1628results about How to "Improve classification performance" patented technology

Quantitative object and spatial arrangement-level analysis of tissue are detailed using expert (pathologist) input to guide the classification process. A two-step method is disclosed for imaging tissue, by classifying one or more biological materials, e.g. nuclei, cytoplasm, and stroma, in the tissue into one or more identified classes on a pixel-by-pixel basis, and segmenting the identified classes to agglomerate one or more sets of identified pixels into segmented regions. Typically, the one or more biological materials comprises nuclear material, cytoplasm material, and stromal material. The method further allows a user to markup the image subsequent to the classification to re-classify said materials. The markup is performed via a graphic user interface to edit designated regions in the image.

The invention discloses a short text classification method based on a convolution neutral network. The convolution neutral network comprises a first layer, a second layer, a third layer, a fourth layer and a fifth layer. On the first layer, multi-scale candidate semantic units in a short text are obtained; on the second layer, Euclidean distances between each candidate semantic unit and all word representation vectors in a vector space are calculated, nearest-neighbor word representations are found, and all the nearest-neighbor word representations meeting a preset Euclidean distance threshold value are selected to construct a semantic expanding matrix; on the third layer, multiple kernel matrixes of different widths and different weight values are used for performing two-dimensional convolution calculation on a mapping matrix and the semantic expanding matrix of the short text, extracting local convolution features and generating a multi-layer local convolution feature matrix; on the fourth layer, down-sampling is performed on the multi-layer local convolution feature matrix to obtain a multi-layer global feature matrix, nonlinear tangent conversion is performed on the global feature matrix, and then the converted global feature matrix is converted into a fixed-length semantic feature vector; on the fifth layer, a classifier is endowed with the semantic feature vector to predict the category of the short text.

A hybrid machine learning methodology and system for classification that combines classical random forest (RF) methodology with discriminant analysis (DA) techniques to provide enhanced classification capability. A DA technique which uses feature measurements of an object to predict its class membership, such as linear discriminant analysis (LDA) or Andersen-Bahadur linear discriminant technique (AB), is used to split the data at each node in each of its classification trees to train and grow the trees and the forest. When training is finished, a set of n DA-based decision trees of a discriminant forest is produced for use in predicting the classification of new samples of unknown class.

The invention discloses a human face detection method and a detection device based on multi-task cascade-connection convolution neural network, wherein the method comprises: establishing a cascade-connection multi-level convolution neural network; using the human face front samples, the human face back samples, some parts of the human face and the human face's key point samples as the training samples to train the multi-level convolution neural network to learn the tasks of human face categorizing, human face area position regression and human face's key point positioning; and utilizing the well trained multi-level convolution neural network to make human face detection from the to-be-detected image wherein in the training stage, both the online manner and the offline manner are combined to extract the human face back samples as the training samples. According to the invention, based on the cascade-connection multi-level convolution neural network, it is possible to learn the characteristics with stronger robustness, and at the same time, through the combination of the online manner and the offline manner to extract the back samples, the categorizing capability of the network is enhanced, so that the detection capability and the accuracy of the network are increased, and that the running speed of the method in an actual product is ensured.

A photo category classification method including dividing a region of a photo based on content of the photo and extracting a visual feature from the segmented region of the photo, modeling at least one local semantic concept included in the photo according to the extracted visual feature, acquiring a posterior probability value from confidence values acquired from the modeling of the at least one local semantic concept by normalization using regression analysis, modeling a global semantic concept included in the photo by using the posterior probability value of the at least one local semantic concept; and removing classification noise from a confidence value acquired from the modeling the global semantic concept.

The invention discloses a pronunciation quality assessment and error detection method based on the fusion of multiple characteristics and multiple systems, which carries out assessment and error detection on pronunciation quality by a method utilizing multiple characteristic parameters to describe pronunciation quality and utilizing multiple inspecting systems to mutually fuse, and comprises the following steps: recognizing voice and automatically segmenting and aligning the voice; extracting the characteristic parameters used for voice quality assessment and error detection; acquiring pronunciation quality assessment and error detection model training data; training a plurality of pronunciation quality assessment and error detection systems; fusing a plurality of pronunciation quality assessment and error detection systems; and assessing pronunciation quality and detecting pronunciation errors. By utilizing the invention, multiple voice characteristics are effectively utilized, and multiple assessment and detection system are fully utilized and perform information fusion, thereby maximally exerting the advantages of various characteristics and systems, and ensuring the accuracy and reliability of pronunciation assessment and error detection.

The invention provides a Chinese text classification method based on a super-deep convolution neural network structure model. The method comprises the steps of collecting a training corpus of a word vector from the internet, combining a Chinese word segmentation algorithm to conduct word segmentation on the training corpus, and obtaining a word vector model; collecting news of multiple Chinese news websites from the internet, and marking the category of the news as a corpus set for text classification, wherein the corpus set is divided into a training set corpus and a test set corpus; conducting word segmentation on the training set corpus and the test set corpus respectively, and then obtaining the word vectors corresponding to the training set corpus and the test set corpus respectively by utilizing the word vector model; establishing the super-deep convolution neural network structure model; inputting the word vector corresponding to the training set corpus into the super-deep convolution neural network structure model, and conducting training and obtaining a text classification model; inputting the Chinese text which needs to be sorted into the word vector model, obtaining the word vector of the Chinese text which needs to be classified, and then inputting the word vector into the text classification model to complete the Chinese text classification.

Methods and apparatus are provided for classifying data packets in data processing systems. A first packet classification method determines which of a plurality of predefined processing rules applies to a data packet, where each rule is associated with a range of possible data values in each of a plurality of dimensions (X,Y) corresponding to respective data items in the packet format. For each dimension (X,Y), it is determined which of a set of predefined basic ranges contains the corresponding data value (I1, I2) from the packet, where the basic ranges correspond to respective non-overlapping value ranges between successive rule range boundaries in the dimension. For the basic range so determined for each dimension, a corresponding basic range identifier is selected from a set of predefined basic range identifiers corresponding to respective basic ranges in that dimension. For each of at least two dimensions (X,Y), the basic range identifiers comprise respective pD-bit strings generated independently for that dimension by a process of deriving a primitive range hierarchy based on the rule ranges in that dimension. The resulting basic range identifiers, one for each dimension, are then combined to produce a search key which is supplied to a ternary content-addressable memory (5). In the memory (5), the search key is compared with a set of ternary rule vectors, each associated with a particular rule and derived for that rule from the aforementioned hierarchies, to identify at least one rule which applies to the data packet. A second method classifies data packets according to the values in respective data packets of a single, predetermined data item (DA) in the data packet format, where a plurality of classification results are predefined for respective ranges of values of the data item (DA). Here the data item (DA) in the packet is first segmented. The resulting segments are then equated to different dimensions (X,Y) of a multidimensional packet classification problem and are processed in a similar manner to identify a classification result for the packet.

The invention discloses an identification method of emotional tendency of network comment texts and a convolutional neutral network model. The method comprises the steps as follows: grabbed network comment texts constitute a data set; word segmentation and text preprocessing are performed; all words subjected to text preprocessing are trained, and word vector representation of all words is obtained; the convolutional neutral network model is constructed and trained on a training set selected from the data set, and network parameters are updated with a back-propagating algorithm; in each step of training, noise is added to word vectors of an input layer for construction of adversarial samples, adversarial training is performed, and network parameters are updated with a random gradient descent algorithm; a classification model is obtained through repeated iteration to identify the emotional tendency of the network review texts. The convolutional neutral network model is used in the method and comprises the input layer, a convolution layer, a pooling layer and a classification layer. The adversarial samples can be classified correctly and the identification accuracy is improved.

The invention relates to a hierarchical classification method for internet flow, comprising the following steps: firstly, coarse grained classification is conducted as follows: a flow statistical property method based on machine learning is adopted for quickly classifying network flow, the network flow is divided into applied categories with different characteristics, if the flow distinguished by the coarse grained classification needs to be used, the flow can be output directly, and if the flow needs further classification, next step is conducted; and secondly, fine grained classification is conducted as follows: classification is further conducted on the applied categories distinguished by the coarse grained classification, the network flow can be intelligently, accurately and effectively identified in real time, good integrity and expandability are realized, and the requirements on the internet flow classification with different application targets and classification grain size can be met.

The present invention discloses an integrated support vector machine mixed intelligent diagnosis method of machine failure. Said method includes the following steps: respectively adopting and lifting small wave packet, utilizing frequency band and empirical mode decomposition process to decompose vibration signal according to the eigen mode component and extract time domain statistical character of decomposed signal to form total characteristic set; providing characteristic distance evaluation technique and characteristic evaluation index; utilizing said characteristic evaluation index to select most sensitive characteristic set from the total characteristic set; using said most sensitive characteristic set as diagnosis characteristics and creating integrated support vector machine mixed intelligent diagnosis model so as to implement intelligent diagnosis of machine failure state.

A text classification method comprises following steps: dividing the initial training text collection into a plurality of subsets including the text in the same category based on the category, extracting the corresponding probability topic model from each subset; generating new text to balance the categories of the subsets by the corresponding probability topic model; constructing a classifier based on the balance training text collection corresponding to plural subsets; and processing text classification by the classifier. The invention can improve the classification effect of the text classification method under the condition of data skew.

The present invention provides a Chinese text classification method based on an attention mechanism and feature enhancement fusion, and belongs to the technical field of data mining. The Chinese textclassification method based on the attention mechanism and the feature enhancement fusion is characterized in that: a feature enhancement fusion Chinese text classification model based on the attention mechanism, a long short-term memory (LSTM) network and a convolutional neural network (CNN), and a feature difference enhanced attentional algorithm model are proposed; and according to the featureenhancement fusion Chinese text classification model, double-layer LSTM and CNN modules are used to sequentially perform enhancement fusion on the text features extracted by using the attention mechanism, the richness of the extracted text features can be continuously enhanced, and the contained text features can be more comprehensive and detailed, so that the recognition ability of the model to the Chinese text features is improved.

The invention discloses a remote sensing image classification method based on an attention mechanism deep Contourlet network, and the method comprises the steps: building a remote sensing image library, and obtaining a training sample set and a test sample set; then, setting a Contourlet decomposition module, building a convolutional neural network model, grouping convolution layers in the model in pairs to form a convolution module, using an attention mechanism, and performing data enhancement on the merged feature map through a channel attention module; carrying out iterative training; performing global contrast normalization processing on the remote sensing images to be classified to obtain the average intensity of the whole remote sensing images, and then performing normalization to obtain the remote sensing images to be classified after normalization processing; and inputting the normalized unknown remote sensing image into the trained convolutional neural network model, and classifying the unknown remote sensing image to obtain a network output classification result. According to the method, a Contourlet decomposition method and a deep convolutional network method are combined, a channel attention mechanism is introduced, and the advantages of deep learning and Contourlet transformation can be brought into play at the same time.

The invention discloses a Chinese network review emotion classification method based on an integrated study frame. According to the method, a part-of-speech combination mode, an order-preserving sub-matrix mode and a frequent word sequence mode are adopted as input characteristics, in the level of characteristics, factors of the influence of Chinese word order information, interval phrase characteristics and the sentence length are considered, and the characteristic vector sparsity problem is solved through semantic similarities; the problem that many review text characteristics exist is solved, the inter-base-classifier independence is guaranteed, and the classification performance of base classifiers is improved as much as possible; a base classifier algorithm constructed based on product attributes is adopted to comprehensively review emotion information of each attribute in a text, and then the sentence-level emotional tendency of reviews is judged, so that a final classification result is more accurate. The Chinese network review emotion classification method based on the integrated study frame is applicable to e-commerce network review emotion classification in various fields, can make a potential consumer know evaluation information of a commodity before purchase and can also make a merchant better sufficiently know the consumer's opinion, and therefore the service quality is improved.

The invention discloses a weak supervision fine-grained image classification method of a multi-branch neural network model. The weak supervision fine-grained image classification method is characterized by the steps of: firstly, randomly dividing a fine-grained image data set into a training set and a test set in proportion; secondly, positioning a local region with potential semantic informationby using a local region positioning network; thirdly, respectively inputting an original image and the positioned local region into a deformable convolution residual network and a rotation invariant coding direction response network to form a feature network of three branches, respectively training the three branches, and respectively carrying out backward propagation learning on the three branches based on cross entropy loss; and finally, combining intra-branch loss and inter-branch loss to optimize the whole network, and performing classification prediction on the test set. According to theweak supervision fine-grained image classification method, the negative influence of various changes such as posture, visual angle and background interference on a classification result is reduced, and a better effect is achieved on a fine-grained image classification task.

The invention discloses a three-dimensional convolutional neural network (3D CNN) based video classifying method and belongs to the technical field of video processing. According to the method, a video is sampled at equal intervals to obtain a plurality of video segments, a video database is amplified, three-dimensional video segments are directly input into a 3D CNN, and time domain and space domain characteristics of the video are extracted, so that the limitation of a conventional video classifying method in manually selecting video characteristics and video modeling modes is improved. A parallel distributed 3D CNN multi-classification model lowers the complexity in learning the 3D CNN and enables a classification system to realize distributed parallel computation more conveniently. Relatively high identification rate can be achieved with only fewer video segments based on a 3D CNN multi-classification system, and videos not belonging to any type can be classified into new type, so that the classification error of the new type is avoided.

The invention discloses a method for automatically identifying a breast tumor area based on an ultrasound image. The method comprises the following steps of acquiring the ultrasound image of the breast, and preprocessing the ultrasound image; segmenting the ultrasound image subjected to preprocessing through an image segmentation method to obtain a plurality of segmented subareas; extracting a grey level histogram, texture features, gradient features and morphological features of the ultrasound image, and combining the grey level histogram, the texture features, the gradient features and the morphological features of the ultrasound image with two-dimensional position information to obtain high-dimensionality feature vectors; selecting the most effective feature subset of the high-dimensionality feature vectors through feature ordering based on biclustering and a selection method; performing learning classification on the selected most effective feature subset through a classifier, and then automatically identifying the breast tumor area. By means of the method, the breast tumor area can be identified automatically from segment results of the breast tumor ultrasound image, therefore, automation performance of computer-aided diagnosis is improved, manual operation of clinical doctors is reduced, and subjective influence of clinical doctors is reduced.

The invention discloses a text classification method based on a cyclic convolution network. The text classification method comprises the following steps: step 1, representing context vectors of all words by using a bidirectional cyclic network; step 2, combining the context vectors and a word vector of a current word into the representation of the current word; step 3, extracting the most important context information by using a maximum pool technology to obtain text representation; step 4, carrying out text classification by using the text representation. According to the method disclosed by the invention, more word order information in a text can be kept and a long-distance text dependence relation is captured; semantics of the words can be accurately described and the words and phrases, which have the greatest influences on the text classification, are found by the maximum pool technology, so that the text classification accuracy rate is effectively improved. A test shows that the efficiency of the method is averagely improved by 1% on the aspect of a plurality of text classification data sets.

The invention discloses a Chinese text sorting method based on correlation study between sorts. The method comprises the following steps of: firstly, dividing words of a document and performing rough selection on characteristics by computing word frequencies; secondly, further determining representative word items according to discrimination indexes between the word items and sorts so as to form characteristic word items which are finely selected; thirdly, training the document to be expressed by a tfidf weight and a discrimination index weight according to an index of the characteristic word items; fourthly, establishing a group of two-sort sorters corresponding to different projection vectors and training to obtain a code array expressing the correlation between two-sort sorters; and finally, projecting a multi-vector expression of a new document to all the two-sort sorters, introducing the code array, computing the similarity between each sort and the document, and outputting the maximum of the similarity as a sort judging result of the new document. The new document is sorted based on a correlation studying result between the sorts, and the running efficiency of an algorithm is improved on the premise of ensuring the sorting performance.

The invention discloses a support vector machine based surface electromyogram signal multi-hand action identification method. The method comprises the following main steps: 1) obtaining electromyogram data, performing smooth filtering on a signal, and generating data samples through sampling windows of different scales; 2) by taking the data samples as units, extracting a novel multi-feature set containing 19 time domain, frequency domain and time-frequency domain features from each data sample, and performing normalization and minimum redundancy maximum correlation criterion based feature selection on eigenvectors; 3) designing a Pearson VII generalized kernel based support vector machine classifier and optimizing parameters of a support vector machine by using a cross validation based coarse grid search optimization algorithm; and 4) training a classification model by using data samples in a training set and optical classifier parameters obtained in the parameter optimization process of the step 3) and inputting data samples in a test set into the classification model to perform classification testing.

The invention discloses a short text characteristic expanding method based on a semantic atlas. The method includes the steps: performing subject modeling by the aid of a training data set of a short text, and extracting subject term distribution; reordering the subject term distribution; building a candidate keyword dictionary and a subject-keyword semantic atlas; calculating comprehensive similarity degree evaluation of candidate keywords and seed keywords based on a link analysis method, and selecting the most similar candidate keywords to finish expanding the short text. Compared with a short text characteristic representation method based on a language model, the method is simple to operate and high in execution efficiency, and semantic correlation information between the keywords is sufficiently used. Compared with a traditional short text characteristic representation method based on a word bag model, the problems of data sparseness and semantic sensitivity are effectively relieved, and the method is independent of external large-scale auxiliary training corpus or a search engine.

The invention discloses an emotion classifying method fusing an intrinsic feature and a shallow feature. The emotion classifying method is characterized in that the intrinsic feature of fusion Doc2vec and the shallow feature of TF-IDF are used for representing features of a text. By adopting a fusion method, the problem of unclear expression of a fixed word feature in the Doc2vec is solved, the problem that semantics among words is not considered in the TF-IDF method is also solved, and the expression of a text vector specific to the text is clearer. An SVM classifying method is adopted, so that better classifying performance of a classifier is achieved. The method is used for solving an emotion classifying problem, so that the emotion classifying accuracy can be improved remarkably.

The invention discloses a hyper-spectral image classification method based on recurrent neural network with the object to solving the problems that in prior art, the input characteristic determination ability is weak and that the extraction of local spatial characteristics is not complete. The method comprises the following steps: 1) extracting the spatial texture characteristics and the sparse representation characteristics of a hyper-spectral image and piling and combining them as the low-level characteristics; 2) extracting from the low-level characteristics the sample local spatial sequence characteristics; 3) according to the local spatial sequence characteristics, creating a recurrent neural network model; and utilizing the training sample local spatial sequence characteristics to train the recurrent neural network model parameters; and 4) inputting the testing sample local spatial sequence characteristics into the well-trained recurrent neural network model; obtaining the highly abstract high-level semantic characteristics and obtaining the classification information of the testing sample. According to the deep learning method of the invention, the correct efficiency for hyper-spectral image classification is increased and the method can be used for vegetation investigation, disaster monitoring, map making and intelligence obtaining.

A calibration device is presented for use in a surveillance system for event detection. The calibration device comprises an input utility for receiving data indicative of an image stream of a scene in a region of interest acquired by at least one imager and generating image data indicative thereof, and a data processor utility configured and operable for processing and analyzing said image data, and determining at least one calibration parameter including at least one of the imager related parameter and the scene related parameter.

The invention discloses a small-sample polarized SAR ground feature classification method based on a deep convolutional twin network, and mainly solves a problem that a conventional method is low in classification precision because the number of polarized SAR data mark samples is smaller. The method of the invention comprises the steps: 1), inputting a to-be-classified polarized SAR image and a real ground object mark of the to-be-classified polarized SAR image, and carrying out the Lee filtering; 2), extracting an input feature vector from the filtered to-be-classified polarized SAR data, andcarrying out the dividing of a training sample set and a test sample set; 3), carrying out the combination of each two samples in the training sample set, and obtaining a sample pair training set; 4), building the deep convolutional twin network, and carrying out the training of the deep convolutional twin network through the training sample set and the sample pair training set; 5), carrying outthe classification of the samples in the test set through the trained deep convolutional twin network, and obtaining the classes of ground features. According to the invention, the method expands thetraining set under the twin configuration, achieves the extraction of the difference features, enables the classification precision of a model to be higher, and can be used for the target classification, detection and recognition of a polarized SAR image.

The invention relates to the technical field of image processing, and proposes a brain disease classification system based on a self-attention mechanism, aiming at solving the technical problems of low classification accuracy caused by the complicated process of preprocessing, feature extraction and feature selection of magnetic resonance images required in the classification and diagnosis of brain diseases. For this purpose, the brain disease classification system based on the self-attention mechanism in the invention comprises the following steps: pre-processing the acquired human brain magnetic resonance images of the brain disease patients to obtain the gray matter density map of the human brain; using a pre-constructed brain disease classification model to classify the gray matter density map to obtain a brain disease category of the brain disease patient, wherein The pre-constructed brain disease classification model is a three-dimensional convolutional neural network model basedon the self-attention mechanism. The system shown in the embodiment of the invention can quickly and accurately classify the categories of brain diseases.

The invention discloses a text classification method based on block partition and position weight. The method comprises the following steps of: after performing basic pre-processing on an input training or test text, extracting section information from the text; performing statistical analysis on block information by taking one section as a basic text block; performing block partition on the text content again, according to the block size distribution or pre-defined block ratio, including operation, such as combining text blocks; extracting characteristic words and quantized weight, acquiring the posterior probability of the characteristic words on class, analyzing the distribution of the characteristic words having the biggest posterior probability class and according with a text class label, and generating a text vector; and finishing classification model training or text classification with a classifier. The method can be applied to a text representation stage of a text classification system. The text classification effect is improved through the rich and traditional expression of the text content when the text vector is created by using the characteristic words.

The invention discloses a text feature quantification method based on comentropy, a text feature quantification device based on comentropy, a text classification method and a text classification device. The text feature quantification method comprises the following steps that: the weight of each feature word in a document is calculated according to the word frequency of feature words in a text document and the comentropy distributed on different text classes; meanwhile, the inter-class distribution entropy of the feature words is calculated in different modes according to the unbalance performance of the scale of each class of a text set; in addition, the inverse document frequency is introduced as required according to the distribution features of each feature word in the text set; local word frequency factors are properly reduced, so that the weight distribution of each feature word in the document is reasonable; and the feature differences of different classes of texts are sufficiently reflected by generated document feature vectors. The text feature quantification device and the text classification device disclosed by the invention have a plurality of options or parameters; and the optimum text classification effect can be achieved through regulation. The text feature quantification method has the advantages that the text classification accuracy is improved, and the performance on different text sets is stable.