814 results about "Binary classification" patented technology

An enterprise-wide query relaxative support vector machine ranking algorithm approach provides enhanced functionality for query execution in a heterogeneous enterprise environment. Improved query results are obtained by adjusting ranking functions using machine learning methods to automatically train ranking functions. The improved query results are obtained using a list of document-query pairs that are modeled as a binary classification training problem, combination function which requires ranking and learning functions to be implemented representing document attributes and metadata utilizing query relaxation techniques and adjusted ranking functions. Machine learning methods implement user feedback to automatically train ranking functions.

The invention discloses a pedestrian detection method based on a deep learning technology. The method comprises the steps that firstly, a binary classification model is trained through a step-by-step migration strategy on the basis of transfer learning to initialize final model parameters; secondly, pedestrian detection work is completed by adopting and modifying a currently popular and efficient Faster RCNN frame, and on the basis of the CNN characteristics of the frame, not only can images with any scales be processed, but also the detection speed is high. Compared with the prior art, the method has the advantages that the network does not need to be specially designed, existing available data is fully utilized, a good experiment result still can be achieved by adopting a general network structure, the advantages of a deep convolution network are fully achieved, and the advantages of being simple in design, good in robustness, high in detection accuracy and low in omission ratio are achieved.

The invention belongs to the technical field of computers, and discloses a machine translation method and system based on a generative adversarial neural network. The method comprises the following steps that: on the basis of an original machine translation generation network, a discrimination network which generates network countermeasure with the original machine translation generation network is imported; a translation used for judging a target language is from a training parallel corpus and is a network machine translation result of the original machine translation generation network; and the discrimination network adopts a multi-layer sensor feedforward neural network model to realize binary classification. The system comprises the discrimination network, a generation network, a mono-lingual corpus and a parallel corpus. While manually annotated bilingual parallel corpus resources are fully utilized, and mono-lingual corpus resources also can be fully utilized to carry out semi-supervised learning; and the mono-lingual corpus resources are very rich and can be easily obtained, and the problem that required training corpora required by the neural network machine translation model are not sufficient is solved.

The subject disclosure presents systems and computer-implemented methods for assessing a risk of cancer recurrence in a patient based on a holistic integration of large amounts of prognostic information for said patient into a single comparative prognostic dataset. A risk classification system may be trained using the large amounts of information from a cohort of training slides from several patients, along with survival data for said patients. For example, a machine-learning-based binary classifier in the risk classification system may be trained using a set of granular image features computed from a plurality of slides corresponding to several cancer patients whose survival information is known and input into the system. The trained classifier may be used to classify image features from one or more test patients into a low-risk or high-risk group.

The invention relates to a mobile robot real-time layered path planning method based on a grid map. The method, which is used in a grid map, comprises the following steps: carrying out binary classification according to external environment information, and establishing a grid map; planning an outer path according to a speed-based artificial potential field method; and when the path is about to fall into local minimum and oscillation, planning an inner path according to an A star algorithm. The problem that the traditional artificial potential field method may fall into local minimum and planned path oscillation and the problem that the A star algorithm is not applicable to real-time path planning are solved. Real-time, safe and stable operation of mobile robots is ensured.

The invention discloses an image type fire flame identification method. The method comprises the following steps of 1, image capturing; 2, image processing. The image processing comprises the steps of 201, image preprocessing; 202, fire identifying. The fire identifying comprises the steps that indentifying is conducted by the adoption of a prebuilt binary classification model, the binary classification model is a support vector machine model for classifying the flame situation and the non-flame situation, wherein the building process of the binary classification model comprises the steps of I, image information capturing;II, feature extracting; III, training sample acquiring; IV, binary classification model building; IV-1, kernel function selecting; IV-2, classification function determining, optimizing parameter C and parameter D by the adoption of the conjugate gradient method, converting the optimized parameter C and parameter D into gamma and sigma 2; V, binary classification model training. By means of the image type fire flame identification method, steps are simple, operation is simple and convenient, reliability is high, using effect is good, and the problems that reliability is lower, false or missing alarm rate is higher, using effect is poor and the like in an existing video fire detecting system under a complex environment are solved effectively.

The invention discloses a deep residual network-based semantic mammary gland molybdenum target image lump segmentation method. The method comprises the following steps of: labelling pixel categories of lumps and normal tissues corresponding to a collected mammary gland molybdenum target image so as to generate label images, and dividing the mammary gland molybdenum target image and the corresponding label images into training samples and test samples; preprocessing the training samples to form a training data set; constructing a deep residual network, and training the network by utilizing thetraining data set, so as to obtain a deep residual network training model; after a to-be-segmented mammary gland molybdenum target image lump is preprocessed, carrying out binary classification and post-processing on a pixel of the to-be-segmented mammary gland molybdenum target image by utilizing the deep residual network training model, and outputting lump segmentation image to realize semanticsegmentation of the mammary gland molybdenum target image lump. The method is capable of effectively improving the automatic and intelligent levels of mammary gland molybdenum target image lump segmentation, and can be applied to the technical field of assisting radiologists to carry out medical diagnosis.

The invention provides a mammary gland lesion area detection method based on deep learning and transfer learning. The method comprises: preparation and amplification of a training set and a test set;according to lump position information marked by a doctor in the breast data set, extracting an available lump image and normalizing the size of the available lump image into a size of 100 * 100 pixels as a positive sample; according to the invention, the AlexNet network is used to train the parameter model of the classification model of the natural image on the ImageNet data set; training and transfer learning are carried out on a specific breast image data set, so that the binary classification problem of the convolutional neural network on a small-scale breast data set can be successfully solved, a lesion area in the breast image can be identified, and the prediction effect on the breast lesion is improved.

The invention provides a convolutional neural network-based remote sensing image building change detection method, and the method comprises the following steps: 1, building a sample library accordingto an existing image and a surface coverage vector file, wherein the sample library comprises a building classification data set and a building change detection simulation data set; 2, training a building extraction network (Mask R-CNN or a multi-scale full convolutional network) by using the building classification data set in the sample library constructed in the step 1, learning building characteristics in the high-resolution remote sensing image, and then extracting a multi-temporal building binary classification graph of the change area by using the trained network; 3, training a change detection network Mask CD-net by using the analog data set in the sample library constructed in the step 1; directly using the Mask CD-net to predict the building change detection data set to obtain achange detection result. If the real change data set is contained, the Mask CD-net is further trained and refined, a better change detection effect is obtained.

An enterprise-wide query relaxative support vector machine ranking algorithm approach provides enhanced functionality for query execution in a heterogeneous enterprise environment. Improved query results are obtained by adjusting ranking functions using machine learning methods to automatically train ranking functions. The improved query results are obtained using a list of document-query pairs that are modeled as a binary classification training problem, combination function which requires ranking and learning functions to be implemented representing document attributes and metadata utilizing query relaxation techniques and adjusted ranking functions. Machine learning methods implement user feedback to automatically train ranking functions.

Embodiments of the present invention provide a system, method, and program product for predicting the classification of a work-item notification generated by an issue tracking system. A computing device analyzes a new work-item notification using a trained binary classifier that was trained using at least a past and / or current work-item history that was collected as a tuple that includes information reflecting the workitem-ID, actor, timestamp, attribute, work-item start-value, and work-item end-value. The computing device, in response to analyzing the new work-item notification, predicts a classification of the new work-item notification. In an embodiment, the classification is used for display of the new work-item notification.

The invention discloses a pedestrian identification method of a camera network based on multi-level depth feature fusion. A new network model is learnt on a pedestrian database by migrating parameters of a pre-training network to the pedestrian database, a plurality of multi-level depth features are extracted using the new network model, and then a Softmax classifier in the last layer of a convolutional neural network is replaced by an SVM classifier to achieve the purpose of making full use of the multi-layer depth features. Furthermore, the multi-level depth features are used for constructing a plurality of groups of SVM classifiers of binary classification and decision values of these binary classifiers are linearly weighted to obtain final classification results. The invention can effectively improve the accuracy of recognizing a pedestrian target by way of multi-level feature fusion in a decision-making layer of the SVM classifier.

The invention provides a method for analyzing and recognizing the structure of a handwritten mathematical formula in a natural scene image. The method comprises the steps of S1, converting the gray matrix of a natural scene image into a local contrast matrix, and conducting the binary classification on the local contrast matrix based on the otsu method to obtain a binary matrix; S2, analyzing the connected domains of the binary matrix obtained in the step S1, and removing non-character type connected domains to obtain character type connected domains; S3, detecting formula structural elements and other special structural elements in the character type connected domains based on the correlation coefficient method, and separately marking out all detected special structural elements; S4, dividing the binary matrix obtained in the step S1 based on the horizontal projection method; S5, recognizing each character type connected domain via a convolutional neural network; S6, defining an output sequence and outputting recognized results according to the corresponding sequence in the latex layout format. According to the technical scheme of the invention, by means of the method, the expression problem of elementary mathematical formulas during the OCR recognition process can be effectively solved.

A method and system for polyp segmentation in computed tomography colonogrphy (CTC) volumes is disclosed. The polyp segmentation method utilizes a three-staged probabilistic binary classification approach for automatically segmenting polyp voxels from surrounding tissue in CTC volumes. Based on an input initial polyp position, a polyp tip is detected in a CTC volume using a trained 3D point detector. A local polar coordinate system is then fit to the colon surface in the CTC volume with the origin at the detected polyp tip. Polyp interior voxels and polyp exterior voxels are detected along each axis of the local polar coordinate system using a trained 3D box. A boundary voxel is detected on each axis of the local polar coordinate system based on the detected polyp interior voxels and polyp exterior voxels by boosted 1D curve parsing using a trained classifier. This results in a segmented polyp boundary.

The invention relates to a defect identification method for a solar panel based on a convolution neural network (CNN). The method comprises the two stages of model off-line training and on-line detection. CNN models are applied to defect identification of the solar panel, and defect detection and classification are progressively realized by two CNN models. Firstly, a CNN binary classification model is used for distinguishing qualified and defective images, and then a CNN multi-classification model is used for classifying images which are classified as defects by the binary classification model. The CNN models adopt the same processing flow for various defect types of the solar panel, namely, feature extraction and feature classification are performed rapidly and automatically through iterative training. For a new defect type, detection of the defect type can be realized by only collecting sample data of the defect type, adding the sample data into a training data set and training the models. Through adoption of the defect identification method, the location of a small defective solar panel can be identified at relatively high accuracy. Moreover, the method can classify various defects, so that the applicability of the method is wider.

The invention provides an electric power communication network fault positioning method based on key alarm sets and supervised classification. The method is suitable for fault positioning of network devices or network links in a communication network, particularly for a positioning scene under large-scale network paralysis. The method first divides a network into an alarm area and a normal area through fault alarm information collection and topology analysis, extracts key alarm sets including a cut edge link set and an edge alarm set, then converts a network fault positioning problem into a series of binary classification problems, adopts a method based on supervised classification learning, uses alarm information as an input, and via a classifier using standardized fault source-fault alarm code records in a fault diagnosis database as training samples, obtains suspected fault sources. Finally, based on the key alarm sets in the first step and a preliminary fault positioning result in the second step, low-probability fault sources are removed and corresponding guessed fault sources are added to match with current fault alarm symptoms, and the fault positioning result is modified and perfected, thereby improving accuracy of fault positioning.

The invention discloses a scholar name duplication disambiguation method which comprises the steps of classification model establishing and iteration disambiguation. In the classification model establishing step, a labeled data set is obtained by labeling based on heterogeneous academic network data, a document-to-binary-classification training data set is established based on the labeled data set, binary classification model training is carried out through a classification algorithm based on the training data set, and a document-to-binary-classification model is obtained. In the iteration disambiguation step, iteration judging is carried out on the data set which needs disambiguation with an iteration classification algorithm based on a binary classification model, a final agglomeration corresponding to a real scholar is obtained, and scholar name duplication disambiguation processing is achieved. The invention further discloses a scholar name duplication disambiguation system.

The invention provides a semi-supervised high-resolution remote sensing image scene classification method based on a generative adversarial network. The method comprises: constructing an EMGAN model:changing the discriminator of the generative adversarial network from binary classification to multi-classification to obtain an EMGAN discriminator, and adding an information entropy maximization network to the generator of the generative adversarial network to obtain an EMGAN generator; training an EMGAN model; dividing a loss function of the EMGAN discriminator into a supervision part and an unsupervised part according to whether a training image has a label or not; dividing a loss function of the EMGAN generator into a feature matching loss function and a generated image information entropy loss function; alternately training the EMGAN discriminator and the EMGAN generator; finely adjusting the VGGNet-16 model; training an SVM model; and fusing the features of the EMGAN model and the VGGNet-16 model, and performing scene classification to obtain a classification result. The remote sensing image scene classification method can effectively improve the precision of remote sensing image scene classification under the condition of few training samples.

The invention discloses a method for automatically classifying academic documents, which comprises the following steps: inputting training documents into a database, wherein the training documents comprise document classification numbers; selecting unitary characteristic words and binary characteristic words, and generating binary word pairs for the training documents; reading the training documents in the database, and respectively calculating the probability relations between the unitary and binary characteristic words and the document classification numbers, thereby forming a unitary classification dictionary and a binary classification dictionary; reading a document to be labelled, calculating the Chinese library classification number corresponding to the document according to the weight of the unitary and binary classification dictionaries, and the unitary and binary characteristic words in the document to be labelled, and automatically labelling; and dividing the unitary and binary classification result into a high-accuracy result set and a low-accuracy result set according to the degree of confidence, and outputting the classification result.

a method for improving stacking schema for classification tasks, according to which predictive models are built, based on stacked-generalization meta-classifiers. Classifications are combined to build a new scheme from at least two layers and multiclass classification problems are converted into binary classification problems. One-against-all class binarization and regression learners are used for each class model and ensemble classifiers are improved using stacking. Accuracy differences, accuracy ratio, and runtime classification in multiclass datasets are also improved and the class of a value is then predicted.

The invention relates to a defect detection and classification method based on FCNs (fully convolutional networks) and applied to galvanized stamping parts. The method comprises steps as follows: collecting samples with various types of defects; performing preliminary binary classification according to image gray standard deviation, and distinguishing qualified workpieces and defective workpieces;preprocessing the preliminarily screened samples to improve contrast, and extracting region of interests to serve as improved FCN inputs for training; calculating pixel values of output workpiece images, and setting a threshold value to judge the types of the workpiece defects and performing classification. Defect detection and classification are performed through combination of image processingand the FCNs, so that complicated preprocessing and feature extraction are avoided, and the defects of the galvanized stamping parts can be better detected and classified.

A machine-learning based artificial intelligence device for finding an estimate of patent quality, such as patent lifetime or term is disclosed. Such a device may receive a first set of patent data and generate a list of binary classifiers. A candidate set of binary classifiers may be selected and using a heuristic search, for example an artificial neural network (ANN), a genetic algorithm, a final set of binary classifiers is found by maximizing iteratively a yield according to a cost function, such an area under a curve (AUC) of a receiver operating characteristic (ROC). The device may then receive patent information for a target patent and report an estimate of patent quality according to the final set of binary classifiers.

An improved system and method is provided for determining web page quality using collective inference based on local and global web page information. A classification engine may be provided for classifying a web page using local features of a seed set of web pages and global web graph information about the seed set of web pages. A dual algorithm based on graph regularization formulated as a well-formed optimization solution may be used in an embodiment for applying collective inference for binary classification of the web page using the local web page information and global web graph information of a web page, the local web page information and global web graph information of an authoritative set of web pages, and the local web page information and global web graph information of a non-authoritative set of web pages.

The invention relates to a secure steganography method and device based on generative adversarial networks. The method includes: generating a carrier image, into which information needs to be embedded, through the generator network under a generative adversarial network framework, and judging authenticity of the generated carrier image through the discriminator network, enabling the carrier image,which is generated by the generator network, to be close to a real image through a dynamic game process of the generator network and the discriminator network; carrying out embedding of the information on the carrier image generated by the generator network; and then utilizing a steganalysis network to carry out binary classification on the input carrier image and a steganographed image to obtainaccuracy rates of classification into an original image and the steganographic image. According to the method, the generated carrier image is closer to the real image visually, a generation speed isfaster, and steganographic security can be improved.

The invention discloses a method for detecting a garbage blog. The method is designed as follows: through analyzing a cheating technology of network garbage, the method aims at an essential attribute of the garbage blog, adopts a technology of text classification in binary classification and surrounds three angles of the content characteristic of a blog text, the link character of a blog page and the time distribution character of the blog text, The method is built on the basis of the comprehensive analysis of the content of a blog webpage and carries out optimization on the operation of character extraction of the blog, thereby ensuring the higher accuracy rate of classifying the garbage blog.

The invention belongs to the technical field of communication control and communication processing, and discloses an isolated forest-based binary classification abnormal point detection method and aninformation data processing terminal. The method comprises the steps of carrying out initial static average blocking on an original data set, and calculating the density in the block and the mean density; after calculating the density in each block of the static block, reducing the data set by taking the mean density of the original data set as a threshold value; constructing an isolated forest byusing a node recursion method; performing corresponding feature extraction and datamation on the original data set, and calculating the spatial position distances between the clustering center pointand other points; adding the abnormal score calculated on the basis of the density and the distance and the abnormal score calculated on the basis of the proof information and comparing with a corresponding threshold value. According to the method, the accuracy of an abnormal point detection algorithm is effectively improved, the actual data size in the abnormal detection process can be greatly reduced, the calculation resources are saved, and the abnormal detection efficiency is improved, and the robustness of an abnormal detection algorithm is enhanced.

The invention relates to a method for automatically filtering a defective image based on multilayer feature. The preset filtering method has poor effect. The method comprises the following steps: inputting an image, preprocessing the input image and carrying out complexion detection on the preprocessed image to obtain a complexion mask image; extracting first-layer features of the images, classifying the images by adopting a first-layer binary system classification tree, carrying out trunk positioning on a suspected image and outputting a classification result; carrying out truck positioning on images which are successfully positioned by adopting a human face detection method; if the positioning is succeeded, extracting a third-layer feature and outputting an image by adopting a decision tree classifier; and positioning a truck of a human body by adopting an elliptical fitting method for the images which are not successfully positioned; after the feature is extracted, outputting classification results by adopting a binary system classification tree and finishing. According to the invention, internet industries carrying more multimedia information such as network video, social network sites, and the like can be healthily and sustainably developed.

The invention relates to the technical field of computer networks, and discloses a link prediction method based on a network structure and text information, and the method comprises the steps: S1, obtaining a structure embedding vector of a node based on a random walk node in the network structure; s2, constructing a convolutional neural network to process the text information of the nodes to obtain text information embedding vectors of the nodes; s3, performing joint embedding on the structure embedding vector and the text information embedding vector of the node; s4, generating a training set and a test set; s5, constructing a neural network for binary classification learning; s6, training a neural network; and S7, result prediction: the link prediction method based on the network structure and the text information is simple in calculation process, small in error and high in prediction accuracy.

The invention relates to a method for the training of a classifier based on weakly labeled images and for the binary classification of an image. The training of the classifier comprises the steps of automatically and iteratively determining initial regions of interest for a training set and further on refining said regions of interest and adapting the classifier onto the refined regions of interest by a classifier refinement procedure. Further on, for a query image with unknown classification, an initial region of interest is determined and refined as to maximize the probability value derived at the output of said classifier. The query image is automatically assigned a negative classification label if said probability value is lower than or equal to a predetermined first threshold. The query image is automatically assigned a positive classification label if said probability value is greater than a predetermined second threshold.

The invention discloses a satellite image segmentation method based on residual network and U-Net segmentation network. The satellite image segmentation method comprises the following steps: constructing a residual network ResNet 34; constructing U-Net segmentation network; constructing a training sample set; training the residual network ResNet 34; training the U-Net segmentation network; inputting the satellite image to be segmented into the residual network ResNet 34 for binary classification, and judging that a ship target is included; using the U-Net segmentation network to perform binarysegmentation on the positive samples in the classification result; for the negative samples in the classification result, directly outputting a single-value mask graph. According to the method, the satellite image is subjected to binary classification by using the residual network ResNet 34, and the U-Net segmentation network is used to only segment the positive samples in the classification result, and an SE-ResNet module is embedded in the U-Net segmentation for extracting finer segmentation masks and the satellite image segmentation method is high in real-time performance and segmentationprecision.