33results about How to "Low classification error rate" patented technology

A method and apparatus for performing discriminative training of, for example, call routing training data (or, alternatively, other classification training data) which improves the subsequent classification of a user's natural language based requests. An initial scoring matrix is generated based on the training data and then the scoring matrix is adjusted so as to improve the discrimination between competing classes (e.g., destinations). In accordance with one illustrative embodiment of the present invention a Generalized Probabilistic Descent (GPD) algorithm may be advantageously employed to provide the improved discrimination. More specifically, the present invention provides a method and apparatus comprising steps or means for generating an initial scoring matrix comprising a numerical value for each of a set of n classes in association with each of a set of m features, the initial scoring matrix based on a set of training data and, for each element of said set of training data, based on a subset of said features which are comprised in the natural language text of said element of said set of training data and on one of said classes which has been identified therefor; and based on the initial scoring matrix and the set of training data, generating a discriminatively trained scoring matrix for use by said classification system by adjusting one or more of said numerical values such that a greater degree of discrimination exists between competing ones of said classes when said classification requests are performed, thereby resulting in a reduced classification error rate.

The invention relates to a deep convolutional neural network method based on information lossless pooling. The method is used for image classification, and comprises the steps of collecting different kinds of images and marking image categories as image label information; performing image set division, and dividing the collected images into a training set, a verification set and a test set; designing a convolutional neural network structure based on information lossless pooling, including the number of convolutional layers and the number of information lossless pooling layers that are used, designing the number of filters in the convolutional layers, designing Gaussian smoothing filter parameters of the information lossless pooling layers, the pooling window size and a convolutional filter structure for feature fusion, designing the number of network training loop iterations and network final convergence conditions, and initializing network parameters; and inputting training data to the network in batches for calculation and training.

The invention belongs to the technical field of intelligent automobile running environment perception, and discloses an automobile running working condition judgment system and a judgment method of the automobile running working condition judgment system. The method comprises the steps that the parameters of the speed and pedal opening degree of an automobile are selected to be treated, character parameters in speed signals and pedal signals representing the working condition are extracted, the character parameters are firstly reduced for the first time through a correlational analysis method, the first fifteen parameters with largest correlation and the working condition are selected for correlation analysis, ten character parameters are selected from large to small according to the correlation as the standard for dividing the working condition, then the ten character parameters are reduced for the second time through kernel principal component analysis, character parameter values are reduced to seven, and finally the running working condition of the automobile is classified based on the semi-supervised nucleus fuzzy C-mean value clustering analysis of an objective function. According to the system and method, the judgment speed is high and accurately, the fuel consumption and tail gas emissions in the automobile running process can be reduced through the judgment result, and the system and method have great significance in the study of various new automobile type development, dynamic property matching of the whole automobile and the like.

A method for classifying enterprise domains and screening enterprise keywords is disclosed by the invention. The method obtains the feature words of the related fields through training the enterprisedomain corpus as a classification dictionary, and then uses the classification dictionary to classify the enterprise profile documents. After classification, the method can also extract the industry label representing the enterprise domain from the enterprise profile documents. In addition, the method can overcome the influence of most noisy words in Chinese text processing, and has low error rate, high classification efficiency and excellent comprehensive performance.

The invention discloses an artificial immunization non-supervision image classification method based on manifold distance and relates to the technical field of image processing. The specific process includes: (1) inputting an image to be classified and setting an initialization parameter to generate initialized antibody population; (2) based on the manifold distance, classifying the category of the sample point of the image to be classified and calculating the affinity of the antibody population; (3) carrying out clonal proliferation operation on the antibody population; (4) carrying out clonal variation operation on the antibody population after clonal proliferation; (5) classifying the category of the image to be classified according to a code of the antibody population after clonal variation and calculating the affinity of the antibody population; (6) carrying out clonal selection operation on the antibody population according to the antibody affinity; and (7) according to set maximum iterations, judging the stop condition of the category classification result of the image to be classified and determining the final classification result. The classification method has the advantages of low sensitivity of image data structure, non-supervision execution, good classification effect and strong robustness, and can be applied to the target identification in the field of image processing.

The invention discloses a deep neural network structure design method inspired by an optimization algorithm. For all layers share the same linear and non-linear transformation classic feedforward network structure, the forward process in the feedforward network is equivalent to an iterative process of minimizing a function F (x) using a gradient descent method. Furthermore, the function F (x) is minimized by using the double sphere method and the Nesterov acceleration algorithm with a faster convergence rate, and a new network structure with better performance is obtained. The method can be used in artificial intelligence, computer vision and other application fields. Through adoption of the technical scheme of the invention, the neural network structure is designed from the optimization algorithm, a traditional design mode depending on experience and experimenting to search can be improved, the more efficient neural network structure can be obtained, and a large amount of time and compute resources can be saved.

The invention discloses a 3D shape image classification method of an isovariant 3D convolutional network model based on a partial differential operator, and the method comprises the steps of: carrying out the parametric modeling of a convolution kernel by employing the partial differential operator, solving a 3D rotating group and the feature domain of each convolution layer, obtaining an isovariant convolution kernel, and building an isovariant 3D convolutional network model PDO-e3DCNN, wherein the input of the PDO-e3DCNN is a 3D shape, and the output of the PDO-e3DCNN is prediction classification of the 3D shape, and the PDO-e3DCNN is used for 3D shape classification and identification visual analysis. According to the method, picture data with direction features can be effectively processed, and a lower 3D shape image classification error rate can be achieved on a data set by using fewer parameters.

The invention provides a fast structural SVM text classification optimization algorithm. For a text classification task of an unbalanced data set, the algorithm directly optimizes the performance evaluation index of the main category by using the performance evaluation method such as on an accuracy rate, a recall rate, an AUC, and the like. The method is different from most conventional text classification algorithms that: instead of learning a single rule to predict a tag of a single sample, the method formalizes the learning problem into a multiple predictive problem on all the samples in the data set, and is different from idea in the conventional method that the reduction of overall classification error rate is taken as the target, so that classification accuracy of the text data set under the unbalanced condition is improved, and the classification performance is effectively improved; and referring to a Structural SVM based sparse approximation algorithm, with better time complexity, the method can be used for evaluation indexes such as the F value calculated from the accuracy rate and the recall rate, and the optimization of the AUC, so that the time complexity is reduced and better results are obtained.

The invention provides an intelligent evaluation method for satellite remote sensing image availability, comprising the following steps: S1, inputting images; S2, according to the user requirements, inputting the surface feature type and calculating the surface feature influence parameters of the surface feature type; S3, performing cloud detection on the image inputted in the step S1; 4, calculating the cloud thickness, the cloud fragmentation degree and the cloud coverage rate according to the cloud detection result; S5, inputting the cloud thickness, the cloud fragmentation degree and the cloud coverage calculated in the step S4 into a remote sensing image availability evaluation model to calculate the objective availability of the corresponding remote sensing image; S6, calculating thefinal availability level of the remote sensing image according to the objective availability calculated in the step S5 and the surface object influence parameter calculated in the step S2; S7: outputting the corresponding remote sensing image according to the final availability level. The invention provides an intelligent evaluation method of remote sensing image availability, which combines theobjective factors influencing the availability and the subjective influence of the related user requirements, and is used for realizing the image availability evaluation through a reasonable step design.

The invention relates to a Dropout regularization method based on the sensitivity of activation values. The method is used for image classification. A data training stage of the method comprises the following steps: 1) data preparation: collecting different types of images and marking the image types as labels; 2) structure design: setting a deep convolutional neural network structure; 3) initialization: (1) determining the weight of a convolution filter, initializing the parameters by using a random initialization method and setting the number of times of iteration and (2) setting a probability density function selected in Dropout; 4) forward computing: performing computing layer by layer from the first layer to the last layer, determining the probability of zero setting of each feature point via the probability density function after the max pooling layer, generating a random number between 0 and 1 by using a uniform distribution function, comparing the random number with the probability of zero setting of each feature point, zero-setting the activation value of the feature point if the random number is less than the probability and maintaining the activation value of the featurepoint if the random number is equal to or greater than the probability; 5) back propagation.

The invention relates to the technical field of computer deep learning, and provides a hybrid theme model construction method for deep learning. The method comprises the following steps of S1, preprocessing; S2, representing the text information; S3, supplementing a background information sub-network; and S4, dividing the theme of a full connection layer network, and outputting a label classification probability. According to the invention, the theme of the data of a Huawei cloud platform and an intelligent learning platform is mined, a hybrid theme model HTM based on deep learning is discovered, the required data volume in the field of theme classification is smaller, and the texts of different lengths can be converted effectively via a Bi-LSTM framework to obtain the better migration capability, so that the migration capability of the model is high, the classification error rate is low, and the overall classification effect of the model is good, and the beneficial attempts are made for the theme classification model of deep learning in small sample learning and transfer learning in future.

The invention discloses a silent living body detection method. The silent living body detection method sequentially comprises the following steps: 1, preprocessing operation is performed on an input video to obtain a face picture with a fixed size; 2, the face image with the fixed size obtained in the step 1 is represented by RGB and YCrCb color spaces, and then the face image is input into a double-flow feature fusion network model to achieve feature extraction; and step 3, a non-living body category is subdivided into a printing attack category and a display attack category, category prediction is performed according to the image feature representation obtained in the step 2, and a multi-classification cross entropy loss supervision model is adopted for training, so that the classification error rate of the algorithm can be effectively reduced, and the generalization performance can be improved. According to the invention, single-frame picture input is adopted, additional auxiliary equipment is not needed, the algorithm implementation cost is low, additional man-machine interaction is not needed, and good user experience is achieved.

The invention discloses a method for detecting the edge of a regularized straight line section of an image of a texture-free metal part. According to the method, the gradient of each pixel point of the input part image is calculated firstly, then a preliminary straight line segment is detected by using an LSD algorithm, and discontinuous straight line segments are connected by using the distance and angle relationship between the straight line segments in the next step, so that the purpose of detecting complete straight line segments is achieved, and the problem that the detected straight linesegments are broken is avoided. The method is improved on the basis of an LSD algorithm, can output a complete straight line segment on the basis of keeping the high speed of the LSD algorithm, is suitable for an RGB image and a gray level image, and can meet the requirements of practical application.

The invention provides a data fusion method and device, electronic equipment and a computer readable storage medium, and the method comprises the steps: classifying to-be-fused data from different sources, and obtaining a plurality of subclass data; searching similar data of each sub-category data in a pre-established standard database, and determining the similarity between the similar data and the sub-category data; and fusing the sub-category data with a standard database according to the similarity. According to the method, the classification error rate can be reduced, invalid data in thefusion database is reduced, and the quality of database data is improved.

The invention discloses a hardware Trojan horse detection method based on a lifting algorithm. Processing according to the topological structure of the integrated circuit to obtain a first controllability value, a second controllability value and an observability value of each node of the integrated circuit; performing density-based clustering according to the first controllability value, the second controllability value and the observability value of each node of the integrated circuit to obtain a matrix of band class distances; and training classification is performed by adopting a weak classifier, the weak classifier is combined to obtain a final classifier, and then the final classifier is adopted to process the matrix X'with the class distance obtained by processing the to-be-detectedintegrated circuit so as to obtain the hardware Trojan horse detection result of the to-be-detected integrated circuit. The method provided by the invention has higher accuracy than a bypass analysismeans under a large sample condition.

The invention discloses a construction engineering material information classification method and system based on part-of-speech and master-slave relations. The system comprises a character processingmodule, a material classification module, a classification screening module, a classification arbitration module, a building material classification noun library, a building material classification auxiliary word library, a building material classification feature word library, a master-slave classification relation library and a material classification library. By adopting the method and the system provided by the invention, construction engineering material information described by human natural language can be automatically collected and classified according to GB / T508512013 Construction Engineering Artificial Material Equipment Mechanical Data Standards, and the automation of the functions of intelligent identification, standard classification, statistical analysis and the like of theconstruction engineering material information can be realized; the material information processing capacity in the field of construction engineering is improved, and third-party application and system development of the construction engineering and the building material industry are helped to improve the efficiency.

The invention discloses a garbage classification processing method, related equipment and a readable storage medium. The garbage classification processing method determines information of a commodityby identifying an image of the commodity to be subjected to garbage classification processing, and then analyzes the information, so as to determine an object generated after the commodity is used anda garbage type corresponding to the object. Based on the determined object generated after the commodity is used and the garbage type corresponding to the object, a user can determine the object generated after the commodity is used and the garbage type corresponding to the object, and garbage classification is carried out based on the object generated after the commodity is used and the garbageclassification error probability can be reduced.

The invention discloses a method for key text detection and classification training of certificate pictures. The method includes the following steps: S1, constructing a certificate template, and generating training samples based on the certificate template; S2, constructing an algorithm model integrating text detection and classification, and Output the text area and its classification results based on the detection feature map and classification feature map output by the network; S3, import the training samples into the integrated algorithm model of text detection and classification for training, and respectively analyze the classification feature map and detection prediction feature map during the training process Carry out two types of negative sampling; S4, sequentially calculate the classification feature map loss value in the classification feature map and the detection prediction feature map loss value in the detection prediction feature map, and fuse the classification feature map loss value and the detection prediction feature map loss value to reverse Optimize the training text detection and classification integrated algorithm model to the propagation. Beneficial effect: the training of deep neural network can be supported by collecting a small number of samples.

The invention relates to the technical field of deep residual networks, and discloses a regional water resource classification evaluation method based on an improved deep residual network, which comprises four steps of image enhancement, image fusion, image marking and image segmentation. According to the regional water resource classification evaluation method based on the improved deep residual network, image enhancement is used for improving image information and quality, so that the features of the image are more obvious under human eye observation, the regional water resource identification effect is enhanced, visual analysis ignores the digital interference degree, and the method plays an important role in various aspects of remote sensing images; according to image fusion, a multi-spectral image with low spatial resolution and a single-band image with high spatial resolution are replayed and sampled to generate a new multi-spectral image with high resolution, so that the generated image has high spatial resolution and multi-spectral characteristics, a composite mode is mainly adopted, information provided by remote sensing image data sources of different sensors is integrated, and multi-spectral image fusion is realized. And high-quality image information is obtained.

The invention provides a classification method and apparatus, a device, and a storage medium. The method comprises the steps of obtaining device information of a target device; determining a target feature vector corresponding to the target device according to the device information of the target device; and inputting the target feature vector into a trained classification model to obtain the category of an application place where the target device is located. The device application place can be automatically determined, and the cost and the classification error rate can be reduced.

The invention belongs to the technical field of intelligent automobile running environment perception, and discloses an automobile running working condition judgment system and a judgment method of the automobile running working condition judgment system. The method comprises the steps that the parameters of the speed and pedal opening degree of an automobile are selected to be treated, character parameters in speed signals and pedal signals representing the working condition are extracted, the character parameters are firstly reduced for the first time through a correlational analysis method, the first fifteen parameters with largest correlation and the working condition are selected for correlation analysis, ten character parameters are selected from large to small according to the correlation as the standard for dividing the working condition, then the ten character parameters are reduced for the second time through kernel principal component analysis, character parameter values are reduced to seven, and finally the running working condition of the automobile is classified based on the semi-supervised nucleus fuzzy C-mean value clustering analysis of an objective function. According to the system and method, the judgment speed is high and accurately, the fuel consumption and tail gas emissions in the automobile running process can be reduced through the judgment result, and the system and method have great significance in the study of various new automobile type development, dynamic property matching of the whole automobile and the like.

The invention discloses a multi-country banknote serial number character recognition method based on image fingerprints, which belongs to the technical field of banknote pattern recognition. Based on the template set and sample set of multi-country banknote serial number character pictures, the perceptual hash algorithm is used to generate training sample data, and the classification model based on SVM classifier is constructed, and the training sample data is used to train the classification model, and the trained classification model is used Identify the serial number character picture of the banknote to be processed. The method of the invention converts a single feature in the banknote serial number character recognition process into multiple features, and adopts a method based on an SVM classifier for recognition, which has a small demand for samples, fast training speed, fast recognition speed, and good performance Anti-interference ability can improve the error tolerance rate of prediction.

The invention discloses a hardware Trojan detection method based on a lifting algorithm. The first and second controllability values ​​and the observability values ​​of each node of the integrated circuit are obtained by processing according to the topology of the integrated circuit; the first and second controllability values ​​and the observability values ​​of each node of the integrated circuit are Based on density clustering, a matrix with class distances is obtained; a weak classifier is used for training classification, and the weak classifiers are combined to obtain a final classifier, and then a matrix X' with class distances obtained by processing the integrated circuit to be tested is used by the final classifier Processing is performed to obtain the hardware Trojan detection result of the integrated circuit to be tested. The present invention has higher accuracy than the bypass analysis method under the condition of large sample.

The invention discloses a method for automatically acquiring genetic knowledge for product concept design: 1. Using the first-order predicate logic method to analyze the function and behavior information of the target product, and using the decomposition and reconstruction principle to decompose the functional information into functional units; 2. From the product instance Find product instances of the same family as the target product in the library to form a sample set; 3. Cluster the samples into K categories, the value of K is set according to the type of target product, and calculate the Euclidean distance between the target product and the center point of each category, The class of the center point with the smallest distance is the class to which the target product belongs; 4. Extract the attribute information related to the functional unit from the samples of the class to which the target product belongs; 5. Use the K-fold cross-validation method to select Sixth, the selected m attributes and behavior information are stored in the product gene bank in the form of coding chains to obtain the genetic knowledge of the target product. The present invention can automatically obtain the information that affects the conceptual design process. key knowledge.

A regularization method for detecting line segment edges in an image of a non-textured metal component. The method comprises: calculating gradient values of respective pixels in an input image of a component; employing an LSD algorithm to detect preliminary line segments; and using distances and angle relationships between the line segments to connect discontinuous line segments. Thus, the invention detects complete line segments and avoids the problem of broken line segments when performing detection. The method is an improvement on LSD algorithms, and can be used to output complete line segments while also maintaining the fast speed of the LSD algorithms. Moreover, the method is applicable to both RGB images and grayscale images and so can meet the requirements of actual applications.

The invention discloses an image fingerprint-based recognition method for multi-national banknotes and belongs to the technical field of banknote pattern recognition. Based on the template set and sample set of the multi-national banknote background image, the perceptual hash algorithm is used to generate training sample data, and the classification model based on the SVM classifier is constructed, and the training sample data is used to train the classification model, and the trained classification model is used for processing Banknote images are used to predict and identify currency types. The method of the invention converts a single feature in the banknote identification process into multiple features, and adopts a method based on an SVM classifier for identification, which requires a small amount of samples, fast training speed, fast recognition speed, and has good anti-interference ability , which can improve the error tolerance rate of prediction.

The invention discloses a deep neural network structure design method inspired by an optimization algorithm. For a classical feedforward network structure in which all layers share the same linear and nonlinear transformations, the forward process in the feedforward network is equivalent to using The gradient descent method minimizes the iterative process of a function F(x); further adopts the re-ball method with faster convergence speed and the Nesterov acceleration algorithm to minimize the function F(x), thus obtaining a new network structure with better performance ; Can be used in artificial intelligence, computer vision and other application fields. Adopting the technical scheme of the present invention and designing the neural network structure based on the optimization algorithm can improve the traditional design method of relying on experience and experimental search, and obtain a more efficient neural network structure, thereby saving a lot of time and computing resources.

The invention relates to a tool wear state evaluation method based on optimal characteristics and lion group optimization SVM. The method comprises the following steps: S1, data preprocessing; s2, feature extraction; s3, performing feature filtering to obtain an optimal feature subset with the highest correlation and the minimum redundancy; s4, constructing a tool wear evaluation model; and S5, evaluating the wear state of the cutter. The feature subset obtained through the method is higher in relevancy and minimum in redundancy, the SVM classifier optimized through the lion group algorithm is low in time complexity, higher in optimization capacity, faster in iteration and lower in classification error rate, the recognition rate reaches 97.125%, the tool wear state evaluation speed and stability can be improved, the production efficiency is improved, and the production cost is reduced.

The invention discloses an image classification method of an equivariant convolution network model based on a partial differential operator. For the input layer and the middle layer of the convolution network model, the equivariant convolution and The equivariant convolution of the middle layer constructs the equivariant convolution network model PDO-eConv and performs model training; the input of the model PDO-eConv is image data, and the output is the prediction and classification of images, so as to realize efficient image classification, recognition and visual analysis . The present invention can provide a better parameter sharing mechanism and achieve a lower error rate of image classification.