195results about How to "Solve classification problems" patented technology

The invention discloses an enterprise industry classification method. According to the method, main business keywords of enterprises are effectively extracted by utilizing semi-supervised learning-based image split clustering algorithm, the extracted keywords are used as features on the basis of a gradient enhancement decision-making tree, and a training cascade classifier is used for classifyingthe enterprises according to industries, so that the problem that artificial classification is tedious is solved. The method specifically comprises the following steps of: 1) extracting main businesskeywords of enterprises by utilizing a word vector and a semi-supervised image split clustering algorithm, getting rid of junk words and constructing a keyword library; and 2) inputting the extractedkeywords which are taken as features into a training cascade classifier, the enterprises are classified by each level of classifier, and the unclassified enterprises are classified according to the next level of classifier. According to the method, keywords can be automatically constructed, updated and classified, the problem of classifying millions and millions of enterprise industries is solved,and the problem of artificial labelling is effectively solved.

The present invention discloses a malicious code analysis method based on semi-supervised learning. The method is characterized in that analysis is carried out based on the multi-dimensional features, and static features and dynamic features of a malicious code are extracted; and the difficulty of the subsequent processing through dimensionality reduction is reduced, and by using the semi-supervised learning method, the problem of the classification of a small number of marked malicious code samples is solved.

The invention discloses an extreme learning machine-based hyperspectral remote sensing image ground object classification method. An original extreme learning machine network is expanded into a hierarchical multi-channel fusion network. In terms of network training, the method is different from the least squares algorithm-based output weight solving strategy of the original ELM (extreme learning machine) and the global iterative optimization strategy of a deep learning network; according to the method of the invention, a greedy layer-by-layer training mode is adopted to train a hierarchical network layer by layer, and therefore, the training time of the network is greatly shortened; and in the layer-by-layer training process, a l1 regular optimization item is added into the training solving model of each layer of the network separately, so that parameter solving results are sparser, and the risk of over-fitting can be lowered. In terms of network functions, A single-hidden layer ELM network focus on solving the fitting and classification problems of simple data, while the different levels of the network model provided by the invention achieve target data feature learning or feature fusion, the network model of the invention integrates the advantages of high training speed and strong generalization capacity of the single-hidden layer ELM network, and therefore, the in-orbit realization of the model is facilitated, and the requirements of emergency response tasks can be satisfied.

Embodiments of the invention provide a method and device for multilingual hybrid model establishment and data acquisition, and electronic equipment. The method includes determining a modeling unit ofan acoustic model according to a speech unit contained by multilanguage, and establishing the acoustic model based on a deep neural network, wherein the modeling unit is the context free speech unit;obtaining multilingual hybrid speech training data, converting a hybrid speech signal in the multilingual hybrid speech training data into an eigenvector sequence, and converting a hybrid labelling text corresponding to the hybrid speech signal into a hybrid label sequence of the modeling unit based on the acoustic model; training the acoustic model by using the eigenvector sequence and the hybridlabel sequence; obtaining multilingual hybrid corpus data to train a language model; and establishing a multilingual hybrid speech recognition system according to the acoustic model and the languagemodel. According to the embodiments, recognition accuracy of speech data mixing multiple languages can be enhanced.

The invention discloses a terahertz dangerous article detection method based on depth learning. The terahertz dangerous article detection method comprises steps that a dangerous article sample image database is established, and images are processed to be gray level images, which have the same size suitable for training and testing; a CNN neural network model is trained, and a final network model is generated and tested; dangerous article detection is carried out, and terahertz equipment is used to acquire terahertz images of to-be-detected objects are acquired, and the CNN neural network modelis used to detect the terahertz images of the acquired to-be-detected objects, and then detection results are acquired; and at the same time, the terahertz images of the acquired to-be-detected objects are added to the dangerous article sample image database. The CNN neural network is directly used for the training and the learning of the sample images, and complexity of selection of network parameters is reduced, and the obvious characteristics of the sample image data can be directly learned, and therefore problems of image classification and mode identification can be solved; working efficiency of security staff can be improved, workload of workers can be reduced, and the abovementioned method and the abovementioned device are suitable for security check of large flow of people.

The invention discloses an image classification method and system based on migration learning. The method comprises the steps that 1, a feature similarity known training set A is utilized to make a training set B of a migration network through a support vector machine; 2, a migration learning network is constructed; 3, the training set B classified in the step 1 is used as a training learning setof the migration learning network, and a migration learning network model with high robustness and good accuracy is obtained through training; and 4, a to-be-classified dataset is introduced into thetrained migration learning network model, and a final classification result is obtained and marked with a tag. Through the image classification method and system, the requirement that a big sample dataset is needed to serve as input when a common RGB image is trained through deep learning is overcome, the problems of overfitting and locally optimal solutions in the training process are avoided, and classification precision is improved to some extent compared with a traditional classification algorithm.

The invention discloses a text classification method based on a capsule network. The problems that in the prior art, the overall precision is not high, the applicability is not high, a large amount ofimportant information is lost in the feature extraction process, and the relation between the local part and the overall part in a text is ignored are solved. The method comprises the following steps: 1, nodes in a capsule network being capsules consisting of a group of neurons, executing complex internal calculation on input by using matrix capsules, outputting instantiation parameters from results in a matrix form, and meanwhile, outputting an activation value of each capsule is output; 2, calculating between two adjacent layers in the capsule network through an EM routing algorithm; representing a higher-dimensional concept through Gaussian cluster, and each activation capsule selecting a capsule of the next layer as a father node through an iterative routing process, so that link prediction is realized between two adjacent layers of networks; and 3, training a weight parameter of the full connection layer, and calculating the prediction probability of the real label by using a softmax activation function.

The invention discloses a network user classification method and device. The network user classification method comprises the following steps of: generating behavior characteristic vectors of users based on the network behaviors of the users; clustering based on the behavior characteristic vectors of the users so as to construct a user-topic matrix; determining the webpage category to which each user belongs on a basis of the user-topic matrix, wherein the network behaviors of the users comprise network access behaviors and search behaviors, and each element value of the user-topic matrix respectively represents the correlation degree between each user and each topic. According to the method and device disclosed by the invention, on a basis of various types of data of the user search behaviors and access behaviors, overall user behaviors are covered within a smaller calculation cost by the adoption of a user-topic model, so that the user information is sufficient to use, and a more accurate user classification result is faster to obtain.

The invention relates to the technical field of garbage classifying, in particular to a smart community garbage classifying system and method. An account number is established for each community userand correlated to the garbage classifying system. The garbage classifying system comprises a server, a community garbage classifying management client side and garbage classifying terminal equipment.The server achieves user data storage and supports data reading and writing through various access ends. The community garbage classifying management client side mainly comprises a garbage classifyingtreatment management module, a garbage paid recycling management module, a garbage bag getting management module, a user management module, a user garbage classifying grade management module and other community equipment management modules. The garbage classifying terminal equipment comprises a garbage classifying treatment module and a garbage bag getting module. According to the provided garbage classifying system, the garbage classifying effect can be better guaranteed, and garbage recycling and treatment are benefited.

The invention discloses a novel garbage disposal manipulator system and an intelligent control method thereof. The overall structure mainly comprises a support, two manipulators, a conveying belt, eight classifying boxes and eight classifying plates. The two four-freedom-degree manipulators of the same structure are arranged; and three clamping claws arranged in a staggered manner is adopted for a tail end executor of each manipulator. The manipulator system is simple in structure, fast in response and convenient to control, and improves the clamping effectiveness. By adopting the above intelligent control system, the manipulators can accurately reach target bottles and cans, and achieves automatic classification. The sorting and classification problem of bottles and cans in travel rubbish is solved, and technical supporting is provided for healthy development of the travel industry.

The invention provides a garbage classifying management system and method. A box body structurally comprises a closed cover body; a movable garbage casting table is arranged below the cover body; a garbage casting opening is formed in the bottom of the garbage casting table; multiple garbage classifying storage boxes are arranged in the box body; the garbage classifying storage boxes comprise a sundries storage box; a picture acquisition device and an infrared scanner are arranged in the box body; the opening and closing of the cover body are controlled by scanning a QR code by an APP client end, or the cover body is opened by a switch button; besides, all kinds of garbage can be cast; the garbage is classified by utilizing the picture acquisition device and the infrared scanner; then thecasting table is moved to place the garbage into a corresponding storage box; and the classifying problem of the garbage is controlled and finished from the beginning.

The invention discloses an urban waste classification treatment monitoring system based on the Internet of Things. The system provided by the invention contains a central processing module; a satellite positioning module which forms a wireless connection with the central processing module; several wet waste treatment monitoring systems which respectively form a wireless connection with the central processing module and are used to monitor the wet waste treatment flow of a wet waste recycling bin, a regional wet waste centralized recycling station, a wet waste transport station, a wet waste transport vehicle and a wet waste disposal factory at real time; and several dry waste treatment monitoring systems which respectively form a wireless connection with the central processing module and are used to monitor the dry waste treatment flow of a dry waste recycling bin, a regional dry waste centralized recycling station, a dry waste transport station, a dry waste transport vehicle and a dry waste disposal factory at real time. According to the invention, dry and wet wastes are separated for processing so as to minimize waste moisture and burning fetor, increase burning calorific value, promote energy conversion and raise waste utilization rate.

The invention discloses a full-automatic stack room which comprises bookshelves, roadway conveying mechanisms and an automatic terminal. Each bookshelf is provided with layer partition plates, each roadway conveying mechanism is provided with a horizontal rail, a vertical rail is disposed on each horizontal rail, a manipulator rail which is perpendicular to each horizontal rail and each vertical rail is disposed on each vertical rail, and a manipulator is disposed on each manipulator rail. Significant improvements include that book fetching and returning manipulators are in butt joint with a book conveying platform, and the book conveying platform is connected with the automatic terminal. The full-automatic stack room can have multiple functions of storage, borrowing, returning, classification, temporary storage and the like of books, and has wider adaptability and more flexible expansibility as compared with a full-automatic closed shelf stack room with basic functions of storage, borrowing and returning.

The invention discloses a method for a computer to identify a vehicle type by a video image. The method is characterized by comprising the following steps of: (1), acquiring a video image by a computer and extracting a target key frame from the video image, exacting a target region to be identified through wiping out background, and pre-processing the video image; (2), carrying out feature extraction on the pre-processed image, selecting an Hu geometric moment invariant in the image shape characteristic as a vehicle characteristic parameter, and calculating the shape characteristic of the target region; and (3), carrying out sample training of all vehicle types through a machine learning method, and carrying out classified predication on the target region to be identified after the sample training is finished so as to identify the vehicle type. The method can semi-automatically classify the vehicle types, so that the classifying accuracy is improved as much as possible.

The invention discloses an industrial control system (ICS) communication network anomaly classification method based on statistical learning and deep learning. According to the invention, the method comprises the steps: designing LSTM deep learning structure parameters on the basis of the flow of a large-data-volume industrial control system communication network during normal operation, and performing modeling analysis; designing a correlation algorithm to analyze a numerical relationship between background traffic and real-time traffic by analyzing a real-time communication traffic data threshold generated based on a SARIMA online statistical learning model in the early stage; and carrying out specific classification on the ICS communication network anomaly according to an ICS network anomaly event classification algorithm. According to the invention, experimental analysis is carried out by using a target range test board combining industrial control safety virtuality and reality inZhejiang Province; meanwhile, a physical simulation platform is built in a laboratory environment to carry out a verification experiment, and detailed examples are given to verify the reliability andaccuracy of the algorithm.

The invention discloses a hyperspectral image semi-supervised classification method based on space-spectral information. The hyperspectral image semi-supervised classification method combines spectral information and spatial information in a hyperspectral image to act on a support vector machine classifier, adopts a self-training semi-supervised classification framework, utilizes an active learning method as a sample selecting strategy of semi-supervised classification, decomposes initial classification results obtained through semi-supervised classification according to classes so as to obtain various classes of binary images as input images of an edge preserving filter, regards a first principal component content as a reference image of the filter, utilizes the edge preserving filter to perform local smoothing, eliminates noise, and classifies image elements according to a class with maximum probability, thus the classification process is completed. The hyperspectral image semi-supervised classification method combines the spectral information and the spatial information to improve the classifiability of classes, utilizes the self-training semi-supervised classification framework to solve the classification problem of hyperspectral image small samples, can effectively eliminate spot-like errors in the initial classification results, and increases classification precision.

The invention relates to a method for estimating vegetation coverage degree in diggings based on a close-up photography mode. The method comprises the following steps of: photographing the diggings to be tested, extracting color features and texture features to form a feature vector according to a particular order, fast classifying vegetation and non-vegetation utilizing a support vector machine classifier, then gathering statistics of vegetation information, and then fast and accurately calculating the vegetation coverage degree for providing reliable basis for surveying the growth status of the vegetation in the diggings. The method also can be used as actual verification of quantitative remote sensing estimation of vegetation coverage algorithm at the same time. The invention also relates to a system for estimating the vegetation coverage degree in the diggings based on the close-up photography mode. The system comprises a feature extraction module, a classifier training module, a vegetation and non-vegetation classifying module and a vegetation coverage degree calculating module.

The invention relates to the technical field of image processing, and provides an Image classification method, apparatus, terminal device, and storage medium. After the image to be classified is acquired, the image to be classified is input into a convolution neural network model constructed in advance to obtain the image characteristics of the image to be classified; Then, the target text corresponding to the image to be classified is selected from the preset text library, the target text is converted into a word vector, and a pre-constructed circulating neural network model is input to obtain the text characteristics of the target text. Then, the weighted image features are obtained by combining the text features with the weighted superposition processing of the image features. Finally,the weighted image features and the text features are fused by bilinear multiplication, and two classes of classification are completed by using the fused features to obtain the image classes of the image to be classified. The invention can solve the problem of high-fine-grained image classification.

The invention discloses a network encrypted traffic recognition method and device. The method comprises a preprocessing stage and a classification stage. In the preprocessing stage, original flow is subjected to flow segmentation, sampling, vectorization and standardization, a sampling scheme in large flow is provided, and the classification problem of the large flow is solved. In the classification stage, spatial feature capture and abstract feature extraction are performed by using a CNN, and then traffic time sequence features are learned by using stacked bidirectional LSTM on the basis ofabstract features so that automatic feature extraction and efficient recognition of encrypted traffic can be realized. The method has universality, can automatically extract the space-time features ofthe encrypted traffic without manual feature design of experts, and can adapt to traffic feature changes caused by different encryption technologies and confusion technologies.

The invention relates to an object-neural-network-oriented high-resolution remote-sensing image classifying method, aiming at solving the problems that the conventional remote-sensing image classifying method is low in classification precision and cannot effectively utilize information of all wave bands of a remote sensor. The method comprises the following steps that: an image of the ground is shot by a high-spatial-resolution sensor and is transmitted to a computer; the computer carries out primary image element division on the input image by a region growing algorithm; the primarily-divided image is subjected to multi-size division according to continuously-set neterogeny degree thresholds and shape features and spectral signatures of the image, thus forming divided images with different sizes; and the obtained divided images with different sizes are used for establishing a BP (Back Propagation) neural network, setting training parameters and establishing training samples to classify the image which is subjected to the multi-size division, thus obtaining a high-resolution image. The method is applicable to the field of obtaining of images with high spatial resolutions.

The invention relates to a garbage classifying device and an application method thereof, and belongs to the technical field of sanitation equipment. Garbage needing to be classified is poured into a garbage collecting box and enters a conveyor belt through the garbage collecting box, the conveyor belt conveys the garbage to a sorting device, a laser sensor above the conveyor belt recognizes the types of the garbage below and feeds the types back to a control chamber, and the control chamber controls baffles in the sorting device; when the recognized garbage is recoverable garbage, the baffle Arotates clockwise, the baffle B is static, and the garbage falls into a recoverable garbage tank along the baffles; and in a similar way, when the recognized garbage is kitchen waste, the baffle A isstatic, the baffle B rotates anticlockwise, the garbage falls into a kitchen waste tank along the baffles, and garbage classification is completed by the baffles which rotate to enable the garbage tofall into different recovery tanks.

The invention discloses a three-decision unbalanced data oversampling method based on a Spark big data platform, and relates to a Spark big data technology in the field of data excavation. The method comprises the following steps: firstly, carrying out data transformation with an RDD (Resilient Distributed Dataset) of Spark to obtain a normalized sample set with the LabeledPoint format <label: [features]>, and dividing the sample set into a training set and a test set; secondly, carrying out data variation by adopting the RDD of Spark, calculating a distance between samples, determining the radius of a domain, and classifying the samples in the whole training set into positive domain samples, boundary domain samples and negative domain samples according to a neighborhood three-decision model; then respectively oversampling the boundary domain samples and the negative domain samples; and finally, calling a Spark Mllib machine learning algorithm to verify a sampling result. According to the three-decision unbalanced data oversampling method based on the Spark big data platform, the problem of classification of a large-scale unbalanced data set in the field of machine learning and mode recognition is effectively solved.

The invention discloses a user behavior fine classification method and system of a mobile application private encryption protocol. The method comprises the following steps: 1) collecting the flow of amobile application, and then identifying the private encryption protocol flow from the collected flow according to set private encryption protocol characteristics; 2) collecting flow data of a set user behavior category from the identified private encryption protocol flow, and labeling the flow data; 3) generating a training set, a verification set and a test set according to the flow data collected and marked in the step 2); 4) performing feature extraction on the flow data in the three sets, and converting the flow data into feature vectors; 5) setting hyper-parameters of the selected classifier, and training the selected classifier; 6) verifying the training classifier by utilizing the feature vectors corresponding to the verification set, and 7) classifying the classifier on the testset, and if the set standard is met, classifying the mobile application traffic to be processed by utilizing the classifier.

The invention provides a reservation type book management system. The reservation type book management system comprises a full-automatic reservation book stack and a reservation type book management module. The full-automatic reservation book stack comprises a book stack body, a book taking and returning mechanical arm and a book conveying platform. The reservation type book management module comprises a reservation data base, a book returning module, a book taking module and a communication module. The reservation type book management system intelligently manages and controls the book stack composed of a bookshelf, the book taking and returning mechanical arm and the book conveying platform through the reservation type book management module, can achieve warehousing and temporary storage of reserved books, and informs, reserves and processes book returning or book borrowing actions of readers and readers making in-place reservation. In addition, returned book classifying and transporting work is completely solved through mechanical automation, the work efficiency of a library is greatly improved, a mode that classifying and warehousing of returned books and processing work of reserved books only can be finished manually at present is completely replaced.

The invention belongs to the technical field of crossing of machine learning and emotion recognition, and discloses an electroencephalogram emotion recognition method and system, computer equipment and wearable equipment, and aims to reduce the influence of non-emotion signals on emotion recognition by removing electroencephalogram signals generated at the beginning of video conversion and subtracting the average value of the signals from remaining data. The method comprises the steps of: extracting time-frequency domain features of the pre-processed electroencephalogram signals by using short-time Fourier transform; putting the features into a convolutional neural network for training, and extracting high-quality features; and performing hypergraph learning on the obtained features, constructing a hypergraph classifier model, and completing emotion classification and recognition. According to the invention, the time-frequency features of electroencephalogram signals are optimized by adopting a deep learning method, and then training and classification are carried out by using a hypergraph learning method for sampling, so that the training time is effectively shortened on the basisof improving the classification accuracy of hypergraph learning, the operation space is compressed, and the method is of great significance to design, research and development of portable wearable equipment.

The invention relates to a method for identifying a green mark of the taxi, relating to the technical field of traffic control computer image process and mode identification, and comprising a camera control module, a movement detecting vehicle snapshot module, a number plate locating module, a green mark locating module and a green mark identifying module, wherein the camera control module is based on an intelligent vehicle information acquisition system. The location rule of the green mark locating module is based on color features and area features of a detecting region, and the color features are judged in an HSV (Hue, Saturation, Value) model space. The identification of the green mark identifying module comprises two stages: carrying out the morphology conversion according to an extracted selecting region of the green mark to obtain a more complete and regular processing region and calculating two circularity form factors of the region according to the fine process result of the first stage, integrating the result of the two circularity form factors to obtain the final identification according to the judgment rule, and outputting the identification result.

The invention belongs to the field of pattern recognition, particularly relates to an object detection method and system based on dynamic sample selection and loss consistency, and aims to solve the problems of insufficient object recognition accuracy and performance. The method comprises the following steps: firstly, acquiring a test image, dynamically selecting a positive sample and a negative sample in a training process, introducing a non-maximum suppression loss, and acquiring a prediction frame position of the test image and a probability that a prediction frame belongs to each categoryby an object detection model; and acquiring the target category and the prediction box position of the optimal test image through non-maximum suppression. Each annotation box generates the same numberof positive samples, the optimizer can fairly treat each training sample, and the regression loss function is re-weighted by predicting a IOU of each prediction box through dynamic sample selection,so that the optimal detection result is more accurate, and the detection accuracy is improved. In the training stage, a non-maximum suppression loss function is introduced to punish false detection generated in training, so that the false detection is reduced in the test stage.

The invention discloses a coal quality sorting method based on cokeability of coking coal. The sorting method comprises the following steps: 1) determining the indexes which affect the cokeability of coking coal: determining the average maximum reflectivity of vitrinite, the Gieseler maximum fluidity, solid-soft temperature interval and optical organizational structure of coke as the indexes of cokeability of coking coal; 2) determining the indexes which affect the cokeability of coking coal; and 3) correspondingly dividing various single coal into determining the indexes which affect the cokeability of coking coal: gas-fat coal, gas coal, fat coal, 1 / 3 coking coal, coking coal, lean coal, inferior mixed coal or special cause coal according to different measuring results in the step 2). The invention further discloses a coal distributing method based on the cokeability of coking coal. By taking the average maximum reflectivity of vitrinite, the Gieseler maximum fluidity, solid-soft temperature interval and optical organizational structure of coke as the sorting indexes, not only are the indexes simple to set, but also the inferior mixed coal or special cause coal can be differentiated, and part of highly degenerative coke is prevented from being misjudged as lean coal, so that resources are scientifically and reasonably configured.

The invention provides an intelligent household system, and a condition configuration and control method for realizing many-to-many communication. The intelligent household system comprises a router, a real time control and configuration end device, a gateway apparatus, and at least one controlled intelligent device, wherein network communication connection is established between the router and the real time control and configuration end device; and data communication connection is established between the gateway apparatus and the controlled intelligent device; and the controlled intelligent device comprises a control object device, a condition control device, a bi-direction controlled device, wherein the control object device is used for completing household practical functions; the condition control device is used for applying condition control for the control object device; and the bi-direction controlled device can be used as a condition control device and also can be used as a control object device. The intelligent household system, and the condition configuration and control method for realizing many-to-many communication can solve the problem about mutual communication among multi-types and variable number of controlled devices; and according to the classification, the condition control device unidirectionally controls motion execution of the control object device directly so as to establish a unified and well-organized communication mechanism without occurrence of disordered situation on logic and data of mutual communication among a plurality of devices.

The invention relates to a hierarchical support vector machine classifying method based on a rejection subspace. The hierarchical support vector machine classifying method based on the rejection subspace is applicable to processing multi-class or unbalance big data classification problems. The hierarchical support vector machine classifying method is capable of realizing hierarchical parallelization processing on big data in virtue of the rejection subspace so as to improve the classification result. The hierarchical support vector machine classifying method comprises the following steps: firstly, acquiring support vector machines low in computation complexity through training; secondly, determining the rejection subspaces of the support vector machine by virtue of a mutual information learning criterion to obtain rejection training sets in original training sets; and thirdly, training high-accuracy support vector machines on the rejection training sets for further judging the rejection training sets; and the training process is repeated for a plurality of times according to actual requirements. The hierarchical support vector machine classifying method has the advantages that the training complexity of the support vector machine of each layer is reduced according to the idea of dividing and ruling, and the optimal rejection subspace is determined by the data through the mutual information; therefore, the hierarchical support vector machine classifying method has the characteristics of low computation complexity, listening to the data and the like; besides, the method can be applied to the fields of big data classification such as medical diagnosis and multi-class object detection.