144 results about "Score vector" patented technology

The present disclosure relates to systems and methods for classifying videos based on video content. For a given video file including a plurality of frames, a subset of frames is extracted for processing. Frames that are too dark, blurry, or otherwise poor classification candidates are discarded from the subset. Generally, material classification scores that describe type of material content likely included in each frame are calculated for the remaining frames in the subset. The material classification scores are used to generate material arrangement vectors that represent the spatial arrangement of material content in each frame. The material arrangement vectors are subsequently classified to generate a scene classification score vector for each frame. The scene classification results are averaged (or otherwise processed) across all frames in the subset to associate the video file with one or more predefined scene categories related to overall types of scene content of the video file.

The present disclosure relates to systems and methods for classifying videos based on video content. For a given video file including a plurality of frames, a subset of frames is extracted for processing. Frames that are too dark, blurry, or otherwise poor classification candidates are discarded from the subset. Generally, material classification scores that describe type of material content likely included in each frame are calculated for the remaining frames in the subset. The material classification scores are used to generate material arrangement vectors that represent the spatial arrangement of material content in each frame. The material arrangement vectors are subsequently classified to generate a scene classification score vector for each frame. The scene classification results are averaged (or otherwise processed) across all frames in the subset to associate the video file with one or more predefined scene categories related to overall types of scene content of the video file.

The invention discloses a multi-operating process monitor method based on principal component analysis and support vectors data. The method establishes a uniform PCA model to various operating mixed data firstly, puts score vectors of principal component space and residual space to high dimension characteristic space. Two new statistics are established in the characteristic space for monitoring the principal component space and residual space. When the process goes wrong, a fault reconstruction method based on SVDD identifies fault. The method establishes two SVDD statistics monitor model to various operating based that the principal analysis method is used for reducing process variable dimension, reduces statistics limit of processing monitor, increases sensitivity of processing monitoring. In addition, the invention provides a fault reconstruction and identifying method aiming at detected process fault which can locate source of fault commendably, is benefit to removing fault as soon as possible, returns process to normal operation.

Disclosed herein are a method and system for training a big data machine to defend, retrieve log lines belonging to log line parameters of a system's data source and from incoming data traffic, compute features from the log lines, apply an adaptive rules model with identified threat labels produce a features matrix, identify statistical outliers from execution of statistical outlier detection methods, and may generate an outlier scores matrix. Embodiments may combine a top scores model and a probability model to create a single top scores vector. The single top scores vector and the adaptive rules model may be displayed on a GUI for labeling of malicious or non-malicious scores. Labeled output may be transformed into a labeled features matrix to create a supervised learning module for detecting new threats in real time and reducing the time elapsed between threat detection of the enterprise or e-commerce system.

The invention discloses a spelling error correction method and system of an ES search engine, and relates to the technical field of information. The method comprises the following steps of: dividing spelling content input by a user into a plurality of entries by adoption of an ansj word segmentation device; carrying out error detection on each entry, if an error entry exists, searching error models matched from the error entry from an error model library, and obtaining correction candidate words corresponding to the error entry from the matched error models; calculating a score, under each matched error model, of each correction candidate word according to the matched error models, and forming a score vector according to the score under each matched error model; processing the score vectors by adoption of an L2R model so as to generate scores of the error models, and determining a total score of each correction candidate word according to the scores of the error models and a language model; and determining the correction candidate word with the highest score in the total score as a correct candidate word, and displaying the correct candidate word. The method and system disclosed by the invention can improve the correctness of spelling error correction.

The invention discloses a partial least squares-based Gaussian regression soft measurement modeling method. The method can be applied to industrial processes with relatively strong time-varying characteristic, coupling, nonlinearity, hysteresis and other complex characteristics. The method comprises the following steps of: firstly, carrying out dimensionality reduction on multi-element input dataon the basis of a partial least squares method, and selecting proper score vectors as input of a Gaussian process regression model; secondly, selecting and combining covariance functions, and constructing different types of Gaussian process regression soft measurement models to predict output data; and finally, evaluating prediction ability of the models by using test set data. Modeling results ofpaper-making wastewater treatment process data prove that a partial least squares-based dimensionality reduction technology for measured variables can improve the prediction ability of the Gaussian process regression model; and the Gaussian process regression models constructed by different covariance functions provide multiple options for effluent indexes, so that the method is more suitable forcomplex and changeable paper-making wastewater treatment environment.

In a center apparatus, a feature space projection processing unit performs a feature space projection process for probe data corresponding to a road section which are stored in a current probe data storage unit to extract the feature data, and a change point detecting unit; an event section partitioning unit and an event assigning unit determine a road section corresponding to the feature data, and assign the event information to the determined road section; and an event information distributing unit distributes the event information assigned to the road section. In a vehicle-installed terminal apparatus, a probe data partitioning unit and an orthogonal component decomposition unit performs processes of partitioning and orthogonal component decomposition of the probe data using a feature score vector obtained from the center apparatus, to thereby reduce the probe data to be uplinked.

The invention discloses a user dynamic preference oriented commodity sequence personalized recommendation method. The method comprises the steps of extracting a commodity sequence under the same usersimilarity score to construct a commodity score vector; extracting a user sequence of similar scores of the same commodity to obtain a user score vector; combining user personal information, commoditybasic attribute information, user and article comments and commodity pictures; achieving feature extraction of a user and a commodity based on multi-task learning, taking a user feature vector and afeature vector of a historical commodity sequence of the user feature vector as input, achieving generation of the commodity sequence by training a coder-decoder, and accurately learning recommendation of the optimal commodity sequence in combination with a search strategy. According to the invention, based on the multi-modal user-commodity data, the user features and the commodity features are highly extracted and fused, personalized recommendation of the commodity sequence for user preferences is realized, and the user experience is improved.

The invention relates to an action recognition method based on a skeleton sequence. The method comprises the steps that all skeletons of the skeleton sequence are projected to the front surface, the side surface and the top surface of a Descartes orthogonal system according to three-dimensional information, and a skeleton distribution diagram is generated; time information is added into the skeleton distribution diagram through color change; convolutional neural network model training is conducted on the skeleton distribution diagram, which is generated based on a training dataset and formed on the three projection surfaces and to which the time information is added, through the convolutional neural network; for each test sample, three Scores vectors are calculated according to three trained convolutional neural network models and targeted at the skeleton distribution diagram which is formed on the three projection surfaces and to which the time information is added; after the Scores vectors of the three projection surfaces are added, the category to which the maximum value belongs is taken as a subordinate category of the video sequence. Through the method, human actions can be recognized accurately and reliably.

The present invention relates to a computation method for detecting remote sequence homologies. The method comprises the following steps: First, a training sequence set of positive and negative examples each having a corresponding binary label is provided together with a database query sequence set (typically large) of unlabeled sequences. Second, each sequence in the training set is converted into a fixed-length vector of real values by computing pairwise sequence similarity scores with respect to the vectorization set to obtain vectorized training sequences each having corresponding binary labels. Third, the vectorized training sequences (along with their binary labels) are used to train a discriminative classification algorithm to obtain a trained discriminative classification algorithm. Fourth, the the database of unlabeled sequences are converted into pairwise score vectors, using the vectorization set to obtain vectorized database sequences. Finally, each vectorized database query sequence is presented to the trained discriminative classification algorithm to produce predicted classifications for the database query sequence.

The invention discloses a diagnosis method for sensor faults of a motor train unit braking system. The method includes: subjecting collected historical sensor signals of a motor train unit to EEMD (ensemble empirical mode decomposition) processing, and creating energy feature vectors of the historical sensor signals; training an FDA (fisher discriminant analysis) model according to the energy feature vectors so as to obtain FDA model parameters; colleting on-line sensor test signals, subjecting the on-line sensor test signals to EEMD processing, and creating energy feature vectors of the on-line sensor test signals; computing FDA score vectors of the energy feature vectors of the sensor test signals according to a projection matrix in the FDA model parameters; classifying the FDA score vectors based on the parameters of the FDA model, and determining fault categories of the on-line test signals. By the method, the defect of modal aliasing effect in the EMD method is overcome, signal features can be extracted effectively, the single FDA model is used for fault classification, and complexity of the SVM (support vector machine) based fault classification algorithm is lowered.

Embodiments of the present application disclose an item recommendation method, apparatus, computer device, and storage medium. The method includes: determining a target user group from a plurality ofuser groups according to a score vector of the target user, and then calculating a similarity value between the target user and each user in the user group to which the target user belongs; Determining a similar user of the target user according to the similarity value; Obtaining items scored by similar users but not scored by target users as recommendation items and generating a first item recommendation table according to the recommendation items; According to the similarity value between the target user and the similar user, the scoring value of each recommended item by the similar user andthe corresponding time attenuation factor, calculating the scoring value of each recommendation item according to the preset calculation formula; sorting a plurality of recommendation items accordingto a preset sorting rule according to an item score value to generate a second item recommendation table and push to a target user. This method can improve the accuracy of item recommendation and effectively avoid the lag problem of recommended projects.

The invention relates to the technical field of machine learning, and discloses an adversarial sample detection method and device, computing equipment and a computer storage medium. The method comprises the following steps: acquiring a training sample and a corresponding training sample label, wherein the training sample label comprises a normal sample and an adversarial sample; inputting the training sample into a target model to obtain a first prediction score vector of the training sample; adding N times of random disturbance to the training sample to obtain N groups of contrast training samples; respectively inputting the N groups of contrast training samples into a target model to obtain a second prediction score vector of each group of contrast training samples; constructing featuredata according to the first prediction score vector and the second prediction score vector of each group of contrast training samples; training a classification model according to the feature data andthe training sample label corresponding to the feature data to obtain a detector; and detecting the input test data according to the detector. According to the embodiment of the invention, reliable detection of the adversarial sample can be realized according to the detector.

A method and system for contrast inflow detection in a sequence of fluoroscopic images is disclosed. Vessel segments are detected in each frame of a fluoroscopic image sequence. A score vector is determined for the fluoroscopic image sequence based on the detected vessel segments in each frame of the fluoroscopic image sequence. It is determined whether a contrast agent injection is present in the fluoroscopic image sequence based on the score vector. If it is determined that a contrast agent injection is present in the fluoroscopic image sequence, a contrast inflow frame, at which contrast agent inflow begins, is detected in the fluoroscopic image sequence based on the score vector.

The invention provides a bilinear neural network false news detection method and system based on style guidance. The method comprises the following steps: acquiring a news text to be subjected to network false news detection, quantifying language style characteristics of the news text through a neural network to obtain a style vector of the news text, and inputting the news text into a text characteristic extractor to obtain a text vector of the news text; inputting the style vector and the text vector into a bilinear neural network, the bilinear neural network comprising a bilinear function for modeling a correlation between the style vector and the text vector to obtain a style-style of the news text; using the maximum score vector in the style-text feature matrix to form a guide vector,and inputting the guide vector to the full connection layer to determine the false news label of the news text. The learning process of the deep learning model is guided according to the language style of the false news generality, and the recognition accuracy and the generalization performance of the model are improved.

The invention relates to an accurate audience advertisement pushing method and system based on artificial intelligence and a readable storage medium, and the method comprises the steps: obtaining userportrait data through big data analysis, carrying out the behavior modeling of the user portrait data, and constructing a user data platform; collecting user behavior information, collecting user traffic advertisement materials, establishing a feature word bank, and generating a user behavior log; establishing an advertisement recommendation model through a user behavior log, inputting user behavior information into the recommendation model, generating a plurality of characteristic spectrums, outputting nonlinear data through an activation function, and extracting similarity characteristics of a user and an advertisement; obtaining a user score vector according to the user behavior information; judging whether the sparsity of the user scoring vector is greater than a preset threshold value or not; if the similarity is smaller than the preset threshold value, calculating the similarity between the advertisement materials and the user interest set to obtain user preference information,weighting the user preference information, obtaining the sequence of the advertisement materials in which the user is interested, and generating a push list.