975 results about "Sparse matrix" patented technology

A deterministic finite automaton uses binary search (and optionally hashing) method(s) of sparse matrix representation to recognize the graphical representations of characters or icons from a bitmap representation of the computer screen. This recognition can be applied to translate data from an unknown form of original specification (file format) into a known form of representation, such as HTML. Alternatively this recognition can be applied to another process that can “see” what is on the screen, and perform programmed actions, based on what it “sees”.

A method and apparatus for performing video compression and decompression using dynamic quantization and / or encoding. According to one aspect of the compression method, an image is recursively filtered into its constituent components each represented by a matrix of coefficients. Next, the level of correlation of a first of the matrices of coefficients is determined. Based on the level of correlation of this first coefficient matrix, one of a first quantization technique and a second quantization technique is selected. The first coefficient matrix is then quantized using the selected quantization technique.According to another aspect of the compression method, an image is digitally filtered into its constituent components, each represented by a matrix of coefficients. At least certain matrices are quantized to generated a first quantized matrix. The first quantized matrix is then recursively divided until the first quantized matrix or each of the resulting submatrices is sufficiently dense or sufficiently sparse. Each of the sufficiently dense matrices are encoded using a first technique, while each of the sufficiently sparse matrices are encoded a second technique.

A data structure, database management system, and methods of validating data are disclosed. A data structure is described that includes a superset of interconnected relational databases containing multiple tables having a common data structure. The tables may be stored as a sparse matrix linked list. A method is disclosed for ordering records in hierarchical order, in a series of levels from general to specific. An example use with address databases is described, including a method for converting an input address having a subject representation into an output address having a preferred representation. Preferred artifacts may be marked with a token. Alias tables may be included. This Abstract is provided to comply with the rules, which require an abstract to quickly inform a searcher or other reader about the subject matter of the application. This Abstract is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the claims.

An efficient method for solving a model predictive control problem is described. A large sparse matrix equation is formed based upon the model predictive control problem. The square root of H, Hr, is then formed directly, without first forming H. A square root (LSMroot) of a large sparse matrix of the large sparse matrix equation is then formed using Hr in each of a plurality of iterations of a quadratic programming solver, without first forming the large sparse matrix and without recalculating Hr in each of the plurality of iterations. The solution of the large sparse matrix equation is completed based upon LSMroot.

The invention relates to a graph-based semi-supervised high-spectral remote sensing image classification method. The method comprises the following steps: extracting the features of an input image; randomly sampling M points from an unlabeled sample, constructing a set S with L marked points, constructing a set R with the rest of the points; calculating K adjacent points of the points in the sets S and R in the set S by use of a class probability distance; constructing two sparse matrixes WSS and WSR by a linear representation method; using label propagation to obtain a label function F<*><S>, and calculating the label prediction function F<*><R> of the sample points in the set R to determine the labels of all the pixel points of the input image. According to the method, the adjacent points of the sample points can be calculated by use of the class probability distance, and the accurate classification of high-spectral images can be achieved by utilizing semi-supervised conduction, thus the calculation complexity is greatly reduced; in addition, the problem that the graph-based semi-supervised learning algorithm can not be used for large-scale data processing is solved, and the calculation efficiency can be improved by at least 20-50 times within the per unit time when the method provided by the invention is used, and the visual effects of the classified result graphs are good.

The invention discloses a high-resolution dictionary based sparse representation image super-resolution reconstruction method. The method comprises the following steps of: (1) constructing a high-resolution brightness image library; (2) generating a sample training set; (3) learning an over-complete dictionary; (4) primarily establishing a high-resolution image brightness space; (5) establishing an image sample test set; (6) updating the high-resolution image brightness space; (7) calculating a weight sparse matrix; (8) reupdating the high-resolution image brightness space; (9) judging whether to repeat execution; and (10) outputting a high-resolution image. The high-resolution over-complete dictionary learned by the invention can be applied to different amplification factors. Sparse representation, non-local prior and data fidelity constraint are fully utilized, so that local information and global information can be comprehensively utilized. The method has higher super-resolution capacity; and the reconstructed image is closer to an actual image.

A method for data encoding and associated decoding is based on the concept of batches that allows transmission of a large data file from a source node to multiple destination nodes through communication networks that may employ network coding wherein sparse matrix codes are employed in a network setting. A batch is a set of packets generated by a subset of the input packets using sparse matrix encoder. A sparse matrix encoder can be called repeatedly to generate multiple batches. The batches are generally independent of one another. During the transmission in a communication network, network coding can be applied to packets belonging to the same batch to improve the multicast throughput. A decoder recovers all or at least a fixed fraction of the input packets using received batches. The input packets can be pre-coded using a pre-code before applying sparse matrix codes. The data file can then be reconstructed by further decoding the pre-code.

A data structure, database management system, and methods of validating data are disclosed. A data structure is described that includes a superset of interconnected relational databases containing multiple tables having a common data structure. The tables may be stored as a sparse matrix linked list. A method is disclosed for ordering records in hierarchical order, in a series of levels from general to specific. An example use with address databases is described, including a method for converting an input address having a subject representation into an output address having a preferred representation. Preferred artifacts may be marked with a token. Alias tables may be included. This Abstract is provided to comply with the rules, which require an abstract to quickly inform a searcher or other reader about the subject matter of the application. This Abstract is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the claims.

A joint code-encoder-decoder design approach and circuit architecture design for (3,k)-regular LDPC coding system implementation. The joint design process relies on a high girth (2,k)-regular LDPC code construction. The decoder realizes partly parallel decoding. The encoding scheme only contains multiplications between sparse matrices and vector and multiplication between a very small dense matrix and vector.

A system and method for visualizing data. Data are provided either in the form of data values of a data array or in the form of a geometric representation. A data array may be, for example, a sparse matrix. A geometric representation, may be, for example, an integrated circuit layout coded in a geometric description language. Data provided in the form of data values are associated with geometric shapes placed on a grid. Information placed on the grid is then reported to a user. If data are provided in the form of a geometric representation, then data values are extracted from the geometric representation. A graphic representation is generated from the extracted data values. The graphic representation is exhibited to a user.

The invention discloses a multiattribute collaborative filtering recommendation method oriented to a social network. The multiattribute collaborative filtering recommendation method includes utilizing mass data information of the social network to collect user, friend and item list information, and establishing an original user-item scoring matrix; utilizing a thought of acquiring a middle average value from nine numbers, and performing prediction filling on a sparse matrix; calculating inter-user attracting similarity through a user-item bipartite graph; calculating interaction similarity, linearly combining the attracting similarity with the interaction similarity to acquire comprehensive similarity among users, and searching to acquire a nearest neighbor set of a target user; performing prediction scoring on items to be recommended by the target user according to the nearest neighbor set of the target user, and generating a Top-N recommendation set. By the method, calculating rules of inter-user similarity in a conventional collaborative filtering method are improved, huge impedance brought to the filtering recommendation method and a recommendation system by sparseness of a scoring matrix is reduced, and accuracy of the recommendation system is improved.

A method for solving a matrix equation AX=B, wherein A represents a block sparse matrix, B represents a right hand side block vector and X represents a solution block vector. In one embodiment, the method includes receiving the block sparse matrix and the right hand side block vector, constructing a reduced transformed block sparse matrix from the block sparse matrix, constructing a reduced transformed residual block vector from the block sparse matrix and the right hand side block vector, and solving for the solution block vector using the reduced transformed block sparse matrix and the reduced transformed residual block vector.

The invention provides an apparatus and method for realizing sparse neural network. The apparatus herein includes an input receiving unit, a sparse matrix reading unit, an M*N number of computing units, a control unit and an output buffer unit. The invention also provides a parallel computing method for the apparatus. The parallel computing method can be used for sharing in the dimension of input vector and for sharing in the dimension of weight matrix of the sparse neural network. According to the invention, the apparatus and the method substantially reduce memory access, reduce the number of on-chip buffering, effectively balance the relationship among on-chip buffering, I / O access and computing, and increase the performances of computing modules.

A Multiple Input Multiple Output (MIMO) radar system and method of using it for target recovery are disclosed. The MIMO radar system comprises an array of distributed radiating elements configured to transmit signals towards a target scene, an array of distributed receiving elements configured to receive signals backscattered from the target scene, a sampling module configured to sample the signals received, and a hardware processor configured to recover from the samples position parameters of one or more targets. Range, direction and optionally velocity, are estimated via simultaneous 2D or 3D sparse matrix recovery, wherein all channels defined by transmitter-receiver pairs are processed together. The digital processing may be applied either in Nyquist or sub-Nyquist scheme, reducing the number of samples, transmit and / or receive antennas. The radar system is optionally further enhanced by cognitive transmission scheme where transmitted signals are distributed over a wide frequency range with vacancy bands left therein.

The invention discloses a method for increasing the computing speed through parallel computing based on an MPI and OpenMP hybrid programming model. The method includes the steps that the callable MPI process number and OpenMP thread number are determined according to the computing node number and the available CPU core number in nodes; an existing sub sparse matrix A, the sub initial vector x0, the block vector b and the maximum computing tolerance Tolerance are read into each process; a multi-thread compiling command is enabled for each process; cycle computing of a precondition conjugate gradient method is conducted on all the processes; if the computed error is smaller than the permissible value, cycle computing is ended, and otherwise, cycle computing is continuously conducted; computing results of all the processes are reduced, and a solution of a problem is output; when parallel computing is conducted, MPI processes are started, multi-thread resolving is conducted on the problem, parallel computing among the nodes is started, all the MPI processes are distributed to one computing node, and information is exchanged through message transmission among the processes; then in all the MPT processes, an OpenMP guidance command is used to create a set of threads, and the threads are distributed to different processors of the computing node to be executed.

The present invention provides a system and method for building fast and efficient support vector classifiers for large data classification problems which is useful for classifying pages from the World Wide Web and other problems with sparse matrices and large numbers of documents. The method takes advantage of the least squares nature of such problems, employs exact line search in its iterative process and makes use of a conjugate gradient method appropriate to the problem. In one embodiment a support vector classifier useful for classifying a plurality of documents, including textual documents, is built by selecting a plurality of training documents, each training document having suitable numeric attributes which are associated with a training document vector, then initializing a classifier weight vector and a classifier intercept for a classifier boundary, the classifier boundary separating at least two document classes, then determining which training document vectors are suitable support vectors, and then re-computing the classifier weight vector and the classifier intercept for the classifier boundary using the suitable support vectors together with an iteratively reindexed least squares method and a conjugate gradient method with a stopping criterion.

The invention discloses a fast decoupled flow calculation method for power systems, which comprises the following steps of: inputting original data and initializing voltage; forming an admittance matrix; forming correction equation coefficient matrixes B' and B'' and performing factor table decomposition; performing P-theta iteration, and correcting a voltage phase angle; performing Q-V iteration, and correcting voltage amplitude; judging whether the iteration is converged; and calculating node power and branch power. The method requires that the P-theta iteration and the Q-V iteration are all converged in the same iteration and the iteration process is finished, so that the algorithm frame is simpler, and the flow is clearer. The sparse matrix technology is not adopted, so the matrix elements are convenient to access and calculate, and the programming is simple; the correction equation coefficient matrixes are stored according to n order, number change of nodes is avoided, and the programming difficulty is reduced; and the calculation amount is reduced through reasonable logic judgment, the calculation speed is obviously improved and the requirement of scientific research can be completely met. The fast decoupled flow calculation method also can process power systems with a plurality of balance nodes.

The invention belongs to the technical field of remote sensing image processing, and specifically relates to a low-rank expression and learning dictionary-based hyperspectral image abnormity detection algorithm. According to the algorithm, a method for introducing low-rank expression in the abnormity detection problems is used for decomposing the two-dimensional hyperspectral image data into the sum of a low-rank matrix expressing background and a sparse matrix expressing abnormity, and then enabling a basic abnormity detection algorithm to act on the sparse matrix to obtain the abnormity detection result; and furthermore, the concept of a learning dictionary is imported in the low-rank expression algorithm, and the learning dictionary is obtained through an algorithm of random selection and gradient descent and is capable of expressing the background spectrums in hyperspectral images. Through the importing of the learning dictionary, the abnormity information can be better separated from the hyperspectral image data, so that better detection result can be obtained; and meanwhile, the robustness of the algorithm for the initial parameters can be improved, so that the computing cost is reduced and important value is provided for the actual abnormity detection application.

The invention provides a distributed source center direction-of-arrival estimation method based on Bayesian compressed perception, and belongs to the technical field of wireless mobile communication. The invention mainly aims to solve the problem concerning inherent error of center direction-of-arrival estimation when the center angle of arrival of an information source is not on an angle sampling grid. According to the invention, an antenna array composed of parallel uniform linear arrays is arranged; an approximate array data reception model of a distribution source is established; the space-domain angle is sampled; a parameterized over-complete redundant dictionary is constructed by using an array steering vector so as to make the problem of distributed source center direction-of-arrival estimation converted into the problem of sparse matrix equation solving; a Bayesian compressed perception method is adopted to solve the equation set and obtain the most sparse solution of an unknown sparse vector; and the estimated value of the center direction of arrival is obtained according to the one-to-one correspondence relationship between sparse solutions and space-domain angles. The method of the invention is low in computing complexity, and has the characteristics of high resolution and accuracy under the condition of a small number of snapshots.

A performance or reliability model representing the behaviour of an application under different system resource conditions is provided. This model may take the form of one or more sparse matrix providing a reliability or performance values for different combinations of conditions. This model is distributed to a user of the application, and is consulted during execution of the application with reference to system resource information provided by the operating system or other monitoring software so as to provide an indication of the expected performance of the application under present operating conditions. This indication may be notified to a user, for example in a case where the indication falls outside predetermined bounds of satisfactory operation. The system may also attempt to renegotiate attributed system resources to as to improve performance.

A computer program product and method for multiplying a sparse matrix by a vector are disclosed. The computer program product includes a computer readable medium for storing instructions, which, when executed by a computer, cause the computer to efficiently multiply a sparse matrix by a vector, and produce a resulting vector. The computer is made to create a first array containing the non-zero elements of the sparse matrix, and a second array containing the end_of_row position of the last non-zero element in each row of the sparse matrix. A variable is initialized, and then, for each row of the second array, the computer is made to do one of two things. Either, it equates the variable to the sum of the variable and the product of a particular element of the first array and a particular element of the vector. Or, it equates a particular element of the resulting vector to the variable, and then equates the variable to a particular value.

The invention discloses a high-efficiency time domain electromagnetic simulation method based on an H matrix algorithm, which can realize electromagnetic simulation on a large three-dimensional target. In the method, a time domain finite element method (TDFEM) is used as a background, a low-rank compression technique is used as a core, and a tree structure is used as a basis for carrying out logical unit (LU) decomposition on a sparse matrix generated by the TDFEM by a four arithmetic algorithm corresponding to the H matrix. The acquired upper and lower triangular factors have low-rank compressible characteristics, and the compressed matrix equation can realize quick solution of high-efficiency time domain electromagnetic simulation by the H matrix algorithm. The high-efficiency time domain electromagnetic simulation method has the advantages of fast computation speed, low memory consumption, controllable computation accuracy, good stability and the like, can reduce the complexity of computation to O(Nlog<2>N) and reduce the memory consumption to O(NlogN), can be widely applied to the solution of a large sparse linear system of equations during high-efficiency time domain electromagnetic simulation, and can provide important reference for analyzing the electromagnetic property of the large three-dimensional target.

The invention provides an image encryption method based on a two-dimensional compression perception and chaotic system. A wavelet sparse basis is used to carry out sparse representation of an image to generate a sparse matrix, then a random sequence generated by a two-dimensional sine-logistic chaotic system iteration is used to generate a measurement matrix, the measurement matrix is used to carry out compression measurement of the sparse matrix, and a preliminary compression encrypted image which is an intermediate result is obtained. In order to enhance the attack resistance ability of the system, the chaotic system is used again to generate a chaotic series to carry out pixel scrambling and cyclic shift operations on the intermediate result, and a final encryption map is obtained. According to the method, the two-dimensional sine-logistic chaotic system is used, the sequence randomness is increased, and the scrambling effect of the image is enhanced. The key space is expanded, the safety of an encryption system is enhanced, an excessive ciphertext data quantity is avoided, and a good encryption effect is achieved.

Systems, methods, and apparatuses are disclosed for identifying anomalies or changes in road conditions on a roadway location. An initial low rank data matrix of initial vehicle probe data at a plurality of different times for a roadway location is provided, where the initial low rank data matrix represents a baseline of road conditions for the roadway location. A plurality of additional vehicle probe data from at least one vehicle at the roadway location is received. The additional vehicle probe data is added to the initial vehicle probe data of the initial low rank data matrix. The updated data matrix with the compiled probe data is decomposed into a low rank data matrix and a sparse data matrix. A change at the roadway location is identified based on the probe data in the sparse data matrix.

The invention discloses a vibration modeling and analyzing method of a crack impeller structure of a centrifugal compressor. The vibration modeling and analyzing method comprises the steps of firstly, providing a modeling method of a centrifugal impeller, obtaining a whole model by rotating and converting a finite element model of a section of the impeller, reducing the DOF (Degree of Freedom) of a system by adopting a hybrid interface modal synthesis method, and simulating a breath effect of a crack by defining a contact mode on a crack interface; secondly, providing a method for solving large-scale symmetrical sparse matrix-inverse matrix, and thus obviously reducing the memory space needed by calculation; finally, providing a method for carrying out statistical analysis on resonant frequency of the crack impeller structure, considering dissonant factors such as manufacturing error and state degradation which exist in an actual impeller, and obtaining statistical regularity of structure resonant frequency under a random dissonant mode. According to the vibration modeling and analyzing method disclosed by the invention, the calculating accuracy is ensured, obvious improvement effect of calculating efficiency is realized at the same time, and a high-efficiency analyzing method is provided for optimal design of the impeller and quantitative diagnosis of the crack.

Embodiments of the present invention relate to systems and methods for optimizing and reducing the memory requirements of state machine algorithms in pattern matching applications. Memory requirements of an Aho-Corasick algorithm are reduced in an intrusion detection system by representing the state table as three separate data structures. Memory requirements of an Aho-Corasick algorithm are are also reduced by applying a banded-row sparse matrix technique to the state transition table of the state table. The pattern matching performance of the intrusion detection system is improved by performing a case insensitive search, where the characters of the test sequence are converted to uppercase as the characters are read. Testing reveals that state transition table with sixteen bit elements outperform state transition table with thirty-two bit elements and do not reduce the functionality of intrusion detection systems using the Aho-Corasick algorithm.

The present invention discloses a coder of LDPC code used in hierarchical quasi-cyclic expansion structure, comprising an input cache, a first depositing and caching pipeline grade, a second depositing pipeline grade, a third caching pipeline grade, a fourth depositing and caching pipeline grade and an output grade. By utilizing the character that the check matrix H is formed by connecting the quasi-cyclic sift matrix, the present invention simplifies the pipeline structure of the RU coding method, reduces the grades of the pipeline from sixth grade to fourth grade, and shortens the coding delay. Besides, the present invention decreases the largest pipeline delay and increases the coding thuoughput according to the fulfilling character of the main functional module. And the present invention also reduces the energy consumption of the coder ROM according to the operational character of the quasi-cyclic sift matrix; replaces the sparse matrixmultiply vector in the RU method with the quasi-cyclic sift unit matrix multiply vector; and replaces the non sparse matrix multiply vector in the RU method with the quasi-cyclic sift matrix multiply vector. A larger storing space can be remained in the ping pong RAM amid the grades to fit the demand of the variable length code and the VBR.

The invention discloses an interactive grayscale image colorizing method based on local linear model optimization. The method includes: inputting a grayscale image to be processed, converting the grayscale image to be processed into a red, green and blue (RGB) color space input image, conducting a little manual line coloring to obtain a colorized image, converting the grayscale image to be processed and the colorized image to a YUV color space from an original RGB color space respectively, calculating a laplace sectional drawing matrix, optimizing and solving a sparse matrix equation by using a local linear model, obtaining the colorized image based on the YUV color space, and finally converting the colorized image based on the YUV color space into the RGB color space to obtain a final colorized image. The method improves an existing image colorizing method based on local color expansion, reduces severe color permeation problems occurring in a grayscale image colorization process under the condition of less manual line coloring, and improves grayscale image colorization quality.

A new data structure and algorithms which offer at least equal performance in common sparse matrix tasks, and improved performance in many. This is applied to a word-document index to produce fast build and query times for document retrieval.