42results about How to "Reduce the probability of misclassification" patented technology

The invention aims to solve the problems that the remote sensing image water area segmentation extraction method in the prior art is poor in self-adaptation due to the fact that a segmentation critical value is manually set, a large number of non-water-area land types exist in a result, and a large number of impulse noise exists in the result. The invention provides a remote sensing image water area segmentation and extraction method for super-pixel classification and identification. In combination with an improved linear clustering super-pixel segmentation method, a remote sensing image is divided into a plurality of super-pixels which are good in homogeneity, compact in layout and capable of well keeping edge information; superpixels are used as a feature extraction unit, water area features in a remote sensing image are extracted from three perspectives of spectrum, texture and terrain, the features of a water area and non-water areas are described more accurately, a typical learning sample library is constructed, and a nonlinear support vector machine is used for supervised classification. Experimental results show that the method can overcome the defects of the prior art and remarkably improve the water area segmentation and extraction precision and speed of the remote sensing image.

The invention provides a temperature sensor adjusted portable intelligent blood sugar instrument and belongs to the technical field of high-accuracy intelligent electric medical measurement instruments. Technology that temperature value is obtained through an intelligent temperature sensor chip and the temperature is compensated according to the real-time temperature is mainly introduced on the basis that a portable series blood sugar instrument researched and developed independently is convenient to carry, simple to use and accurate in measuring algorithm, and the like and on the premise of meeting national regulations. Influence of environmental temperature on blood sugar test paper measurement is greatly eliminated, and blood sugar value stability and accurate measurement algorithm are obtained.

The invention provides an electrocardiogram classifying method based on fuzzy inference and weighted similarity measurement and relates to an electrocardiogram classification method. The electrocardiogram classifying method aims to solve the problem that by the adoption of an existing fuzzy inference classification method, due to the fact that an electrocardiogram knowledge base cannot be established, the influence of electrocardiogram knowledge and different combinations of different wave band forms on classification is omitted, and as a result, the error rate of classification is high and to solve the problem that by the adoption of the existing fuzzy inference classification method, due to the facts that attribute concepts are not screened and comparison of membership degrees of the attribute concepts is directly used for classification, the error rate of classification is high. The electrocardiogram classifying method comprises the following steps that, firstly, electrocardiosignals are preprocessed; secondly, feature parameter extraction is conducted on all wave bands; thirdly, a classification feature attribute value vector Yi=[yi1, yi2, yi3, yi4 and yi5] and a to-be-detected feature attribute value vector X=[x1, x2, x4, x4 and x5] are established, and an ECG body ecg.owl is established according to electrocardiogram knowledge; fuzzy concept lattices are established; fuzzy attributes are converted into specific membership degree values, and effective screening is conducted on the specific membership degree values; final classification is conducted through a weighted classification method. The electrocardiogram classifying method is suitable for electrocardiosignal classification.

The invention belongs to the technical field of dam body deformation monitoring, and discloses a tailing pond dam body deformation monitoring system and method. The method comprises steps of detectingpressure data of the dam body of the tailing pond by using a pressure sensor; measuring the displacement data of the dam body of the tailing pond by using an inclinometer; enhancing the contrast of the acquired image by using image processing software; evaluating the deformation state of the tailing pond dam body according to the collected data; using the alarm for timely alarming and notifying according to the detected abnormal data; and displaying the interface of the tailing pond dam body deformation monitoring system and the acquired image, compression and displacement data through a display. According to the method, the contrast adjustment between the shadow part and the highlight part in the dam body image can be well processed to obtain a proper contrast, and image distortion caused by overlarge contrast adjustment is prevented; The image definition of the dam body is enhanced, and the deformation of the dam body is more conveniently monitored.

The invention discloses a channel environment self-adaptive spectrum sensing method based on a Catboost algorithm, comprising the following specific steps: 1, a secondary user collecting an energy value in a current channel environment and sending the energy value to a secondary user serving as a fusion center; 2, the main user discontinuously sending the channel resource occupation condition to afusion center; 3, the fusion center constructing the information sent by the primary user and the secondary user into a data set, and further constructing a feature vector set; 4, the fusion center using a Catboost algorithm to train the model; 5, the secondary user continuing to send the energy value to the fusion center, and the energy value serving as a test vector and being input into the trained model; 6, the fusion center sending the available channel resources to all the secondary users after obtaining the result, and the secondary users making a response according to the judgment of the fusion center. According to the method, under the condition that the false alarm rate is 0.1, the detection rate is increased by 10% compared with an SVM, and meanwhile, the false classification rate and the false classification risk are also remarkably reduced.

The invention relates to an unsupervised Markov random field image segmentation method, which is technically characterized in that: the scope of the classification number of an image M is firstly determined, the image M is then carried out the K-means segmentation from m min to m max type, the BICm which is corresponding to the image M under each type is calculated according to the BIC criteria, a potential energy function is applied to calculate the total energy of the image M, and finally the ICM method is selected to complete the Markov random field image segmentation. The method has the advantages that: the new potential energy function is applied to image segmentation, the anti-noise performance thereof is obviously improved, the segmentation result can remove most of the noise of the image with great noise, the error classification phenomenon can be simultaneously effectively reduced, the error classification rate is reduced by more than 60 percent; the classification number does not need to be manually determined, thus basically realizing the unsupervised image segmentation; and the judgment accuracy of the classification number can achieve more than 95 percent by the artificial synthetic image test.

The invention discloses a method for quickly identifying and classifying steel materials based on a support vector machine combined with laser-induced breakdown spectroscopy. First, a series of steel samples of known grades are detected through a LIBS system to obtain steel data matrices of different grades. Use the support vector machine to establish a classification model for the known category data. In the modeling process, an improved modeling method-combined model is used. When the sample data to be tested is input into the model, it is first fuzzy classified by the one-to-many method , to screen out the candidate categories, and then finely classify them through the one-to-one method to finally determine the category of the data to be tested. This method combines the traditional one-to-many and one-to-one modeling methods, and makes full use of the advantages of the two, so that the data to be tested can pass through the two-layer analysis system of fuzzy classification and fine classification, reducing the impact of useless category information on the prediction process. impact, thereby significantly improving the prediction accuracy and reducing the computational cost.

The invention relates to a mobile application traffic identification method and system based on machine learning, and belongs to the field of traffic identification. The method comprises a flow acquisition stage, a flow processing stage, a feature extraction stage, a flow marking stage, a flow balancing stage and a model training stage. The system comprises a flow monitoring module, a flow processing module, a flow display module, a feature extraction module, a feature display module, an application identification module and a result display module. According to the method, a multi-feature fusion feature extraction scheme is provided, the information richness is improved, the model training effect is optimized, and the classification accuracy is improved; a model training mode combining an SMOTE + ENN sample balance algorithm and a random forest algorithm is designed, so that the misclassification rate of minority class samples is reduced, and the classification and recognition capability of a classifier is improved.

The invention relates to a reeling steel state identifying method based on KNN reeling steel picture data of an optimized KAP sample, wherein KAP belongs to a fast efficient clustering algorithm, clustering is performed by the mutual relation among computation features, a new sample space is formed by clustering and selecting feature samples with representation in various samples, and the balance among each sample number in the new sample space is ensured by fixing the parameter K in the KAP algorithm. A PCA algorithm is utilized for performing dimension reducing on the features and the operation speed of the method is improved under the condition of ensuring the classifying precision. The classifying identification on the linear array CCD (charge couple device) picture-based reeling steel loading state under the complex natural environment is realized. The method has very high classifying precision and lower wrong classifying rate, and the actual needs are met with the faster classifying speed.

The invention relates to the field of garbage treatment, in particular to a garbage classification management method and system.The garbage classification management method comprises the steps that bag body type information is determined; based on the bag body type information, classified garbage bags are provided for the target user; and the garbage types of the classified garbage bags in the throwing points are matched with the garbage types of the garbage in the classified garbage bags. By means of the technical scheme, the classified garbage bags used for collecting garbage are subjected to garbage classification, so that only the garbage of the same garbage type is collected in each classified garbage bag, the garbage of different garbage types can be treated in a centralized mode, subsequent efficient garbage recycling treatment is facilitated, and the environment is protected. Therefore, the utilization rate and economic value of renewable resources are improved. According to the garbage classification method, the garbage types of the garbage in the classification garbage bags are detected, whether the garbage in the classification garbage bags is correctly classified or not can be analyzed, so that the behavior that the garbage is not correctly classified is warned or corrected in time, and the garbage classification method is more efficient and systematized.

The invention relates to an integrity detection method for a graphical model structure of a safety net, which comprises the steps of constructing a marking matrix of railway wagon tarpaulin rope net, and detecting the integrity of a graphical model geometric structure. The integrity detection method is based on non-rigid network modeling, and can realize automatic detection and recognition for the state of the railway wagon tarpaulin rope net based on linear array CCD pictures under the background of a complicated natural environment. Whether a topological structure of the safety net is integral or not is judged through acquired wagon pictures, and thus whether the structure of the safety net is integral or not is determined. The method provided by the invention has very high detection precision and low classification error rate, and can reduce the workload in checking the state of the safety net manually. Meanwhile, the high detection speed also meets actual engineering requirements.

An electrocardiogram classification method based on fuzzy reasoning combined with weighted similarity measure relates to an electrocardiogram classification method. In order to solve the problem that the existing fuzzy reasoning classification method cannot construct the electrocardiogram knowledge base, thus ignoring the influence of different combinations of electrocardiogram knowledge and different band forms on the classification, the problem of high classification error rate is solved, and the existing fuzzy reasoning classification method The method directly uses the comparison of membership degree values ​​to classify attribute concepts without screening, which leads to the problem of high classification error rate. The present invention first preprocesses the ECG signal, then extracts the characteristic parameters of each band, and constructs the classification characteristic attribute value vector Yi=[yi1 yi2 yi3 yi4 yi5] and the characteristic attribute value vector X=[x1 x2 x3 x4 x5 ], and create ECG ontology ecg.owl according to ECG knowledge; construct fuzzy concept lattice, convert fuzzy attributes into specific membership value, and effectively screen the membership value; then use weighted classification algorithm for final classification. The invention is applicable to the classification of electrocardiographic signals.

The invention discloses a stable evaluation method for electronic blood glucose measurement, which belongs to the field of medical measurement devices. The method comprises the steps of monitoring the variation of current in the reaction process in real time by utilizing an electrochemical analyzer, extracting n sample current values under the situation that the detection current is stable, getting an average value, obtaining a real-time amplified voltage value through a voltage amplifier, and obtaining a measurement value of the blood glucose through an analog-digital conversion and functional conversion method. By adopting the method, the error of the blood glucose electronic measurement caused by uncontrollable factors can be reduced, so that a stable and precise blood glucose value can be obtained.