40 results about "Sample classification" patented technology

Embodiments of the present application provide a method, device, medium, equipment and product for classifying wine samples, relating to the technical field of artificial intelligence. The method comprises: contacting odor molecules of a wine sample with at least one recombinant cell expressing an olfactory receptor and a reporter protein, the reporter protein generating a detectable signal after the receptor binds to the odor molecules; obtaining time series data of the signal, calculating the maximum value and baseline value thereof, and determining a response characteristic value of each olfactory receptor according to the same; and inputting the characteristic value into a trained machine learning model to obtain a wine sample classification result. The present method accurately extracts a stable characteristic value of the response of an olfactory receptor to a wine sample, thereby eliminating interference caused by initial value fluctuations and improving the accuracy of model classification.

Disclosed is a loop filter method based on an adaptive pixel classification criterion. The loop filter method based on an adaptive pixel classification criterion in an image decoding device includes a step of classifying a restored sample according to an absolute classification criterion or a relative classification criterion, a step of obtaining offset information according to a result of classifying the restored sample, a step of adding the offset value to the restored sample with reference to the obtained offset information, and a step of outputting the restored sample to which the offset value is added. Thus, errors of the restored sample can be corrected.

The application provides a sample classification method and device for first arrival wave deep learning, and relates to the technical field of oil exploration. The method comprises the following steps: acquiring training sample data and classification information to which the training sample data belongs; the training sample data comprises first arrival sample data and first arrival label data corresponding to the first arrival sample data; a classification network model is trained by using the training sample data and the classification information to obtain a training result; and the training result is used for classification prediction processing on to-be-classified first arrival sample data to obtain a first arrival sample classification result. The application can accurately and efficiently classify and screen the first arrival sample data, enriches the type feature information of the first arrival sample data, and improves the pertinence and effectiveness of first arrival wave sample deep learning.

The application provides an industrial defect sample classification method and system based on prototype learning, comprising: acquiring an industrial defect image; based on a pre-constructed industrial open set defect detection dataset and the industrial defect image, generating an unknown class feature sample set through Monte Carlo sampling; based on the unknown class feature sample set, generating a reliable pseudo label through an adaptive pseudo label generation mechanism; based on the reliable pseudo label, performing joint prototype learning on known class samples and the unknown class feature sample set in the industrial open set defect detection dataset to construct known class feature prototypes and unknown class feature prototypes; and classifying the industrial defect image according to the known class feature prototypes and the unknown class feature prototypes; and the application performs joint prototype learning on the known class samples and the unknown class feature sample set in the industrial open set defect detection dataset based on the reliable pseudo label, can realize collaborative learning of the known class and the unknown class feature, and adapt to complex and diverse defect classification scenarios under the industrial open set.

This application relates to a method, apparatus, and device for classifying metal-based solid waste samples based on multimodal fusion. The method includes: inputting the elemental information of the metal-based solid waste sample to be tested into a pre-trained elemental classification model, outputting a first confidence matrix and a first reliability score of the first confidence matrix; inputting the image information of the metal-based solid waste sample to be tested into a pre-trained image classification model, outputting a second confidence matrix and a second reliability score of the second confidence matrix; when the first category corresponding to the highest confidence in the first confidence matrix is ​​inconsistent with the second category corresponding to the highest confidence in the second confidence matrix, and the category difference between the first and second categories is within a preset range, calculating the weighted total score of each category associated with the first and second matrices, and determining the category with the highest weighted total score as the category of the metal-based solid waste sample to be tested. This method has high classification accuracy and strong anti-interference ability.

The application provides a visual large model staged training data construction method and device, the method comprises the following steps: obtaining image data of a target application scenario and analyzing the image data to obtain a basic data pool comprising a structured scene attribute description; calculating a single probability of each attribute label, calculating a joint probability of attribute label pairs meeting a first screening condition, and judging as a long-tail sample if the attribute label pairs do not meet the first screening condition; classifying the image data according to the single probability and the joint probability to obtain a sample classification result; in a basic training stage, randomly sampling and screening image data meeting a preset quality condition to construct a basic training data set; in a target scene adaptation stage, weighted sampling according to the single probability and the joint probability to construct a target scene adaptation training data set; in a difficult sample and long-tail sample optimization stage, screening difficult samples and long-tail samples according to a prediction result of a current visual large model and the sample classification result to construct a difficult sample and long-tail sample optimization training data set. The application is self-adaptive to a scene and is efficiently trained in stages.

This invention relates to the field of surrounding rock stability evaluation technology, providing a method for evaluating the stability of coal seam roadways based on DBO-PPM and machine learning. The method involves adaptively gridding the roadway area to be evaluated and extracting multi-source stability influencing factors. DBO is introduced to globally optimize the projection direction of PPM, obtaining objective and accurate evaluation index weights. The K-means algorithm is combined to achieve automatic sample classification. The Smote algorithm is used to oversample scarce accident samples, constructing a class-balanced evaluation index system. Machine learning models, including a radial basis function classifier optimized with GridSearchCV hyperparameters, XGBoost, and Stacking models, are constructed for accurate prediction and visualization using ArcGIS. This invention improves the accuracy and efficiency of surrounding rock stability evaluation in complex geological conditions, providing a scientific basis for support optimization and disaster early warning in deep mines.

The intelligent specimen receiving system and receiving method of the present application, the system includes a sample classification and conveying assembly, the sample classification and conveying assembly includes a test tube feeding conveying line, a test tube transfer mechanism, a code scanning and rotating mechanism, a test tube and tray grabbing robot, a defective product placement position, an empty tray placement position, a test tube placement position and a manual material taking position, the test tube feeding conveying line is used for carrying specimen test tubes, the test tube transfer mechanism is used for transferring the specimen test tubes on the test tube feeding conveying line to the code scanning and rotating mechanism, the code scanning and rotating mechanism scans the specimen test tubes, the test tube and tray grabbing robot selects to transfer the specimen test tubes to the defective product placement position or the test tube placement position according to the code scanning result, after the test tube placement position meets the predetermined condition, the test tube and tray grabbing robot transfers the test tube placement tray on the test tube placement position to the manual material taking position, the cooperation between each mechanism automatically completes the classification and transportation of the sample, and errors caused by fatigue of manual work are avoided.

The application belongs to the field of deep learning and remote sensing image processing, and specifically discloses a hyperspectral few-shot classification network construction method and a hyperspectral feature classification method, which comprises the following steps: acquiring hyperspectral remote sensing images and sample label data thereof; performing cross-domain reconstruction and spectrum curve extraction, and filtering common information between categories through eigenvalue decomposition to obtain discrete spectrum curves; extracting physical invariance features and taking them as constraint rules to generate virtual samples; inputting the reconstructed hyperspectral data and spectrum self-supervised auxiliary information into a double-branch variational autoencoder network to perform multi-loss constraint of cross reconstruction, so that the hyperspectral data and the spectrum self-supervised auxiliary information of the same category show greater similarity in the latent space; and outputting final feature prediction results based on a multinomial logistic regression classifier to complete the construction of the hyperspectral few-shot classification network. The application can realize high-precision and high-robustness hyperspectral feature classification under the condition of sample scarcity.

This invention discloses a point cloud tree segmentation and forest parameter extraction method based on RandLA-Net, comprising the following steps: S1, data preprocessing; S2, RandLA-Net point cloud classification; S3, tree segmentation based on CHM+watershed algorithm; S4, forest parameter extraction; S5, region filtering and result output. This invention constructs a standardized point cloud preprocessing workflow, achieving accurate mapping, format conversion, downsampling, and index construction of multi-category labels, improving the quality of model training data, and adapting to the complex category distribution of transmission line point clouds; it optimizes the RandLA-Net model training strategy to solve the class imbalance problem and improve sample classification accuracy; it optimizes the tree segmentation algorithm to solve the segmentation problems of overlapping canopies and dense vegetation, improving the completeness of tree segmentation; it improves the accuracy of forest parameter calculation, optimizes the calculation methods for tree height and canopy area, and reduces tree height measurement errors; and it constructs a complete technical chain without manual intervention, adapting to the high-efficiency processing needs of large-scale transmission line point clouds.

The application is suitable for the field of automobile aerodynamic performance prediction technology, and provides a cross-dimension 2D-3D automobile aerodynamic prediction method based on an artificial intelligence algorithm. First, sensitive parameters of geometric features under the normal viewing angle of the automobile are selected for parameterized deformation to generate a parameterized deformation table and an automobile model set, and two-dimensional and three-dimensional CFD simulations are respectively carried out. Second, based on the mapping rule of the parameterized deformation table and the aerodynamic force coefficient, a K-Means and GMM hybrid clustering strategy is adopted for sample classification, the optimal clustering category number is determined in combination with MAE and MSE indexes, and a multi-branch neural network model set is constructed and trained for each type of sample. Finally, the deformation parameters of the new three-dimensional vehicle body model to be predicted are input into the model set, after clustering positioning and branch regression preliminary prediction, a scene-specific cross-dimension gain coefficient is introduced for correction and compensation, and the resistance coefficient prediction value is output. The method can greatly reduce the three-dimensional CFD simulation workload, ensure the prediction accuracy, and significantly save the computing resources.

The utility model provides a kind of sample storage box, it is related to sample storage technical field, comprising: box body, identification component and separating component, the top side of box body is rotatably connected with cover plate, the inside rotatable connection of cover plate has rotating rod.The sample storage box, by the setting of identification component, when using, after placing sample by opening cover plate, close cover plate, rotating rotating plate drives rotating rod to rotate, rotating rod rotation drives driving wheel to rotate, by mechanical engagement drive driven gear to rotate, and then drive fixed rod to rotate, fixed rod rotation can rotate prompt board and adjust the display picture of prompt board top, time recording module can manually record the time of placement, reach the effect that multiple sample classification storage and information correspond, improve sample management efficiency, avoid traditional sample storage device after placing multiple samples to easily mix different kinds of sample, leading to need to find when taking, and the situation that the convenience of taking is reduced.

This invention discloses a large-scene hyperspectral zero-shot classification method and system based on a visual language model. The method includes the following steps: generating an initial pseudo-label map for the hyperspectral image based on a visual language model; calculating the spectral angle between each pixel in the initial pseudo-label map and the spectral fingerprint vector; filtering the initial pseudo-label map based on the spectral angle to obtain an optimized pseudo-label map; training a hyperspectral segmentation network based on the hyperspectral image and its optimized pseudo-label map to establish a mapping relationship between the hyperspectral image and its land cover categories for land cover category classification. The hyperspectral segmentation network consists of an encoder, a decoder, and a classifier, wherein the encoder is composed of N-level cascaded multi-kernel convolutional residual modules. This invention aims to accurately identify and segment different land cover categories in complex hyperspectral images without manual annotation.

The application belongs to the technical field of computer vision, and discloses a few-shot classification method based on an image-text pre-training model.For the optimal transport of the base element level, an image block reweighting mechanism based on prototype distance is proposed.The Euclidean distance between the prototype and the image block is converted into the weight of each image block, so that different weights are assigned to the image blocks.For the prompt level stage, a cascaded optimal transport module is proposed, which not only considers the optimal transport between the few-shot image-text features, but also considers the cross optimal transport between the zero-shot and few-shot image-text features.A prototype-based unbalanced consistency loss function is proposed to supervise the network.The loss function is divided into a prototype distance loss and an unbalanced consistency loss.The test is carried out on Caltech01, DTD and EuroSAT data sets, and the experimental results prove that the network model proposed in the application is superior to the most advanced algorithm at present, and the effectiveness of the algorithm is verified.

The present application relates to a kind of based on adaptive noise's adversarial sample detection method and system, belong to computer network security technical field.First, the reconstructed residual is obtained as noise distribution by image compression and reconstruction network, then dynamic noise intensity is generated using convolutional neural network, and adaptive noise is formed by combination;By comparing the label consistency and confidence similarity of original sample and noise sample in the output of target model, score is calculated and compared with threshold to realize adversarial sample detection.The present application does not need internal information of target model, can be deployed by relying on a small amount of clean samples training, suitable for various image classification models and adversarial attack types, while ensuring the classification accuracy of clean sample, significantly improve the detection rate of adversarial sample, reduce false positive rate, can be used as third-party detection service adaptation commercial image classification API.

The application relates to the field of pattern recognition and machine learning, and discloses a dynamic feature selection method for a high-dimensional small sample classification task, which comprises the following steps: based on a gate network, a first importance score corresponding to the feature dimension of a high-dimensional data training sample is generated; based on the first importance score and an adaptive temperature adjustment module, a feature selection probability distribution is generated and input into a threshold screening module to obtain a first sparse mask; according to the first sparse mask and the high-dimensional data training sample, a first screening feature vector is determined and input into a preset multi-task back-end network to output a joint loss value; based on the joint loss value, the dynamic feature selection network is iteratively trained, a high-dimensional data to-be-tested sample is input into the trained dynamic feature selection network to obtain a second screening feature vector, and a classification result is determined according to the second screening feature vector. The scheme of the application realizes adaptive selection and accurate extraction of features, and improves the accuracy and stability of the high-dimensional small sample classification task.

The application discloses a positive and negative unlabeled image classification method and system based on deep metric learning, relates to the technical field of sample classification, and comprises the following steps: obtaining labeled positive sample images and unlabeled images, and inputting a deep metric learning model; adjusting the labeled positive sample images in a deep feature space; minimizing the distance between the labeled positive sample images and the center of the positive sample images; aligning the unlabeled images and their corresponding enhanced views in a representation space through self-consistent metric learning; and outputting relevant feature representations; selecting sample images in a multiple ratio of the number of positive sample images as reliable negative samples by using the relevant feature representations; taking the relevant feature representations as input, using the positive sample images and the reliable negative samples as supervision signals to train a binary classifier, and classifying positive and negative unlabeled images to be detected. The application avoids the dependence on heuristic negative sample selection and can learn high-quality feature representations that can distinguish positive and negative samples.

Disclosed is a loop filter method based on an adaptive pixel classification criterion. The loop filter method based on an adaptive pixel classification criterion in an image decoding device includes a step of classifying a restored sample according to an absolute classification criterion or a relative classification criterion, a step of obtaining offset information according to a result of classifying the restored sample, a step of adding the offset value to the restored sample with reference to the obtained offset information, and a step of outputting the restored sample to which the offset value is added. Thus, errors of the restored sample can be corrected.

This invention discloses an intelligent rice variety classification system and method based on image recognition, aiming to address the technical pain points of existing rice variety classification methods, such as reliance on manual experience, low efficiency, large errors, incomplete feature extraction, insufficient classification accuracy, and poor adaptability. The system includes a dual-optical-path controllable image acquisition module, a rice-specific preprocessing module, a dual-branch feature decoupling and fusion extraction module, a few-sample classification inference module, and a result output and verification module. The classification method sequentially performs image acquisition, preprocessing, feature extraction, variety identification, and result verification. Dual-path images of rice varieties are acquired through dual optical paths, and after preprocessing, the dual-branch module extracts fused features, which are then identified by the few-sample classification module. This invention achieves non-destructive, rapid, and accurate intelligent classification of rice varieties, improves the accuracy of distinguishing highly similar rice varieties, reduces usage and maintenance costs, adapts to multiple scenarios, and is applicable to fields such as agricultural seed testing and breeding research.

A method for detecting security vulnerabilities, comprising: generating a corpus of input samples each labeled to indicate a threat level when executed by an input processing code; training a neural network (NN) using the plurality of input samples to classify inputs according to a plurality of labels of the plurality of input samples; for each input sample: iteratively altering the input sample to correspond to a process of gradient change of the NN, until the NN classifies the altered input sample to a different label than a respective label of the input sample; assigning the different label to the altered input sample; using the plurality of relabeled altered input samples to further train the NN and augment the corpus of input samples.

1. The name of the design product: storage box (portable eye microbial sample classification type). 2. The use of the design product: for storing microbial samples. 3. The design points of the design product: in shape. 4. The picture or photo that best indicates the design points: perspective view 1.

This invention relates to the field of image processing technology, and in particular to a semi-supervised domain generalization system based on causal inference guidance. The system includes a feature extraction module for primary feature extraction of images; a cross-domain prototype self-learning module for data acquisition from the feature extraction module, which performs prototype comparison learning on the primary features and constructs a cross-domain discriminative loss to align instance-level and prototype-level cross-domain features; a causal guidance enhancement module for data acquisition from the feature extraction module, which enhances image samples through domain style feature mixing, key semantic feature suppression, and non-key region noise intervention; and a classification decision module for combining the cross-domain prototype self-learning module and the causal guidance enhancement module to obtain pseudo-labels. This invention can learn more robust causal feature representations and effectively utilize unlabeled data to improve the model's cross-domain performance.

The application discloses a multi-view generalized zero-shot classification auxiliary method and device based on PU learning, and belongs to the technical field of picture classification. The method comprises the following steps: acquiring pictures, extracting multi-view visual features of the pictures in a training set and a test set; constructing a benchmark model; constructing a multi-view PU classification model; inputting the multi-view visual features corresponding to the pictures in the training set and the test set into the multi-view PU classification model, and training the multi-view PU classification model by using an alternating direction multiplier method; resetting semantic prototypes of each unseen class based on a first semantic projection; resetting semantic prototypes of each seen class based on a second semantic projection; and classifying pictures to be identified by using the benchmark model. The application separates unseen class samples from the test samples to be identified by using a PU learning framework, and uses the unseen class samples for class semantic calibration, so that the bias of the model to the seen class samples can be effectively relieved, and thus the generalization performance of the generalized zero-shot classifier is significantly improved.

The embodiment of the application discloses a method and device for judging mitral valve prolapse and pathological segmentation based on a heart ultrasound image; the method comprises the following steps: acquiring a heart ultrasound image to establish a heart ultrasound image data set; training and testing a small sample classification model based on a picture after noise reduction and preprocessing; obtaining a heart mitral valve medical image data set and a heart mitral valve prolapse image data set; inputting, training and verifying a pathological segmentation neural model; preliminarily verifying the pathological segmentation neural model to obtain a classification-segmentation model; sequentially inputting a new medical ultrasound image into the classification-segmentation model to detect whether classification and pathological segmentation are completed; and realizing automatic judgment and segmentation of the heart ultrasound image for the heart mitral valve prolapse, short analysis and processing time, high efficiency, and high medical image judgment accuracy.

A process evaluates an AI assistant by collecting feedback tuples and a list of topics of the artificial intelligence assistant, wherein each feedback tuple include a question, a corresponding answer generated by the artificial intelligence assistant, and qualitative feedback corresponding to the question and the corresponding answer. A few-shot classifier outputs derived feature data. The process generates a feature-target training dataset by combining multiple derived features with metafeatures corresponding to each question-answer pair of the derived feature data and adding a quantitative feedback target corresponding to each question-answer pair of the derived feature data, and trains an attribution model using the feature-target training dataset to yield a trained attribution model. The process extracts feature importance vectors from the trained attribution model. Each feature importance vector indicates a relative importance of a given feature of the feature-target training dataset on a corresponding target of the feature-target training dataset.

The application relates to a radar target detection method based on a deep manifold network and an NP criterion. The method comprises the following steps: acquiring radar echo data, obtaining a time-frequency spectrum array through time-frequency transformation, constructing a time-frequency feature SPD matrix through symmetric positive definite processing and normalization processing. The SPD matrix is input into a preset SPDNet network, and features are extracted through a bilinear mapping layer, an eigenvalue correction layer, an eigenvalue logarithm layer and a full connection layer in sequence, and a sample classification score is output. Based on the differentiable loss function layer of the NP criterion, a detection probability and a false alarm rate are calculated according to the classification score, and a loss value is output. In combination with the loss value, the SPDNet network parameters are updated through gradient back propagation to complete training. The radar echo data to be detected is input into the trained network, and target and clutter are distinguished and detected according to the output classification score. The method has high radar target detection precision.

This invention relates to the field of machine learning technology and provides a few-shot classification method and system based on cross-memory attention. The method includes: acquiring support set and query set features; in multi-shot scenarios, generating fused features from the support set features via an adaptive fusion module; inputting the support and query features into a cross-memory attention module, where the memory encoder compresses the support features into a memory query vector using depthwise separable convolution; using this vector as the query and the support features as the key, generating memory-enhanced support features; concatenating this with the query features using a memory decoder to generate memory-enhanced query features; obtaining relation-enhanced features through residual connections; enhancing robustness through domain adaptation; calculating structural similarity using Earth Mover's Distance; and outputting the classification result. This invention replaces traditional pairwise attention with memory query vectors, reducing computational complexity, establishing a deep association between support and query, and improving the accuracy and robustness of few-shot classification.

The application belongs to the technical field of point cloud data processing, and particularly relates to a small sample point cloud semantic segmentation method. First, a meta-learning training strategy is used to train a constructed semantic segmentation network model; the semantic segmentation network model comprises a feature extraction network, a multi-prototype generation model and a relationship learning network; the feature extraction network is used for respectively performing semantic feature extraction on a support set and a query set to obtain support set features and query set features, the multi-prototype generation model is used for performing feature selection on the support set features to obtain prototype features, and the relationship learning network is used for learning the similarity relationship between the prototype features and the query features; then, to-be-classified point cloud data is input into the trained semantic segmentation network model to obtain a classification result. The application designs a small sample classification meta-learning method, which can realize classification of a new class by using only a small amount of labeled samples when facing a brand-new scene, and can improve overall classification precision compared with a supervised deep learning method.