98results about How to "Reliable classification" patented technology

The invention relates to a neural network text classification method based on a multi-knowledge map. The method comprises the following steps: inputting a text in a training set into a long-term and short-term memory network to obtain a context vector of the text; extracting entities from each text in the training set, and performing entity matching in the knowledge map; calculating the attentionweights of matched entities and relationships in the knowledge map under context vector to obtain the overall entity vector, the overall relationship vector and the fact triple vector of the text; computing the fact triple vectors under different knowledge maps, calculating the attention weights of these fact triples, obtaining the text representation vectors and inputting them to the full connection layer of the neural network, and using a classifier to calculate the probability of each text belonging to different categories to train the network; using the trained deep neural network model to predict the category of the text to be predicted. This method improves the model's understanding of text semantics and can classify text content more reliably, accurately and robustly.

A system and method are provided for psychographically profiling a target population of individuals, programmably implemented in a server apparatus. The psychographic segmentation portion includes a questionnaire processing part, a segment solution part, and a typing tool part. The questionnaire processing part acquires a plurality of responses to a predetermined set of questionnaire questions by an initial pool of target population members. The questionnaire processing part uses a combination of weighted statistical methodologies to arrive at a rule-based decision on whether the question would serve as a good wedge variable. Optionally, the rule based decision is then compared to an expert's opinion and the weights and threshold parameters of the weighted statistical combination are adjusted accordingly to more accurately match the expert's decision. The identified wedge variables are used for effectively segregating the target population into groups of homogenous individuals. The typing tool part selectively generates an abbreviated set of questions from the predetermined set of questionnaire questions for profiling an individual respondent to at least one of the segments.

A character reading technique recognizes character strings in grayscale images where characters within such strings have poor contrast, are variable in position or rotation with respect to other characters in the string, or where portions of characters in the string are partially obscured. The method improves classification accuracy by improving the robustness of the underlying correlation operation. Characters are divided into regions before performing correlations. Based upon the relative individual region results, region results are combined into a whole character result. Using the characters that are read, a running checksum is computed and, based upon the checksum result, characters are replaced to produce a valid result.

The current invention concerns a method for detecting cancerous cells and / or classifying cells in a cell sample comprising the following steps: providing a cell sample; obtaining holographic information from said cell sample by digital holographic microscopy (DHM); deriving at least one cellular parameter from said holographic information, and; classifying said cells of cells sample; characterized in that said classification occurs by appointing a Scoring Factor to said cells of cell sample, based on said cellular parameters. In a second aspect, a system for the detection of cancerous cells and / or classification of cells in a cell sample is provided, employing the method as disclosed in the invention. In a final aspect, a method for updating and / or improving a database comprising thresholds linked to holographic information and the database related thereof is equally disclosed.

A character reading technique recognizes character strings in grayscale images where characters within such strings have poor contrast, are variable in position or rotation with respect to other characters in the string, or where portions of characters in the string are partially obscured. The method improves classification accuracy by improving the robustness of the underlying correlation operation. Characters are divided into regions before performing correlations. Based upon the relative individual region results, region results are combined into a whole character result. Using the characters that are read, a running checksum is computed and, based upon the checksum result, characters are replaced to produce a valid result.

The unattended surveillance device may include a housing to be positioned for unattended surveillance, a video camera associated with or carried by the housing to capture video, and an image processor carried by the housing and cooperating with the video camera. The image processor extracts moving objects in the foreground of the captured video, generates a profile image or sequence of profile images of the extracted moving objects, compresses the sequence of profile images, and generates a surveillance information packet based upon the compressed sequence of profile images. Also, a wireless transmitter or transceiver may be associated with the image processor to transmit the surveillance information packet to a surveillance monitoring station.

A digital image is automatically combined with text data by automatically classifying the image according to a predetermined set of categories, and automatically selecting text data from a repository, wherein the selected text data match the category of the image according to a predetermined criterion.

The invention provides a multi-label classification method of electronic medical record based on symptom extraction and feature representation. Considering the influence of disease and symptom and thecorrelation between symptoms on the multi-classification problem of disease label of electronic medical record, two different symptom representation methods are adopted: using TF-IDF builds symptom vectors and learns symptom vectors using word2vec. The two symptom vector sequences extracted from the electronic medical record are respectively used as input sequences of the two two-way LSTM models,and the two-way LSTM models are trained. For the electronic medical records of unknown disease tags, two symptom vectors corresponding to the symptoms are extracted to form two symptom vector sequences, and two trained bi-directional LSTM models are input respectively to obtain two probability vectors. The weighted combination of the two probability vectors is used to obtain the final classification vector. This method has good classification effect and applicability.

The invention is directed to a method for classifying a moving vehicle. The object of the invention is to find a novel possibility for classifying vehicles moving in traffic which allows a reliable automatic classification based on two-dimensional image data. This object is met according to the invention in that an image of a vehicle is recorded by means of a camera and the position and perspective orientation of the vehicle are determined therefrom, rendered two-dimensional views are generated from three-dimensional vehicle models which are stored in a database in positions along an anticipated movement path of the vehicle and are compared with the recorded image of the vehicle, and the vehicle is classified from the two-dimensional view found to have the best match by assignment of the associated three-dimensional vehicle model.

A detection system is provided which is configured to have a transmitter capable of interacting with an object by generating a field, and a multiplicity of receivers operative to measure changes in the environment caused by the object's response to the generated field and mounted to a closed-looped belt, which is displaceable in a proximity to the object.

The invention is an a document classification and marking engine / method that functions in a real-time compatible mode with off-the-shelf word processors, e-mail programs and presentation or other document development software applications. The software engine is used for the security classification of sensitive or national security classified information in electronic format and is enhanced by methods and processes that ensure that the software classification engine considers all document informational elements regardless of attributes assigned to the text that may hide text from the user. The software engine provides a complete and reliable document classification determination interface method based on user selections and uniquely codes the full text classification determination in a persistent manner within the electronic shell of the document in real-time and dynamically displays the text based full classification determination in the banner of the host document development application. The unique codes of the full classification embedded in the electronic shell of the document enable effective and reliable software processes and methods that establish controls for access, movement, storage etc. for electronic documents, as well as virtual electronic security perimeters, on a computer, networks of computers and / or among computer networks and domains of networks. The full or complete and persistent classification codes embedded in the document shell also enables reliable software processes and methods that immediately warn or alert security personnel of a beach of a physical security perimeter between or among computers networks or domains of networks established to protect the information contained in electronic document format.

The invention relates to a power system transient state voltage stability evaluating method based on misclassification cost classified learning, which belongs to the power system stability analyzing and evaluating field. Based on dynamic measurement datum of a synchronous phasor measuring unit, a key subsequence closely related to the power system state is extracted from a time sequence constituted of a large amount of dynamic measurement datum; by setting different misclassification costs of stable power system and unstable state, a weight coefficient is introduced into a learning sample; a decision-making tree algorithm infused with the sample weight coefficient is used to perform classified learning so as to acquire a decision-making tree model; and the decision-making tree is used for online monitoring to implement evaluation to the power system transient state voltage stable situation.

The invention discloses an equipment fault classification method based on dynamic weight combination of integrated increment, and relates to an equipment fault diagnosis technology. The method is mainly divided into the following parts: performing unbalanced data processing, performing wavelet denoising and reconstruction, performing feature extraction through empirical mode decomposition, building a long-short-term memory neural network, building a support vector machine model, using a support vector machine for dynamically adjusting the weight of each classifier in the long-short-term memoryneural network and support vector machine combined model, and using an integrated incremental model for realizing rapid dynamic incremental learning. Based on the research, the dynamic weight combination classification model based on the integration increment is finally provided and applied to rolling bearing equipment fault diagnosis, and the classification accuracy of equipment fault diagnosisis improved.

The current invention concerns a method for detecting cancerous cells and / or classifying cells in a cell sample comprising the following steps: providing a cell sample; obtaining holographic information from said cell sample by digital holographic microscopy (DHM); deriving at least one cellular parameter from said holographic information, and; classifying said cells of cells sample; characterized in that said classification occurs by appointing a Scoring Factor to said cells of cell sample, based on said cellular parameters. In a second aspect, a system for the detection of cancerous cells and / or classification of cells in a cell sample is provided, employing the method as disclosed in the invention. In a final aspect, a method for updating and / or improving a database comprising thresholds linked to holographic information and the database related thereof is equally disclosed.

The invention relates to a system (100) arranged to delineate the acute intracerebral haematoma in non-contrasted CT images in two stages. The first stage, performed by the extraction unit (110), employs an analysis of gray values of the image data in order to extract the candidate region. The candidate region may comprise both an acute haematoma and other regions having similar gray values, e.g., regions resulting from partial volume effects at the interface of the bony structures of the skull and the brain. The novel second stage, performed by the classification unit (120), analyzes spatial features of the candidate region such as, for example, the size, shape, and connectedness to the skull bone of the candidate region. Using spatial features of the candidate region improves the correctness of classification of the candidate region as a true or false acute haematoma.

A method for operating e.g. a robot. The method includes the steps: acquiring a sequence of images of a space surrounding the robot within which a potential object is situated; ascertaining, using a first machine learning system, a respective first variable characterizing the potential object as a function of each of the images; ascertaining, using a second machine learning system, at least one second variable that characterizes the potential object as a function of a plurality of the first variables characterizing the potential object; controlling the robot as a function of the second variable characterizing the potential object. A computer program, a device for carrying out the method, and a machine-readable storage element on which the computer program is stored, are also described.

A bibliographic-information impression word is generated from a bibliographic information segment about selected one of music pieces. An acoustic feature quantity of an audio signal representing the selected music piece is calculated. A feature-quantity impression word is generated from the calculated acoustic feature quantity. A degree of conformity between the bibliographic-information impression word and the feature-quantity impression word is determined. Both the bibliographic-information impression word and the feature-quantity impression word are. selected as final impression words when the determined conformity degree is greater than a predetermined threshold value. One is selected from the bibliographic-information impression word and the feature-quantity impression word as a final impression word when the determined conformity degree is not greater than the predetermined threshold value. A signal representing the final impression word or words is stored into a storage in relation to the selected music piece.

The present invention relates to a device, system and method for pulsatility detection. The proposed device comprises an input interface (30) for obtaining image data of a scene, said image data comprising a time sequence of image frames, an extraction unit (31) for extracting a time-varying signal indicative of a periodic physiological phenomenon from a region of interest of said image data, a transformation unit (32) for transforming said time-varying signal into a spectral signal, a partitioning unit (33) for partitioning the spectral signal into at least an in-band sub-signal covering a first frequency range of said spectral signal and an out-band sub-signal covering a second frequency range of said spectral signal, wherein said first frequency range comprises at least a frequency range of the periodic physiological phenomenon, an analysis unit (34) for separately deriving an in-band measure from the in-band sub-signal and an out-band measure from the out-band sub-signal, the in-band measure and the out-band measure representing a descriptor, and a classifier (35) for classifying said region of interest as a pulsatile region of a living being or as a non-pulsatile region based on the descriptor.

The present invention discloses a method for analyzing a liquid cell sample, comprising the steps of: a) providing at least one liquid cell sample in a sample vial; b) obtaining data linked to the cells in the sample by performing differential digital holographic microscopy on said liquid cell sample in said sample vial. In a second aspect, the invention also provides for a system for analyzing a liquid cell sample, comprising (i) a differential digital holographic microscope comprising illumination means, a differential interferometer and a digital recording device connected to a processing device such as a computer; (ii) at least one exchangeable sample vial comprising a liquid cell (iii) a movable sample vial holder; characterized in that (iv) said sample vial holder is adapted to receive said sample vial; (v) said sample vial holder is adapted to position said sample vial such that the focal plane of the objective lens of said differential digital holographic microscope lies in the vial.

The embodiment of the invention provides a visual classification method and device for Internet of Things malicious software and an electronic device, and the method comprises the steps: carrying outthe binary file extraction of to-be-classified Internet of Things malicious software, obtaining a binary malicious code, and carrying out the bit combination of the binary malicious code to generate agray-scale image; and performing feature extraction on the grayscale image, and based on the extracted grayscale image features, utilizing a deep neural network classification model based on a 2-bitnetwork to obtain a classification category to which the to-be-classified Internet of Things malicious software belongs. According to the embodiment of the invention, a classification model based on a2-bit network is adopted, model size and calculation time can be substantially reduced, the processed malicious software is converted into the gray level image, feature extraction is carried out according to the gray level image, visual features of extracted features can be improved, the classification process is made to have higher interpretability, and classification of the Internet of Things malicious software can be more reliably and efficiently accomplished.

An apparatus according to the invention for the classification of physiological events has a signal input for the input of a physiological signal representing or constituting a physiological event and a classification unit 1 for classifying the physiological signal on the basis of its signal shape. The classification unit 1 includes a transformation unit 3 which is designed to carry out transformation of the physiological signal in such a way that as the output signal it outputs a number of values representing the physiological signal and based on the transformation; and a probabilistic neural network which is connected to the transformation unit 3 to receive the values and which contains a number of event classes which represent physiological events and which in turn are each represented by a set of comparative values, which probabilistic neural network is adapted on the basis of the comparison of the values with the comparative values to effect an association of the physiological signal represented by the values with one of the event classes.

The invention provides a visual classifier construction method with the consideration of characteristic reliability. The method includes the steps that video characteristics of each video sample in a video sample set are extracted to obtain a video characteristic set; a label is endowed to each video sample to indicate that the video sample belongs to a first category or a second category; reliability assessment is carried out on each video sample to obtain a reliable factor of each video sample; based on the video characteristic set, the label of each video sample and the reliable factor of each video sample, a weighted support vector machine algorithm is utilized to obtain a visual classifier. The visual classifier construction method can be applied to services of internet harmful video filtration, video supervision and the like to maintain safety of contents of the internet.

The invention discloses a fruit sugar degree nondestructive detection device and method based on a silicon-based multispectral chip, and relates to the technical field of apple sugar degree nondestructive testing. A fruit bracket is arranged above the detection device; a fruit sample to be detected can be placed; a light source, a multispectral camera adopting a silicon-based multispectral chip and a detection module are arranged in the detection device; the multispectral camera is connected with the detection module and forms an included angle with the light source, the multispectral camera collects scattered multispectral image data of the fruit sample to be detected and inputs the scattered multispectral image data into the detection module, and the detection module detects the sugar degree of the fruit sample to be detected according to the received multispectral image data. The adopted silicon-based multispectral chip is made of silicon-based wafers and has the advantages of smallsize, light weight, simple structure, low cost and the like, and the sugar degree detection device is more convenient to operate and higher in detection speed. According to the fruit sugar degree detection method, the sugar degree of the fruit can be graded relatively reliably without a complex mathematical model and long-time operation.

A method to establish virtual security perimeters for classified electronic documents on a computer system. The security perimeters are based upon a full classification determination of all informational content of an electronic document file. The full classification determination is uniquely coded to identify a classification value, the classification regime used to classify the information as well as ownership of the electronic information of the electronic document, and is embedded in the electronic document. The classification determination code is matrixed with identification codes for elements of a file management system and used to control computer events initiated on a computer involving the electronic document. Computer events on computers are monitored for the coded full classification determination. The code scheme is also used to identify a breach of a security perimeter on a computer of an unauthorized classified electronic document and warning of the breach.

The invention discloses an all-media news intelligent cataloguing method based on multi-modal information fusion. The method comprises the steps: carrying out the unified representation and fusion understanding of video information, voice information, subtitle bar information and character information in all-media news, so as to obtain multi-modal fusion features; and realizing automatic slicing, automatic cataloguing description and automatic scene classification of news by using multi-modal fusion features. The full-media news-oriented automatic comprehensive cataloguing method has the beneficial effects that a complete process of full-media news-oriented automatic comprehensive cataloguing is realized, and the accuracy and generalization of the cataloguing method are improved by generating the splitting mark, the news cataloguing description, the news classification label, the news keyword and the news character on the basis of fusion of multiple modalities of the video, the audio and the text; the manual cataloguing time is greatly reduced.

The invention discloses an intuitive microearthquake event classification method. The technical scheme includes: firstly, performing de-noising processing on original observation records and increasing signal-to-noise ratio of the records; finding out possible microearthquake events by a long time window and short time window proportion method; collecting first arrivals of the microearthquake events according to continuity of synthetic shafts, performing arrangement processing on the acquired first arrivals and finally superimposing all the first arrivals of the microearthquake events to display. According to distribution of the superimposed first arrivals, the microearthquake events similar in first arrival forms are defined as the same microearthquake event group, unified selection of parameters in post-processing is facilitated, and the events with the curve forms different from the overall trend are defined as stray events or invalid microearthquake events. The intuitive microearthquake event classification method has the advantages that classification of the microearthquake events is convenient, intuitive and reliable.

A method of image analysis, includes: obtaining a sequence (37;51) of images, each represented by pixel data; performing a vision-based analysis on at least one of the sequence (37;51) of images to obtain data for classifying a state of a subject represented in the images;—determining at least one value of a physiological parameter of a living being represented in at least some of the sequence (37;51) of images; and classifying a state of the subject using the data obtained with the vision-based analysis and the at least one value of the physiological parameter. The at least one value of the physiological parameter is determined through analysis of image data from the same sequence (37;51) of images from which the at least one image on which the vision-based analysis is performed is taken. A method of enabling remote photoplethysmographic analysis includes: obtaining a sequence (37;51) of images from at least one camera (3), each represented by pixel data representative of reflected ambient light in at least a limited range of wavelengths; and, for at least one measurement zone (41), providing a signal representative of at least variations in a time-varying value of a combination of pixel values at least a number of image points in the measurement zone (41) for use in determining at least one value of a physiological parameter. At least part of a selected subject represented in the sequence (37;51) of images is tracked, and a directable light source (4) illuminating at least part of the selected subject is controlled.

There is provided an implantable cardiac pacing system or other cardiac monitoring system having an enhanced capability to classify intracardiac signals through a combination of DSP techniques and software algorithms. The implantable device has one or more DSP channels corresponding to different signals which are being monitored. Each DSP channel most preferably amplifies the incoming signal, converts the signal from analog to digital form, digitally filters the converted signals to provide a filtered signal, operates on the filtered signal to provide a slope signal, determines from the filtered and slope signals when an intracardiac event has been detected, signal processes the filtered and slope signals for a predetermined analysis interval after threshold crossing, and generates a plurality of wave parameters corresponding to the signal. The generated wave parameters are further operated on by a programmable algorithm to classify the detected event based upon DSP-generated parameters, and then monitor or detect the onset, development or presence of an undesired heart condition in a patient. The system may further provide for the delivery of treatment, storage of intracardiac data, or provision of a warning to a patient or physician in response to the detection of such a heart condition.