35 results about "Statistical cluster" patented technology

Provided are, among other things, systems, methods and techniques for forecasting information. One representative embodiment involves obtaining data that has been provided by each of a number of individual people; dividing the people into clusters by utilizing a statistical clustering technique; calculating statistics of the data in each of at least some of the clusters; and combining the statistics for such clusters so as to obtain a combination forecast.

A digital computer system stratifies in a set of patients, based on a set of observations. The observations can include physical, biochemical, histological, genetic, and gene-expression data, among other types of information. Adjustments can be made to account for the possibility that observations of several patients may begin at different points in the progression of their respective disease processes. Once these adjustments are made, the data are subjected to a statistical cluster analysis. Each cluster of patients potentially represents a different disease stratum, with its own underlying cause, optimum therapy, and prognosis. Once the strata are defined and patients are assigned to them, adjustments to the data can be refined. The cluster analysis then can be repeated, and so an iterative process of stratification and staging takes place.

The invention discloses a semi-automatic knowledge map construction method, Most of the existing relationship extraction methods rely on the pre-determined relationship type system, This process is complex and time-consuming, the invention is based on dependency analysis, aiming at several Chinese sentence patterns, combined with a semantic dictionary, While exporting open relationships, semantictagging of words in relation, The semantics of unknown words are inferred based on statistics, and the semantic relation patterns on a large number of corpus are statistically clustered to form a relation type system. In this process, most of the links are carried out automatically, in which the results of semantic annotation and relation clustering of unknown words can be checked manually. Compared with the existing open relation extraction method, the invention optimizes and expands, and the extraction of the open relation and the formation of the semantic relation type complement each other, so as to improve the accuracy rate of the two.

A technique for monitoring source addresses through statistical clustering of packets is disclosed. In one particular exemplary embodiment, the technique may be realized by a method for monitoring source addresses through statistical clustering of packets. The method may comprise identifying at least part of a source address of a packet. The method may also comprise searching at least one recorded source address based on the at least part of the source address, the at least one recorded source address being organized into at least one cluster. The method may further comprise routing the packet if the at least part of the source address falls within one of the at least one cluster and the one of the at least one cluster contains at least a predetermined number of source addresses.

The invention relates to a POI name determination system based on clustering and a method thereof, wherein the method comprises the following steps: capturing address data from network data, the address data includes name field and address information; clustering name fields corresponding to the same address information according to key words; counting frequency of the clustered name fields occurred in each cluster, using the frequency as a second frequency; determining the POI name of the cluster corresponding to the address information based on the second frequency. By the provided technical scheme, users can quickly and precisely search the POI name corresponding to the POI address at the same longitude and latitude, thereby improving the user experience.

The invention discloses a method for retrieving a three-dimensional object, comprising the following steps: carrying out bipartite graph maximum matching on the three-dimensional retrieval object view sets and the view sets in a database, filtering the view sets not satisfying the preset matching conditions in the database to obtain the remaining view sets; carrying out bipartite graph optimal matching on the three-dimensional retrieval object view sets and each remaining view sets to acquire the distances between the three-dimensional retrieval object view sets and each remained view sets; sequencing each remaining view sets and outputting each sequenced remaining view sets as the retrieval results. The embodiment of the invention adopts pretreatment of statistical clustering, combines bipartite graph maximum matching and bipartite graph optimal matching to acquire the distances between the three-dimensional retrieval object view sets and each remaining view sets, thus improving the correctness of three-dimensional object retrieval, and simultaneously causing the three-dimensional object retrieval to be capable of not depending on collecting environment information, and the three-dimensional object retrieval method to be more vastly applied.

The disclosed embodiments provide a system for processing data. During operation, the system obtains a set of feature additions and an evaluation metric for assessing the performance of a statistical model. Next, the system automatically builds treatment versions of the statistical model using a set of baseline features for the statistical model and feature combinations generated using the feature additions. The system then uses a hypothesis test and a fixed set of feature values to compare a baseline value of the evaluation metric for a baseline version of the statistical model that is built using the set of baseline features with additional values of the evaluation metric for the treatment versions. Finally, the system outputs a result of the hypothesis test for use in assessing an impact of the feature combinations on a performance of the statistical model.

The invention relates to an image segmentation processing method based on an area matching optimization K-means clustering algorithm, comprising steps of firstly extracting vehicle feature points of front and back frame images, and then comparing an area overlapping situation of front and back frame vehicles, extracting positions of the feature points in an area overlapping region and the positions of the rest feature points, respectively calculating a mean value of the two groups of the feature points as two types of initial clustering central points to be segmented, and then implementing K-mean segmentation, correcting a classification situation of the feature points in the area overlapping region according to an output clustering result, and meanwhile, judging whether the clustered vehicles are reasonable or not; and if not, re-clustering the clustering result and recounting clustering centres, ending clustering segmentation until the reasonable vehicles are found, and then feeding back a tracking result. The method is on the basis of area matching optimization, and adopts fixed clustering numbers to implement the segmentation; and vehicle targets obtained by the K-mean segmentation do not need the next round of matching treatment, thereby a processing speed is accelerated, and time is saved.

The invention discloses an adaptive two-dimensional clustering-based signal pre-sorting method. The method can be used for the signal pre-recognition of a new system radar in a complex environment. The method includes the following steps that: an antenna array is adopted to perform two-dimensional positioning on a target, so that distance and azimuth information can be obtained; the initial valuesof the thresholds of the distance and azimuth are set; similarity between the distance dimension and azimuth dimension of the target is calculated; two-dimensional clustering of the distance and azimuth is obtained according to the calculated similarity and thresholds; and the thresholds of the distance and the azimuth are adaptively adjusted, the average value of the similarity between the distance dimension and azimuth dimension after clustering is put into statistics until the average value is smaller than the threshold of the distance or the azimuth, and signal pre-sorting is completed. According to the adaptive two-dimensional clustering-based signal pre-sorting method of the invention, pre-positioning and pre-sorting are sequentially carried out, so that the limitations of traditional radar signal sorting are broken; and the two-dimensional clustering processing of the distance and azimuth is utilized, so that increasingly serious batch increase and batch missing phenomena of signal sorting can be eliminated, and therefore, the effectiveness of radar signal sorting can be improved.

The present invention belongs to the communications and information technical field and relates to a method for processing received radio voice communication signals and information, in particular, a method for demodulating radio voice speech communication, and then performing data detection, segmentation and identification on the demodulated radio voice speech communication so as to identify periods of invalid signals such as noise and interference. According to the method, on the basis of the principles of statistical mathematics, statistical clustering analysis and fuzzy signal processing, a unified framework integrated processing method for identifying non-speech segments after receiving and demodulating non-cooperative radio voice communication, a detection method for finding sudden transient disturbance, a detection method for finding co-channel crosstalk disturbance from other radios, a speech signal and noise signal segment detection and identification method that does not depend on signal power strength and zero-crossing rate, a method for finding and identifying radio fault data, and a method for performing integrated identification on analog speech data and other digitally modulated data are provided.

The invention provides a voltage sag homologous identification method based on a Pearson correlation coefficient and OPTICS. The method comprises the following steps: S1, acquiring sag monitoring data recorded by a voltage sag monitoring device; S2, quantifying the sag monitoring data similarity based on a Pearson correlation coefficient; S3, obtaining homologous recognition features of the sag monitoring data, performing homologous clustering based on the homologous recognition features of the sag monitoring data through an OPTICS algorithm, and outputting a clustering result reachable graph, wherein the homologous recognition features comprise sag monitoring data similarity and sag duration; and S4, obtaining a clustering result cluster number based on a reachable graph sag number, and outputting a homologous recognition result. The problem that homologous identification features are too single is avoided, clusters of different densities can be found, the voltage sag homologous identification result is finally obtained, repeated redundant information can be effectively removed through the voltage sag homologous identification result, the real occurrence level of regional sag can be obtained, the value density of data is improved, and the intensity and difficulty of calculation and analysis are reduced.

The invention provides a method for evaluating a set of input data, the input data comprising at least one of: clinical data of a subject; genomic data of a subject; clinical data of a plurality of subjects; and genomic data of a plurality of subjects, using a deep learning algorithm. The method includes obtaining a set of input data, wherein the set of input data comprises raw data arranged into a plurality of data clusters and tuning the deep learning algorithm based on the plurality of data clusters. The deep learning algorithm comprises: an input layer; an output layer; and a plurality of hidden layers. The method further includes performing stabstical clustering on the raw data using the deep learning algorithm, thereby generating statistical clusters and obtaining a marker from each statistical cluster. Finally, the set of input data is evaluated based on the markers to derive data of medical relevance in respect of the subject or subjects.

The invention provides a malicious traffic detection method in a high-bandwidth scene based on frequency domain analysis. The method comprises the following steps: carrying out data packet granularity feature extraction on network traffic to obtain a data packet granularity feature; encoding the features of the data packet granularity to obtain matrix representation, performing fitting operation to obtain a plurality of frames, and performing frequency domain analysis on each frame to obtain a corresponding frequency domain feature; calculating the power of the frequency domain features to obtain power representation, performing logarithmic transformation to obtain frequency domain feature representation, cutting and averaging the frequency domain feature representation to serve as the input of a statistical clustering algorithm, and outputting a clustering center; and calculating the distance between the frequency domain feature representation and the corresponding nearest clustering center, if the distance is greater than a predetermined multiple of a training error, determining that the flow corresponding to the frequency domain feature representation is abnormal flow, otherwise, determining that the flow is normal flow. The method has the advantages of high detection throughput, high precision, low time delay and the like, and malicious traffic can be accurately detected in a high-bandwidth scene while calculation overhead and storage overhead are considered.

The invention relates to a university innovation entrepreneurial ability evaluation method based on gray clustering. The method includes defining a white function, adopting three kinds of statistic clusters of a high standard, a middle standard and a low standard to analyze the university innovation entrepreneurial ability and obtain numerical value classification of the high standard, the middle standard and the low standard of each index of each performance object.

The invention relates to the technical field of computer vision, in particular to a child detection frame filtering algorithm based on grid clustering, and the method comprises the following steps: obtaining video data in a real scene through a camera, transmitting the video data to a computer through the camera, obtaining a target image, storing the target image, selecting a target camera, and pulling detection frame data corresponding to one week from a production environment; according to the child detection frame filtering algorithm provided by the invention, a detected human-shaped frame can be directly used as statistical data, then the average frame height in the grid range is obtained by using a grid clustering method, whether a child is a child is judged by using the average frame height and the target frame height, and meanwhile, the average frame height of each region in a target image can be judged; misjudgment caused by shielding is avoided, and the problem that due to the fact that many shielding and false detection exist in a real scene, statistical clustering is directly carried out through a detection frame, the whole data distribution is disturbed by an abnormal detection frame, and consequently many misjudgments of a child frame are caused is solved.

The invention relates to the air-conditioning technology. The problem that a system or method for statically analyzing a wind velocity usage habit of a user is inexistent at present is solved, the invention provides a clustering method of the wind velocity usage habit of the air-conditioner user. The technical scheme can be generalized as follows: firstly defining a plurality of clustering centers according to the selection of the user to the wind velocity; acquiring behavior data of the air-conditioner user through a database, and then acquiring the behavior data of each air-conditioner user from the database through the system, computing the usage rate of each wind velocity by the user, and computing the distance between the usage rate of each wind velocity by the user to each clustering center, comparing the distances, selecting the corresponding clustering as the classification of the user according to a distance rule, finally counting the user number of each kind through the system after the clustering, and outputting the usage preference of the user to the wind velocity on the whole. The method disclosed by the invention has the beneficial effects that the method is convenient for the developing and designing of the air-conditioner, and is suitable for the air-conditioning system.

The invention discloses a power distribution network operation data abnormity determination method based on data mining, and the method comprises the steps: setting original power grid operation data D and the number m of outliers, and putting the standardized data into a K-means + + clustering model; after the result of the clustering model is obtained, counting the data number n (i) of each cluster after clustering, and judging whether the value of i is greater than the set number m of outliers; if n (i) is greater than or equal to m, calculating LOF outlier factors of all objects of the class by adopting an LOF algorithm; a final outlier candidate set is generated, outlier factors of all data points are calculated and sorted, and a line loss abnormal condition set is formed; and carrying out inductive reasoning on the operation data in the line loss abnormal condition set to obtain abnormal occurrence time, tracing the abnormal occurrence time to a power distribution network structure, and positioning an abnormal occurrence place. According to the method, the occurrence of the abnormality in the operation data of the power distribution network can be efficiently and accurately judged, and the abnormality occurrence time and place are determined.