language model. The method shows a remarkable effect in Chinese and Vietnamese industry text classification tasks, and is particularly suitable for a few-sample learning scene with data scarcity and language imbalance.

The invention relates to the technical field of patent information analysis, in particular to a patent image few-sample classification method based on multi-modal representation fusion. The method comprises the following steps: constructing a hybrid pre-training model by designing a language branch, a visual branch and a visual auxiliary branch; obtaining input corresponding patent image few samples, and carrying out image feature parallel extraction and text description feature generation to obtain patent image text prompt description features; a cross-modal fusion layer is designed by adopting an attention mechanism, and multi-modal representation fusion and feature fusion loss calculation are carried out, so that comparative learning loss and visual auxiliary branch loss between a language branch and a visual branch are obtained; and performing model fusion optimization on the mixed pre-training model to output fusion features, and constructing a patent image few-sample classification framework to execute corresponding patent image few-sample classification work. The feature representation corresponding to the patent image sample can be fully fused and learned to improve the classification effect of the model.

A processor may gather raw data comprising a plurality of characteristic data samples of a target user group. The processor may categorize the characteristic data samples into a plurality of user-related classes and triggers. The processor may build an input property graph for each characteristic data sample. The processor may augment the input property graph by a concept of hierarchies. The processor may determine a modification vector from the augmented input property graph. The processor may train an encoder / decoder combination machine-learning system. An embedding vector and a modification vector are used as input for the decoder to build a trained machine-learning generative model.

The invention discloses a few-sample classification method research based on feature cache optimization, and the method comprises the steps: (1), employing an improved ResNet-based SwAV model as a backbone, and enabling the SwAV model to be used for generating auxiliary features; (2) taking the image data and the text description as original input, and generating text input with rich downstream language semantics by utilizing GPT-3 to serve as text prompt of a CLIP model; (3) through a feature selection method based on feature similarity and difference, the problems of feature redundancy and inaccurate selection in the feature selection process are solved, feature dimensions with high selection value are identified, and normalization and enhancement zooming processing are performed on the features; and (4) utilizing visual contrast knowledge of SwAV, introducing a learnable cache model, and adaptively mixing prediction results from CLIP and SwAV. According to the method, under the CLIP framework without extra training, the accuracy of few-sample image classification is effectively improved through feature cache optimization and a self-adaptive hybrid strategy.

The invention provides a collaborative classification method and system fusing advantages of large and small models, and the method comprises the steps: inputting to-be-classified data into a trained zero-sample classification small model, and outputting candidate label domain screening information and an initial classification result; inputting the to-be-classified data, the candidate label domain screening information and the initial classification result into the trained large model, and outputting a final classification result; the training process of the small model comprises the following steps: inputting training data into the zero sample classification small model to obtain a preliminary prediction result; screening and obtaining pseudo label data based on an active learning strategy; checking and re-marking the pseudo-label data by using the large model to obtain a modified pseudo-label data set; and carrying out iterative training on the zero sample classification small model by utilizing the modified pseudo label data set. The method has the advantages that a small model has classification performance close to that of a large model while keeping lightweight calculation characteristics; the calculation burden of a large model is reduced, and the accuracy of classification decision is improved.

A method for technical issue resolution. The method includes: receiving, from a user, a text query concerning a technical issue; obtaining query-related context relevant to the text query; and processing, through a large language model (LLM), the text query and the query-related context to produce a multimodal query response used by the user to address the technical issue. More specifically, embodiments described herein utilize text topic and zero shot classification models to translate multimodal technical documentation (e.g., including text and images) into topic relevant metadata; and process queries, pertaining to technical issues, using a multimodal LLM provided with query-related text and image context derived from said topic relevant metadata.

The invention discloses an image classification method and device, equipment, a medium and a product, and relates to the field of image classification, and the method comprises the steps: obtaining to-be-classified image data; according to the to-be-classified image data, performing classification by using a classification model to obtain a classification category; the classification model is a trained pulse neural network; the spiking neural network comprises a plurality of neural layers; each neural layer comprises a plurality of neurons; each neuron comprises an emission threshold value and an input resistance; the training process of the classification model specifically comprises the following steps: taking sample image data as the input of a spiking neural network, taking a sample classification category as the output of the spiking neural network, and determining a total loss function of the spiking neural network according to cross entropy loss and adaptive sparse loss; and optimizing the synaptic weight, the emission threshold and the input resistance of the spiking neural network by using a back propagation algorithm to obtain a classification model. The method can improve the classification precision.

The invention discloses a visual analysis and screening method for fish maw appearance features, and relates to the technical field of computer vision and intelligent sorting, and the method comprises the steps: carrying out the polarization reflection feature analysis of a fish maw main region image, constructing a surface direction-reflection relation tensor, and carrying out the clustering to generate an anisotropic reflection texture feature pattern; gradient pyramid fusion is carried out on the anisotropic reflection texture feature map and the RGB image, contour information is extracted, and a fish maw boundary enhanced image is generated; inputting the fish maw boundary enhanced image into a contour detection network, performing automatic modeling, outputting a high-precision boundary thermodynamic diagram, and generating a high-precision boundary contour diagram; and carrying out feature fusion on the anisotropic reflection texture feature map and the high-precision boundary contour map, constructing a fish maw appearance description vector, carrying out appearance grade classification through a multi-layer neural network model in combination with a historical evaluation sample, and outputting a fish maw sample classification result. The accuracy and efficiency of fish maw quality detection are remarkably improved.

The embodiment of the invention provides a backdoor attack method and device, processing equipment, a program product and a medium, and is applied to the technical field of artificial intelligence. The method comprises the steps that target attention distribution corresponding to a target vision converter model is constructed based on a sample set of a target category, and the target vision converter model is trained to be capable of classifying all samples in the sample set into the target category; based on the target attention distribution and the to-be-processed source image, generating a trigger corresponding to the to-be-processed source image; based on a dual-objective loss function, parameters of the trigger are iteratively optimized, a target trigger is obtained, and the dual-objective loss function is constructed based on Wasserstein distance and classification loss; and generating a poisoning sample corresponding to the to-be-processed source image according to the target trigger and the to-be-processed source image. By the adoption of the method, the problem that an existing backdoor attack method is difficult to adapt to a self-attention mechanism, and consequently efficient hidden attack on a visual converter model is difficult to achieve is solved.

The invention relates to a road crack detection method and system based on a zero-sample large language model, and relates to the technical field of computer vision and artificial intelligence. The method comprises the steps that a road surface image is collected through an unmanned aerial vehicle or a vehicle-mounted camera, after denoising and contrast adjustment preprocessing, visual features are extracted through a multi-modal model and converted into text description, and an image-text embedding vector is generated; semantic reasoning is performed on the embedded vector based on a large language model, zero sample classification is realized through similarity matching of predefined text tags, and crack types and severity are identified; and generating a restoration suggestion in combination with the road environment and the traffic load data, and outputting a detection result and a user feedback optimization model through a visual interface. The method can realize high-precision crack detection without labeling data, has the advantages of strong generalization, multi-modal fusion, intelligent decision support and the like, and can be widely applied to health monitoring and maintenance of infrastructures such as roads, bridges and the like.

The invention discloses a semi-automatic labeling method and system for rail transit engineering construction video images, and the method comprises the steps: removing redundant frames from a video stream, extracting key frames, and forming a block index set; and driving the multi-modal large model to pre-annotate the image by using the constructed cue word, and carrying out binarization processing on an annotation result. Based on active learning, migrating unmarked samples to a marked set for multiple times and training a key sample classifier CD until the scale of the marked set reaches the standard; meanwhile, a confidence classifier CC is trained by comparing manual and pre-labeling results. And for residual samples in the unmarked set, after the residual samples reach the standard through a CC precision test, marking tasks are divided according to a threshold value theta: high-confidence samples are pre-marked, and low-confidence samples are manually marked. And finally, updating the set of the manually labeled samples, and determining whether to retrain the model or not according to category distribution. On the premise of ensuring the labeling quality, the blindness of manual labeling is effectively reduced, and the efficiency of the labeling process is improved.

The invention discloses a small sample abnormal signal identification method and system, a storage medium and a program product. The method comprises the following steps: acquiring a to-be-tested signal; identifying a to-be-detected signal by using the abnormal signal identification model; wherein the abnormal signal identification model comprises an isolated forest model and a small sample classification model. According to the method, the to-be-detected signal in the actual industrial field can be accurately judged, and the recognition reliability is effectively improved.

The invention provides a multivariate data-based printing machine top project approval design method, which relates to the technical field of printing equipment, and comprises the following steps of: acquiring printing machine operation data comprising printing quality parameters, equipment working condition parameters and environment parameters; performing feature analysis and parameter extraction through an adaptive fuzzy clustering algorithm, dynamically adjusting a sample classification membership degree, identifying a key parameter combination influencing the printing quality, and constructing a correlation matrix; constructing a dynamic Bayesian probability network model based on a parameter mapping relation, constructing a core network structure through an information entropy value and information gain, and establishing probability association between a printing quality parameter and an equipment working condition parameter; and determining the optimal parameter configuration of the function modules of the printing machine by utilizing a bidirectional constraint dynamic programming algorithm and combining the coupling weight and the state transition equation among the function modules, and generating a top design scheme.

Some embodiments of the current disclosure disclose methods and systems for training for training a natural language processing intent classification model to perform few-shot classification tasks. In some embodiments, a pair of an utterance and a first semantic label labeling the utterance may be generated and a neural network that is configured to perform natural language inference tasks may be utilized to determine the existence of an entailment relationship between the utterance and the semantic label. The semantic label may be predicted as the intent class of the utterance based on the entailment relationship and the pair may be used to train the natural language processing intent classification model to perform few-shot classification tasks.

The invention discloses a positive and negative unmarked image classification method and system based on depth metric learning, and relates to the technical field of sample classification, and the method comprises the steps: obtaining a marked positive sample image and an unmarked image, inputting the marked positive sample image and the unmarked image into a depth metric learning model, adjusting the marked positive sample image in a depth feature space, and obtaining the unmarked image; minimizing the distance between the marked positive sample image and the center of the positive sample image, aligning the unmarked image and the corresponding enhanced view in a representation space through self-consistent metric learning, and outputting related feature representation; using related feature representation to select sample images which are in multiple proportion to the positive sample images as reliable negative samples; and taking the related feature representation as input, using a positive sample image and a reliable negative sample as supervision signals to train a binary classifier, and classifying positive and negative unmarked images to be detected. According to the method, dependence on heuristic negative sample selection is avoided, and high-quality feature representation with discrimination on positive and negative samples can be learned.

The invention relates to the technical field of ore raw material analysis, and discloses an intermediate fusion strategy ore sample classification method based on random model combined data, and the method comprises the following steps: S1, collecting LIBS spectrum and Raman spectrum data of a manganese ore sample; s2, preprocessing the spectral data; s3, performing dimensionality reduction on the preprocessed LIBS and Raman data by adopting principal component analysis; s4, splicing the feature matrixes after dimension reduction to construct an intermediate fusion data set; s5, training a random forest classification model based on the intermediate fusion data set; and S6, carrying out classification and identification on the manganese ore samples by utilizing the trained model. According to the method, metal element characteristic spectral lines captured by the LIBS spectrum and molecular vibration characteristics identified by the Raman spectrum are organically integrated through an intermediate fusion strategy, two-dimensional identification is achieved, redundant information is effectively eliminated while main characteristics of the spectrum are reserved, and the accuracy and reliability of manganese ore sample classification are remarkably improved.

For ticket classification in a building automation system, a large language model (LLM) is used to classify. In one approach, a prompt is generated for zero-shot classification, and a prompt is generated for few-shot classification. In another approach, a hybrid annotation provides corrections (review) by an expert to correct LLM classification for sample tickets to be used as examples in the few-shot classification. The LLM may operate on a diverse and complex range of tickets in an efficient and scalable manner.

The invention provides an auxiliary rock core geological logging method based on drilling parameters, and relates to the technical field of geological investigation, the method is applied to a logging control terminal, and the method mainly comprises the following steps: obtaining regional typical rock core samples and five types of drilling parameter signals, classifying and extracting static characteristics of the rock core samples, and carrying out parallel noise reduction analysis on the parameter signals; obtaining a matching relationship between the rock core and the parameter signal, and constructing a lithologic parameter bidirectional association reference library; on the basis of the matching relation of the lithologic parameter bidirectional association reference library, the model is embedded into an incremental learning module to update the weight, and a subsection lithologic pre-judgment result is output; and calling a preset catalog template to load a pre-judgment result, generating a catalog report after man-machine interaction recheck and correction, collecting correction data, reversely transmitting the correction data back to the reference library to update a matching relationship, and triggering incremental learning of the model to complete a closed loop. The technical problems that in traditional rock core logging, regional lithology adaptation is poor, parameter interference is large, a model is not dynamically optimized, the deep operation risk is high, and efficiency is low are effectively solved.

The invention relates to a hyperspectral image classification method based on a bidirectional interactive fusion space-spectrum multi-order gating aggregation network, and belongs to the field of remote sensing image processing. The method comprises the steps of determining a data set; data preprocessing: performing sample block extraction on the hyperspectral data set, and dividing the hyperspectral data set into a training set, a verification set and a test set; network construction: constructing a space-spectrum multi-order gating aggregation network based on bidirectional interactive fusion, wherein the space-spectrum multi-order gating aggregation network is used for hyperspectral image classification; the hyperspectral samples in the training set are input into the constructed network in batches for training, and after each training batch is completed, the classification performance is evaluated by using the verification set samples; and sample classification: inputting the hyperspectral samples in the test set into the trained classification network to obtain a final classification result. According to the method, efficient feature extraction can be realized, high-accuracy classification is performed on the hyperspectral image, and the method can be widely applied to remote sensing application fields such as hyperspectral image surface feature category detection and recognition.

The invention provides a text classification method and device, electronic equipment and a storage medium, and relates to the technical field of natural language process.The method comprises the steps that a training sample set is obtained, and the training sample set comprises sample texts and sample classification labels and sample reasoning reasons corresponding to the sample texts; performing fine tuning on a first pre-trained large language model through the training sample set to obtain a text classification model; in response to a text classification request, through the text classification model, based on a preset reasoning constraint parameter, only performing classification processing on a to-be-classified text to obtain a prediction classification label; wherein the preset reasoning constraint parameters comprise an output length limiting parameter and a logit probability intervention parameter. According to the method, the text classification efficiency can be improved while the accuracy and the reliability of a text classification result are improved.

The embodiment of the invention provides a tool calling method, a tool calling model training method and a data generation model training method. The tool calling model training method comprises the steps that a target intention and target tool calling parameters corresponding to the target intention are determined; the target intention is input into a data generation model, target query data corresponding to the target intention is obtained, the data generation model is obtained by training an initial data generation model through a positive and negative sample pair composed of an intention sample and a query data sample, and the positive and negative sample pair is determined in a preset sample classification mode; the query data sample is determined by performing data enhancement on the intention sample; inputting the target query data into the initial tool calling model to obtain a prediction tool calling parameter; according to the prediction tool calling parameter and the target tool calling parameter, training an initial tool calling model to obtain a tool calling model; the tool calling model is obtained by training the high-quality target query data, so that the accuracy and efficiency of processing the query data by the tool calling model are improved.

A method is proposed for sample processing. A first group of data are received, here data in the first group of data comprises a sample and a classification of the sample, and the classification belonging to a first group of classifications in a plurality of classifications associated with the data. A plurality of data with the classification are selected from the first group of data. A first and a second loss function are determined for training a classification model that represents an association relationship between samples and classifications of the samples based on a plurality of samples comprised in the plurality of data and the classification, the first and second loss functions represent classification accuracy and a feature distribution for the classification model. The classification model is trained based on the first and second loss functions. Therefore, the accuracy of the classification model may be increased.

The invention provides a model training method, device and equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring an image sample corresponding to a label type, and classifying the image sample through an image classification model to obtain a first classification probability of the image sample; determining a loss gradient corresponding to the label type based on the first classification probability of the image sample corresponding to the label type; determining a first loss based on the loss gradient corresponding to the label type, and determining a second loss based on the loss gradient corresponding to the target label type and the loss gradient corresponding to the label type; and performing fusion processing on the first loss and the second loss to obtain third loss, and updating the image classification model based on the third loss to obtain an updated image classification model. According to the method and the device, the model can equally learn all the label types when the data of the image samples corresponding to the label types are not uniformly distributed, and the situation that the high-frequency label types dominate the optimization direction of the model is avoided.

The invention belongs to the technical field of hyperspectral image small sample classification, and particularly discloses a hyperspectral image small sample classification method based on joint domain adaptive weight self-learning, which uses a spectrum-space attention double-branch dense network to avoid gradient disappearance. Meanwhile, a channel and a space attention mechanism are combined to efficiently extract spectrum-space discrimination features of the hyperspectral image; a domain projector and a domain discriminator are introduced to process significant distribution differences existing in different domains of the hyperspectrum; then designing an adaptive learner to carry out weight adjustment, realizing effective unification of joint domain adaptation, and further realizing domain distribution alignment efficiently; and finally, performing small sample classification of the hyperspectral image of the target domain by utilizing metric learning. According to the method, feature alignment and adjustment are carried out in different directions by using a joint domain adaptation technology, and inter-domain distribution differences can be better processed, so that the generalization ability and robustness of a model facing an unknown domain are improved, and the rationality and quality of a hyperspectral image small sample classification result are improved.

The invention belongs to the technical field of spectral analysis, and particularly relates to a coal coke residue sample classification detection method and system, and the method comprises the steps: obtaining the noise reduction spectrum of each coke residue sample; according to the difference between the spectrums of the coke residue samples and the corresponding noise reduction spectrums and the fluctuation frequency of the spectrums of the coke residue samples, noise indexes of all wavelengths of the spectrums of all the coke residue samples are obtained; according to the law of the difference of the noise indexes of each wavelength in different coke slag samples of the same kind of coke slag, obtaining the structural performance of each wavelength; according to the absorbance difference of the same wavelength of different coke residues and the absorbance difference of the same wavelength of coke residue samples of the same type of coke residues with different crushing degrees, the component expression of each wavelength is obtained; and in combination with the structural expression and component expression of the wavelength, obtaining a characteristic wavelength, and establishing a classification detection model of the coke residue sample. According to the method, the classification detection model of the coke residue samples is constructed, and the accuracy and the detection efficiency of classification detection of the coke residue samples are improved.

A method is proposed for sample processing. A first group of data are received, here data in the first group of data comprises a sample and a classification of the sample, and the classification belonging to a first group of classifications in a plurality of classifications associated with the data. A plurality of data with the classification are selected from the first group of data. A first and a second loss function are determined for training a classification model that represents an association relationship between samples and classifications of the samples based on a plurality of samples comprised in the plurality of data and the classification, the first and second loss functions represent classification accuracy and a feature distribution for the classification model. The classification model is trained based on the first and second loss functions. Therefore, the accuracy of the classification model may be increased.

The invention discloses a pork safety traceability quality nondestructive testing method and system for livestock breeding. The method comprises the following steps: constructing a stable detection environment by utilizing a constant temperature and humidity device, a dynamic environment monitoring unit and an optical shielding material; a Fourier transform infrared spectrometer is adopted to collect spectral data, and a non-contact texture detection device is combined to measure physical parameters; denoising, standardization processing and feature fusion are carried out on the collected data, and key variables are extracted through principal component analysis; grouping the samples by using a K-means algorithm, and verifying sample classification by using linear discriminant analysis; quantitative prediction and classification of pork quality are realized based on partial least squares regression and a support vector machine model; the model is applied to an actual scene, real-time processing and visual display are achieved through online data collection, and a data sample library is continuously expanded. The method has the advantages of high detection speed, high precision, wide application range and the like, and can be widely applied to the fields of meat quality detection and food safety.