39 results about "Stage classification" patented technology

A method and apparatus for two-stage packet classification, the two-stage packet classification scheme including a first stage and a second stage. In the first classification stage, a packet is classified on the basis of the packet's network path. In the second stage of classification, the packet is classified on the basis of one or more transport (or other) fields of the packet. Also disclosed are embodiments of most specific filter matching and transport level sharing, and either one or both of these techniques may be implemented in the two-stage classification method.

The application relates to the technical field of bone maturity analysis, and in particular to a bone age recognition method and system and an electronic device. The bone age recognition method includes the a step a of inputting a bone image into a target bone position detection model, and detecting position coordinates of a target bone in the bone image through the target bone position detectionmodel; a step b of cutting out a target bone image according to the position coordinates of the target bone; and a step c of inputting the target bone image into a bone stage classification model forbone age recognition. The bone age recognition method and system and the electronic device in the application adopt deep learning to automatically classify and recognize ulna and radius images withoutmanual intervention, improve the effectiveness and efficiency of bone age recognition and have better performance than the prior art; and also eliminate the need for manual acquisition, save more time and efforts, and are highly efficient.

The invention relates to natural language processing and model recognizing technical field, which discloses a two-stage combine text classifying method based on probability subject word, wherein the first stage classification is based on Bayes classifying method to use probability subject word and reject condition judge couple to test the text classification, and the second stage classification is based on traditional character extraction method to extract character word to classify the test text rejected by the first stage. The inventive method classifies the text and fuses the characters of different classifiers to quickly classify various texts in the first classification, to improve classifying efficiency, to provide better treatment on the text classification system. And the invention provides probability subject word based on text character, while the probability subject word can effectively classify various texts in right reject condition. Compared with traditional single classification, the invention can reduce time consumption and improve system classification correct rate.

The invention provides an automatic sleep stage classification method based on dual character filtering. The stage classification method comprises steps of firstly extracting a two-guidance sleep electroencephalogram signal and an one-guidance horizontal electrooculogram signal; performing filtering for the original electroencephalogram signals and electrooculogram signal; extracting multiple characters from the filtered electroencephalogram signals and the filtered electrooculogram signal; selecting the optimal character subsets by use of a dual character filtering method combining the Fisher score method and the sequential forward selection method. Via the dual characteristic filtering method, character dimension is greatly reduced and redundancy among the characters is reduced. At last, a support vector machine classifier is used for identifying the optimal characters, so automatic stage classification of sleep is finished. According to the invention, objectivity, precision and convenience of automatic sleep stage classification can be well increased; the automatic sleep stage classification method is characterized by high precision, low calculation complexity, simple operation and easy popularization; and considerable social and economic benefit can be gained.

The invention discloses a tree network-based multistage server networking communication method which can be applied to a video conference system. Forwarding servers are deployed according to conference participant distribution and network bandwidth condition; stage classification and code management are carried out on each server; a sending end sends data to a server (S10), the server (S10) distributes the data to a receiving end, and at the same time forwards the data to an upper-stage node server (S0) and forwards the data to a plurality of lower-stage node servers (S200, S201); and the lowest-stage node servers directly distribute the data to the receiving ends connected with the lowest-stage node servers. According to the tree network-based multistage server networking communication method, the problem of poor intercommunication among internet operators can be solved, and the method can be better adaptive to a physical network structure of an enterprise private network.

Disclosed are an apparatus and a method for determining a multi-view specific object. The apparatus comprises an input device for inputting image data; and cascade classifiers formed of stage classifiers corresponding to a same detection angle, the stage classifiers corresponding to different features. Each cascade classifier is for calculating a degree of confidence of the image data of a specific object corresponding to the detection angle based on the aspect of the corresponding feature, and determining whether the image data belongs to the specific object based on the degree of confidence. A self-adaptive posture prediction device is disposed between two stage classifiers in each cascade classifier, and is used to determine whether the image data enters the cascade classifiers corresponding to the detection angles and located after the self-adaptive posture prediction device.

The present invention is mass spectral model for detecting liver cancer serum characteristic protein and its preparation process, and belongs to the field of protein fingerprint detecting technology. The present invention screens 17 characteristic proteins including 6 up regulation proteins and 11 down regulation proteins, selects two or more proteins out of the 17 characteristic proteins and establishes two or more stage classification tree with the clinical peak values of the proteins to constitute the detection model for detecting the liver cancer serum characteristic protein. The present invention lays foundation for further discovering new liver cancer biological mark, and the model has over 90 % specificity, sensitivity and positive predicting rate in liver cancer diagnosis.

The invention relates to a hematite stage grinding-magnetic separation-reselection-acid positive flotation process which comprises the steps of raw ore, first-stage grinding and first-stage classification. The process is characterized in that the first-stage classification overflow is fed to medium magnetism; the concentrate of the medium magnetism is subjected to rough and fine classification; the tailings of the medium magnetism is fed to strong magnetism after carrying out concentration and deslagging operation before strong magnetism; the strong magnetic ore is subjected to rough and fineclassification; a rough particle product obtained by rough and fine classification is reselected; the reselected tailings is subjected to two-stage classification; the middle ore of a trochus is returned to the trochus; the two-stage classification overflow is returned to the medium magnetism; the two-stage classification sank sand is subjected to two-stage grinding; the discharge ore of the two-stage grinding is returned to the two-stage classification; the separated fine particle product is subjected to acid positive flotation operation after concentrated; the flotation concentrate and the reselected concentrate are mixed into final concentrate; and the flotation failings and the strong magnetic tailings are mixed into final tailings for discarding. The invention has the advantages of simple process flow, convenience in control, stable technical index, good effects of energy saving and emission reduction and more ideal effect on processing low-grade ore.

The invention relates to a semi-supervised electroencephalogram signal-based sleep staging method under multi-domain features, and belongs to the field of human-computer interfaces. The method comprises the steps of processing original electroencephalogram signals in combination with an ant colony algorithm and a semi-supervised Bayesian classification method to obtain an original feature set under the multi-domain features; optimizing the feature set through the ant colony algorithm and extracting an optimal feature subset; and performing stage classification in combination with an active learning policy by taking the optimal feature subset as a classification feature through using an improved semi-supervised Bayesian classification method. According to the method, the sleep staging can be effectively realized, the calculation resource waste caused by low-efficiency feature attributes and a blind search process are avoided, and a demand quantity of labeled samples can be reduced; and compared with multiple sleep staging methods proposed in recent years, a result shows that the method provided by the invention not only can achieve the staging precision effect of a mainstream algorithm but also can greatly reduce the demand quantity of the labeled samples, thereby avoiding manual misjudgment situations.

The embodiment of the invention provides a stage classification method based on a resilient distributed dataset (RDD) and a terminal, which relate to the field of big data and can shorten the execution time of Spark operation. The method comprises the steps that firstly, a directed acyclic graph (DAG) of the RDD of an application program is created by the terminal; subsequently, stage classification is perfromed on the RDD in the DAG by the terminal, and a sub-RDD, a parent RDD with wide dependency on the sub-RDD and a parent RDD with narrow dependency on the sub-RDD are classified into different stages when the wide dependency and the narrow dependency simultaneously exist in any sub-RDD; then corresponding tasks of the parent RDD with the wide dependency on the sub-RDD and the parent RDD with the narrow dependency on the sub-RDD are executed in parallel by the terminal, and then a corresponding task of the sub-RDD is executed. The stage classification method and the terminal are applied to memory computing so as to classify and execute the stages based on the RDD.

The embodiment of the invention provides a sales stage identification method and device based on client problems. The method comprises: responding to a currently received first client problem, obtaining each second client problem before a first client problem in a current session which the first client problem belongs to, and a third client problem in last session before the current session, wherein first, second and third client problems are all bound to a current client account number, the third client problem is last client problem in the last session, and the time interval between the current session and the last session is greater than a preset time threshold; extracting a first characteristic, a second characteristic and a third characteristic respectively associated with the first client, the second client and the third client, and forming a first characteristic vector associated with the first client problem according to the first, second and third characteristic; and identifying the sales stage corresponding to the first client problem according to the first characteristic vector by utilizing a sales stage classification model.

The embodiment of the invention discloses a method and a device for establishing a threat activity topological graph and storage equipment. The method and the device are used for solving the problem that the process of threat activity of an attacker cannot be accurately restored due to the fact that analysis on threat activity of a network space stays in fuzzy evaluation based on network assets and an influence range in the prior art. The method comprises the steps of performing stage classification on a target threat activity based on a network space threat framework; in combination with a threat activity relationship, inferring related threat activities of all stages before and/or after the stage where the target threat activity is located; establishing a sequential association relationship between the target threat activity and the related threat activity; obtaining a topological graph of a plurality of complete threatening activities according to the sequential incidence relation;and assessing the influence range of each complete threat activity according to each topological graph, comparing the influence range with the actual network asset influence range, and determining thecomplete threat activity topological graph of the target threat activity.

The invention discloses a classification method and device for multi-level classification objects, and relates to the technical field of computers. One specific embodiment of the method comprises thefollowing steps: constructing a cascade classification model step by step in a joint training mode by utilizing sample feature data of multi-stage classification objects, and a construction process: constructing a first-stage cascade classification model according to a first-stage sub-classifier; when K is larger than or equal to 2 and smaller than or equal to N, the sample feature data of the multi-level classification objects and K-1-level classification information, output by the K-1-level cascade classification model, of the multi-level classification objects are jointly input into a Kth-level sub-classifier, so that the K-1-level cascade classification model and the Kth-level sub-classifier are jointly trained, and a K-level cascade classification model is obtained; and classifying the to-be-classified data of the multi-stage classification object by utilizing the final N-stage cascade classification model to obtain N-stage classification information. According to the embodiment,the inter-class hydrophilicity and hydrophobicity are considered, the classification effect is good, error transfer can be avoided, the classification accuracy is improved, the overall model complexity is low, and the model development difficulty and workload are reduced.

The invention discloses an encrypted traffic classification method based on two-stage judgment, which comprises the following steps: taking the classification output of a random forest classifier as the input of a K-nearest neighbor classifier, and finally fusing and comparing the results of the two classifiers to finish the classification of traffic data. The output characteristics of the randomforest algorithm and the reliability guarantee of two-stage judgment are fully utilized, and the classification accuracy can be improved. Accoridng to an encrypted traffic classification algorithm based on two-stage judgment, the method comprises the following steps of 1, training a meta-feature sequence to obtain a first-stage classification model, that is, a random forest model containing k trees, then, taking a label output by each tree as a new composite feature vector and input into a K-nearest neighbor classifier to be classified in the second stage, comparing results obtained after twotimes of classification wherein classification succeeds only when the results obtained after two times of classification are the same. Such a two-stage judgment model enables the final classificationaccuracy to be significantly improved.

The invention provides an information acquisition method and system based on social media emergencies. The method comprises the following steps: S1, constructing a corpus of emergencies; S2, using a support vector machine classifier to carry out non-emergency event classification filtering, and achieving first-stage classification; And S3, performing positive and negative class prediction classification by using a naive Bayes classifier to realize second-stage classification. According to the invention, corpus acquisition of related keywords is carried out on the social media through the crawler; A support vector machine classifier is used for carrying out non-emergency event classification filtering, first-stage classification is achieved, a naive Bayes classifier is used for carrying outpositive and negative class prediction classification, second-stage classification is achieved, the information classification precision is improved by 2.9% compared with the result of non-instant seismic information screening, and the accuracy of information classification is improved by 2.9%. Compared with the prior art, the method has the advantages that the value of the F-Masure is increasedby 2.6%, the problem that the text classification result is low in precision in the prior art is solved, the classification precision is improved, a decision maker can control disaster events easily,and a basis is provided for decision making.

The invention discloses a medicine frying sorting machine. The machine comprises a medicine frying device and a sorting device. The sorting device is arranged on a discharging opening of the medicine frying device, fried medicine in the medicine frying device falls in the sorting device from the discharging opening, the sorting device comprises a vibration frame, a motor, a gearbox, four springs, a cam, a rotating shaft, an upper-layer first-stage screen net, a middle-layer secondary screen net, a lower-layer third-level screen net, a first-stage medicine guide-out sliding way, a second-stage medicine guide-out sliding way, a third-level medicine guide-out sliding way and a fourth-stage medicine box, the four corners of the vibration frame are supported on the four springs, the four springs are supported on the ground, the cam is supported at the bottom of the vibration frame, and the upper-layer first-stage screen net, the middle-layer secondary screen net and the lower-layer third-level screen net are distributed on the vibration frame and are mounted in an inclined manner. The motor drives the cam through a transmission mechanism, the cam presses the vibration frame in a top manner, and the vibration frame does the vertical reciprocated vibration. Due to the adoption of the above structural manner, fried medicine can be subject to four-stage classification.

The invention relates to a multi-stage fingerprint database search method. The method comprises the following steps of: performing first-stage classification by using singular points; performing second-stage classification by using information between the singular points; and performing third-stage classification based on invariant moment characteristic. In the invariant moment characteristic-based multi-stage fingerprint database search method, the invariant moment characteristic is introduced on the existing basis to establish a three-stage search mechanism; and at the third stage, the defect of being influenced by rotation and displacement of an input image is overcome by extracting orthogonal invariant moment characteristic of a fingerprint image ROI (Region Of Interest) with the same reference point, so that the reliability and the stability of the system are improved. A research result has a higher academic application value in the fields of fingerprint image analysis, recognition, image retrieval, information security and the like.

The invention relates to an embryo developmental stage classification method in an embryo time sequence image. The method comprises the steps that M time sequence images to be detected in the embryonic development process are acquired and sequentially input into a single-input-multiple-output convolutional neural network, m probability sequences corresponding to m adjacent images to be detected ina one-to-one mode are obtained under input of each image to be detected, and m is smaller than M; based on all the m probability sequences, integrating is performed to obtain m probability sequencesof each to-be-detected image, and the m probability sequences are fused to obtain a probability fusion sequence of the to-be-detected image; and a matrix formed by the probability fusion sequences ofthe to-be-detected images is smoothed by adopting a dynamic programming method meeting monotonically increasing constraints, and a development stage corresponding to each to-be-detected image is identified. According to the method, the single-input multi-output convolutional neural network is adopted, integrated fusion processing is combined, single-input multi-output is converted into single-input single-output, and finally the development stage of each image is obtained by adopting a dynamic planning method, so that the classification accuracy is high, and the calculation complexity is low.