128 results about "Entropy model" patented technology

A system and method for protecting user privacy using social inference protection techniques is provided. The system executes a plurality of software modules which model of background knowledge associated with one or more users of the mobile computing devices; estimate information entropy of a user attribute which could include identity, location, profile information, etc.; utilize the information entropy models to predict the social inference risk; and minimize privacy risks by taking a protective action after detecting a high risk.

A system and method for protecting user privacy using social inference protection techniques is provided. The system executes a plurality of software modules which model of background knowledge associated with one or more users of the mobile computing devices; estimate information entropy of a user attribute which could include identity, location, profile information, etc.; utilize the information entropy models to predict the social inference risk; and minimize privacy risks by taking a protective action after detecting a high risk.

The invention discloses a Chinese word segmentation method, which comprises the following steps of: performing word segmentation on a Chinese text according to word semantics, segmenting ambiguous fields and outputting a first text string taking words as units; and identifying and combining Chinese names in the first text string to generate a second text string taking words as units. The ambiguous fields are segmented by combining a dictionary rule method with a statistical method; and the ambiguous fields are segmented and the names are identified by word standard a maximum entropy model in the statistical method. The invention also discloses a Chinese word segmentation system, which comprises a word segmentation module, a name identification module and the like. The method and the system improve word segmentation efficiency and accuracy.

The invention discloses a debris flow risk degree evaluation method. The method comprises a step of determining the characteristic parameters associated with debris flow three elements, a step of establishing the comprehensive information evaluation system of the three elements, obtaining a three-element initial information evaluation matrix, and calculating the information entropy of the three elements through matrix operation and an entropy method, a step of establishing a debris flow information entropy model, taking three sub information entropy as an input factor, through BP neural network effect, and outputting debris flow information entropy, a step of defining a debris flow risk degree level standard according to the relation between the information entropy theory and the fact whether debris flow occurs, and carrying out risk degree evaluation on a search object. The invention discloses a debris flow probability model, the mutual interaction mechanism of debris flow three elements in a debris flow inoculation process can be comprehensively reflected, the complex nonlinear and dynamic processes of debris flow can be represented, the risk degree of the research object (single channel / regional debris flow) can be forecasted.

The invention discloses an event extraction based sensitive information monitoring method. The method comprises the steps of 1) creating a trigger word dictionary and an event element role dictionary; 2) training models marked with training corpus by the machine learning method to acquire a maximum entropy model MT for determining the type of an event and a maximum entropy model MR for extracting event elements from event sentences; 3) filtering corpuses of the event to be extracted according to the trigger words, and treating the sentences which are matched with the set trigger words as the candidate events; 4) classifying the candidate events through the maximum entropy model MT to obtain the event sentence with the set event type; 5) extracting each element term of the event from the event sentences obtained in step 4) according to the event element role dictionary and the maximum entropy model MR so as to finish the event extracting; matching the extracted event with the monitored event; if matching successfully, determining that the extracted event is sensitive information. With the adoption of the method, the monitoring efficiency of the sensitive information is greatly increased.

The invention provides a method for assessing voltage sag sensitivity of sensitive equipment. The method includes steps of determining an uncertain region of voltage tolerance curves of the sensitive equipment to be assessed; creating an entropy model and a maximum entropy model for the uncertain region by the aid of a probability density function; importing historical monitoring data and solving the maximum entropy model by an optimization algorithm to obtain analytic expressions of the probability density function; integrating the analytic expressions of the probability density function to obtain the voltage sag sensitivity of the sensitive equipment to be assessed. Artificial assumption that voltage sag amplitudes and duration are independent from one another is omitted in the entropy model and the maximum entropy model. The method has the advantages that the type of the probability density function is not set in advance, the artificial assumption that variables, namely the voltage sag amplitudes and the duration, are independent from one another is omitted, and the accuracy of the voltage sag sensitivity, which is assessed on the basis, of the sensitive equipment can be improved.

The invention discloses a method for modelling based on a maximum entropy model and named entity recognition. The method comprises that training texts marked by named entity are input; the role marking of the characters in the training texts is carried out so as to obtain the character role marking of the training text; according to the character role marking, the characteristic items of the characters are established; the characteristic items of the characters are input into a modelling tool of the maximum entropy to obtain the data model which is based on the character role marking. The method requires no word division, thus solving the problems that when the named entity recognition is carried out, the word division errors and information loss which is caused by the word division errors affect the recognition effect.

The invention relates to a highway real-time operation risk calculation method. According to the highway real-time operation risk calculation method, a multi-factor-integrated entropy model is adoptedto analyze and calculate the real-time operation of the traffic flow and environment of highways. In accordance with the entropy model, collected data are pre-processed on the basis of real-time traffic flow data and real-time environmental change data acquisition; and on the basis of the real-time operation conditions of the highways, correlation analysis, a random forest model, normalization processing, and the like are adopted with traffic flow risk factors and external environmental factors integrated so as to establish a risk entropy model that can analyze the real-time operation conditions of the highways. With the highway real-time operation risk calculation method of the invention adopted, risk identification, evaluation and analysis can be performed on the real-time operation risks of the highways; reference can be provided for the real-time risk management of the highways; and administration staff can conduct risk disposal and prevention so as to avoid or reduce economic andproperty losses caused by high-risk situations.

The invention belongs to the technical field of named entities recognition, and discloses a Korean named entities recognition method based on a maximum entropy model and a neural network model. The method comprises the steps that a prefix tree dictionary is built, when any one combined noun template or any one proper noun template is matched in an input sentence, the combined noun template or the proper noun template are recognized into a target word; the target word is obtained in a target word selection module, the target word is searched in an entity dictionary, and when only one subclass is matched, the subclass serves as a tag of the target word; the maximum entropy model is adopted, and various linguistics information is utilized; a feedforward neural network model is constructed; adjacency words form an entity tag through a template selection rule. All data used in the method is extracted in a training corpus with tags and a field-independent entity dictionary, the data is very easily migrated to other application fields, and the performance cannot be reduced obviously.

The invention provides a novel prosodic structure boundary division forming method based on prosodic phrases. The method combines machine learning with rules to greatly improve the accuracy of the prediction of Chinese text prosodic structure boundary. Prosodic phrase boundaries are firstly identified on the premise that input files goes through word segmentation and part of speech tagging, then prosodic word boundaries are formed by combining prosodic phrase boundary information, and finally a plurality rules are artificially added to carry out integral modification. In prosodic phrase and prosodic word boundary identification, characteristics are respectively designed and selected for establishing a characteristic template, and a prosodic phrase model and a prosodic word model are established by utilizing the maximum entropy algorithm for respectively identifying prosodic boundaries of two stages. In addition, aiming at the errors in identification of a maximum entropy model, an optimal rule is selected by utilizing an error-driven rule learning method to further improve the accuracy. Based on the method, the prosodic structure boundary division forming method based on prosodic phrases is provided, and the method can effectively improve the accuracy of prosodic structure prediction and the naturalness of speed synthesis.

The invention discloses an event extraction method based on maximum entropy. The method comprises the following steps of (1), constructing a trigger word dictionary and an event element role dictionary; (2), as for labeled training corpus, training a model by use of a machine learning method, acquiring a maximum entropy model MT which is used for judging event types and a maximum entropy model MR which is used for extracting event elements from event sentences; (3), filtering corpus needing event extraction according to trigger words, and utilizing sentences which are matched with the set trigger words as candidate events; (4), classifying candidate events by virtue of the maximum entropy model MT and acquiring the event sentences which belong to a set event type; (5), extracting each element word of events from the event sentences which are obtained in the step (4) according to the event element role dictionary and the maximum entropy model MR, thereby finishing event extraction. The event extraction method disclosed by the invention is extensive in use and high in accuracy; by virtue of the event extraction method, the event extraction effect is greatly improved.

The invention discloses network abnormality detection method and system based on information entropy and a sliding window. The method comprises the following steps: defining the size of a time window and a sliding distance of the time window; progressively calculating the entropy and the entropy ratio of each time window orderly according to the sliding window; judging that network abnormality occurs when data mutation is generated in the time window or does not accord with the previous law if the entropy of the time window is smaller than a given threshold or the entropy ratio is greater than the given threshold. By adopting the method and the system, an information entropy model and a sliding window technology are led in, so as to find out the problem of network abnormality; the network abnormity can be quickly found out, the model is simplified to a certain extent, and the network abnormality can be quickly found out.

The invention discloses a phrase division model establishing method, a statistical machine translation method and a decoder. The phrase model establishing method comprises the following steps of: acquiring a training sample from a bilingual corpus; inputting the acquired training sample to a parameter training tool of a maximum entropy model, and performing parameter training to acquire a weight parameter of the maximum entropy model; and substituting the weight parameter into the maximum entropy model to generate a phrase division model.

The invention has the observation characteristic that road traffic time domain characteristic and road traffic space domain characteristic are used for fusing multiple data so as to obtain the service level predicting result by a maximum entropy model, so that the predicting result is ensured to be more accurate; and by adopting the technology for prediction, the traffic control department can issue the crowding situation of the roads in the urban area, the traffic state and the service level in advance according to the predicting result, so that the invention provides reference for going out, leads the public to avoid the rush hour and the crowding road sections, is beneficial to inducing and evacuating the traffic, effectively eases the traffic pressure, and provides decision support for traffic guidance.

A method for calculating the steering ratio at road cross based on genetic algorithm and maximum entropy model includes inputting flow rate data in certain time slot at study road cross and adjacent road cross, setting operator and control parameter of said algorithm and defining its search space and total frequency of solving solution, initializing calculation environment of said algorithm, executing genetic algorithm, calculating entropy of obtained solution and judging whether said total frequency is up or not if precision of obtained solution is satisfied and using solution corresponding maximum entropy as final solution if it is.

A method for performing conditional maximum entropy modeling includes constructing a conditional maximum entropy model, and incorporating an observation confidence score into the model to reduce an effect due to an uncertain observation

The subject invention provides for systems and methods that facilitate optimizing one or mores sets of training data by utilizing an Exponential distribution as the prior on one or more parameters in connection with a maximum entropy (maxent) model to mitigate overfitting. Maxent is also known as logistic regression. More specifically, the systems and methods can facilitate optimizing probabilities that are assigned to the training data for later use in machine learning processes, for example. In practice, training data can be assigned their respective weights and then a probability distribution can be assigned to those weights.

The invention discloses an application for detecting and tracking an infrared weak object under a complicated background. The application is characterized by comprising the following steps of: 1. suppressing clutters and keeping the topological structure of an image, and constructing a bionic vision weighted entropy model with an adjacent airspace and a preferred direction to realize conversion for the image from a grey mode to an entropy model; 2. analysing the movement state of the weak object with burst and stationary characteristics, and constructing a self-adaptive entropy flow target movement estimation model meeting the maneuvering features of the weak object by virtue of the nonlinear diffusion smoothing and self-adaptive local restriction criterion of an entropy flow to realize the approximation of an estimation speed to the real movement state of the weak object; 3. searching a weak object tracking method adopting generic multi-feature and measurement, and constructing a multi-feature fused sequential filter model to realize accurate, robust and real-time identification for the weak object. The invention discloses a self-adaptive entropy flow detection and tracking algorithm for the infrared weak object, and enriches a detection and tracking technology for the weak object.

The invention discloses a classification method and system of emotions of news readers. The classification method comprises the following steps: acquiring a news text and a comment text as well as word characteristic information from target linguistic data; fusing the word characteristic information and converting the word characteristic information into available linguistic data with a corresponding format of a maximum entropy model; dividing the available linguistic data into training linguistic data and testing linguistic data according to a pre-set rule, and dividing the training linguistic data into marked samples and unmarked samples; training the marked samples to obtain a maximum entropy model; classifying emotion classes of the unmarked samples by using the maximum entropy model to obtain posterior probability of each emotion class corresponding to the unmarked sample; carrying out emotion class marking on the unmarked samples with the preset quantity and maximum uncertainty of the posterior probability to form new marked samples, and updating the current marked samples and unmarked samples; and circulating the last step until all the unmarked samples are marked. The classification method and system can be used for efficiently classifying the emotions of the news readers when the scale of marking the linguistic data is relatively small.

The invention discloses a method for predicting a camellia oleifera suitable area. The method comprises the following steps that: S1: obtaining camellia oleifera distribution point data; S2: obtaining the environmental data of the distribution point; S3: selecting a maximum entropy model to construct an ecological niche model, and dividing a suitable level for the potential distribution area of the camellia oleifera according to a distribution probability value for ecological niche model prediction; and S4: carrying out the reliability analysis of a prediction result. By use of the method, existing camellia oleifera distribution data and environment data can be utilized, the maximum entropy model is adopted to construct the ecological niche model, the camellia oleifera suitable area is predicted, a prediction result is accurate so as to be favorable for knowing a camellia oleifera distribution rule, and the method has a high reference value for guiding camellia oleifera production and searching wild camellia oleifera resources.

The invention provides a sentiment analysis engine based on fine-granularity attributive classification. An adopted model is a maximum-entropy model. An algorithm thought of a sentiment polarity intensity quantitative method of the system comprises the following steps: firstly, utilizing a traditional sentiment dictionary to calculate a sentiment tendency value of each word through a method of word frequency statistics; and then, utilizing the sentiment tendency value of the word to design a corresponding formula so as to calculate the sentiment tendency value of the word. The sentiment analysis engine is suitable for the MicroBlog, the WeChat and the like on the Internet, calculates a PMI (Purchase Management Index) value by aiming at a possible situation that a sentiment word lacks of object attributes so as to determine an association probability between an evaluation object attribute class and the sentiment word, realizes a purpose that a reasonable attribute class is assigned to sentiment information which lacks of the evaluation object attribute, and enables sentiment summarization calculation to be more reasonable and effective and is better in perception performance.

The invention belongs to the field of computer vision mode recognition and image understanding and relates to an automatic image segmentation method of continuous quantum goose group algorithm evolution pulse coupling neural network system parameters. The method comprises the steps that a minimum combination weighting entropy model of automatic image segmentation of the evolution pulse coupling neural network system parameters is established; a continuous quantum goose group population space is initialized; a simulation quantum rotating door is used for updating the position of each wild goose; the position of each wild goose corresponds to a pulse coupling neural network system parameter, a pulse coupling neural network system is activated for image segmentation, and a fitness value of a new position of an i wild goose is computed; the history optimal quantum positions and the history optimal positions of all wild geese are updated; whether the maximum iteration algebra is reached is checked; and a pulse coupling neural network model is substituted to carry out segmentation on images and output the images after segmentation. The method has the advantages of being small in computing amount, high in convergence rate and high in optimizing capacity.

The invention relates to an HEVC quantization parameter optimizing method based on a total code rate and information entropy model. The total code rate and information entropy model is used in the code rate control process to adjust the quantization parameters (QP)s in the HEVC coding process according to different sequence characters of different videos, the purposes of improving the rate-distortion performance and controlling the code rate more accurately are achieved, and the coding complexity cannot be obviously affected. The quantization parameter optimizing method is also suitable for calculating the QPs of H.264 / AVC, AVC and other video coding standards.

The invention discloses a bus operation status data adjustment processing method, an intelligent terminal and a storage medium. The bus operation status data adjustment processing method comprises thesteps of: comparing bus shift departure time data corresponding to an expected bus departure time table with dynamically-predicted bus stop-return time data, so as to obtain and a result of shift andbus time table feasibility; constructing a bus dynamic scheduling entropy model based on data of shifts not expected to depart according to an expected time table, and adjusting a bus time table according to the predicted bus stop-return time data; and verifying the bus dynamic scheduling entropy model through example analysis, and adjusting a load factor of buses before and after the shift according to validity of the bus dynamic scheduling entropy model. Based on the bus dynamic scheduling entropy model predicting stop-return time, the departure intervals at the early stage are prolonged inadvance so as to eliminate shift breaks, the fairness and rationality of the adjustment of the departure intervals in the early stage are realized, the bus shift breaks are reduced, the passenger load factors of the bus shifts are balanced, the stability of bus operation is improved, and passengers are facilitated to travel.

The invention provides a key distribution and reconstruction method based on the mobile internet. The method includes the steps of S1, constructing an identity-based key packaging model and a verifiable random function; S2, distributing keys; S3, reconstructing the keys. A calculable collusion-proof equilibrium method is designed, a protocol collusion-proof game-with-entropy model is constructed, and thus collusive attack of participants is prevented; a cryptographic protocol communication game model is constructed, and the defect that password protocols constructed in the broadcast communication network are unable to be implemented in the mobile internet is overcome; the key packaging mechanism applicable to the verifiable random function is studied, the rational key sharing protocol requiring no pubic key infrastructure is designed, and calculation fairness and delivery in the mobile internet is guaranteed; finally, the protocol is subjected to security analysis and proving through the theory of provable security.

The invention discloses a punctuation addition method and a device in speech recognition to solve the problem that an obtained recognition result through the speech recognition is lack of pragmaticality. The method comprises a step of extracting features of present words in a sentence obtained through the speech recognition, a step of recognizing the extracted features of the present words in a preset maximum entropy model to obtain identification characters after the present words, and a step of choosing punctuations corresponding to the identification characters after the present words from a known identification character set according to an incidence relation of the obtained identification characters and each punctuation to be added after the present words. The punctuations (the punctuation can be empty) which need to be added after the present words are forecasted according to logical relations between the present words and several words before and after the present words and the preset maximum entropy model. The pragnaticality of the speech recognition is improved in the speech recognition result after the punctuations are added.

The invention discloses an emotion component analyzing method and system based on emotion distribution learning. The method includes marking a basic emotion of each image; calculating a correlation coefficient between emotion distribution vectors and each pair of basic emotion mark vectors and calculating a weight matrix based on the correlation coefficient; by using an image feature vector and emotion distribution thereof as a training set, combining a maximum entropy model with Jeffrey divergence and weight matrix and combining with two regularization items for generating a target function, and optimizing the target function for obtaining a parameter model for forecast of emotion distribution; and performing feature extraction on an image waiting for emotion distribution estimation and using the model obtaining through training for emotion distribution prediction. If the value corresponding to the emotion mark is greater than a constituent ratio of the virtual mark, the emotion is judged to be a main emotion component. By using the method and system provided by the invention, a model for emotion component analysis can be obtained quickly and effectively through training and which emotions are contained in an expression and the proportion of the emotions can be calculated out.

A traffic incident detection method based on depth learning and entropy model relates to the technical field of intelligent traffic. The traffic incident detection method based on depth learning and entropy model includes the following steps: 1) training a convolutional neural network model for traffic incident classification; 2) performing incident classification on input video stream images or video subsegments according to the convolutional neural network model; 3) calculating the entropy value in a period of time according to the result of the incident classification; 4) judging whether atraffic incident occur according to the entropy value. Compared with the prior art, the reliability parameter is calculated according to the proportional and inverse relation between different incident probability, the entropy value and stability. The mutation of the image is analyzed from the global and local characteristics of video images, the occurrence of traffic incidents is detected with the advantages and methods of CNN and the characteristics of mutation detection by entropy model, and the method has characteristics of fast speed and accurate detection.

The invention discloses an electric power supply system based on the maximum entropy principle. A method for calculating probabilistic energy flow of a gas integrated energy system comprise the stepsof: solving electric power; obtaining steady state energy flow of natural gas integrated energy system, node voltage, branch power, node pressure of natural gas system and pipeline flow correspondingto the reference operating point, and calculating the reference sensitivity matrix of electric power system and natural gas system. The central moments of each order are transformed into semi-invariants of each order, and the electric power is considered simultaneously. Coupling relationship between natural gas; According to the product of the semi-invariant and the sensitivity matrix, it is transformed into the semi-invariant of the node voltage, the branch power, the node pressure of the natural gas system and the disturbance part of the pipeline flow rate, and then transformed into the final central moments of each order. Based on the final order center moments and the maximum entropy model, the probabilistic energy flow results for electricity-natural gas integrated energy systems. Theinvention can effectively solve the problem of electric power. Probabilistic Energy Flow of Natural Gas Integrated Energy System.

The embodiment of the invention provides a method and a system for eliminating ambiguity of Chinese word segmentations. The method comprises the following steps: segmenting a to-be-segmented word, thereby acquiring an initial segmenting result; extracting a segmentation ambiguity point according to the initial segmenting result; constructing a new segmented word containing the segmentation ambiguity point and calculating the maximum entropy model score of the new segmented word; judging if the new segmented word is a valid segmented word according to the maximum entropy model score of the new segmented word; correcting the initial segmenting result with the valid segmented word. According to the embodiment of the invention, the defect of requirement for a large amount of training corpus data and ambiguity corpus of the prior art is overcome and the word segmentation effect can achieve the practical precision.