326 results about "Real arithmetic" patented technology

A relevancy ranking and clustering method and system that determines the relevance of a document relative to a user's query using a similarity comparison process. Input queries are parsed into one or more query predicate structures using an ontological parser. The ontological parser parses a set of known documents to generate one or more document predicate structures. A comparison of each query predicate structure with each document predicate structure is performed to determine a matching degree, represented by a real number. A multilevel modifier strategy is implemented to assign different relevance values to the different parts of each predicate structure match to calculate the predicate structure's matching degree. The relevance of a document to a user's query is determined by calculating a similarity coefficient, based on the structures of each pair of query predicates and document predicates. Documents are autonomously clustered using a self-organizing neural network that provides a coordinate system that makes judgments in a non-subjective fashion.

The embodiment of the invention discloses a neural network processor and a convolutional neural network processor. The neural network processor can comprise a first weight preprocessor and a first operation array, wherein the first weight preprocessor is used for receiving a vector V<x> including M elements; the normalized value domain space of the element V(x-i) of the vector V<x> is a real number which is greater than or equal to 0 and is smaller than or equal to 1; an M*P weight vector matrix Q<x> is used for performing weighting operation on the M elements of the vector V<x> to obtain M weighting operation result vectors; the first operation array is used for accumulating the elements with the same positions in the M weighting operation result vectors to obtain P accumulated values; a vector V<y> including P elements is obtained according to the P accumulated values; and the vector V<y> is output. The technical scheme provided by the embodiment has the advantage that the expansion of the application range of neural network operation is facilitated.

Methods and apparatus for calculating Single-Instruction-Multiple-Data (SIMD) complex arithmetic. A coprocessor instruction has a format identifying a multiply and subtract instruction to generate real components for complex multiplication of first operand complex data and corresponding second operand complex data, a cross multiply and add instruction to generate imaginary components for complex multiplication of the first operand complex data and the corresponding second operand complex data, an add-subtract instruction to add real components of the first operand to imaginary components of the second operand and to subtract real components of the second operand from imaginary components of the first operand, and a subtract-add instruction to subtract the real components of the second operand from the imaginary components of the first operand and to add the real components of the first operand to the imaginary components of the second operand.

The invention provides a short text sentiment tendency analysis method combining BiLSTM with an Attention mechanism. A model combining a convolutional bidirectional long-short time memory network withan attention mechanism is constructed; for the problems that high dimensions represented by data in text classification are difficult to train and vector representation features are irrelevant, the text data are mapped to a low-dimension real number vector, and the problem that a model is subjected to dimension disasters due to high-dimension input is avoided; and for the feature selection problem, constructing a feature extraction model, and carrying out text sentiment classification on the extracted features.

A system and method for training a statistical machine translation model and decoding or translating using the same is disclosed. A source word versus target word co-occurrence matrix is created to define word pairs. Dimensionality of the matrix may be reduced. Word pairs are mapped as vectors into continuous space where the word pairs are vectors of continuous real numbers and not discrete entities in the continuous space. A machine translation parametric model is trained using an acoustic model training method based on word pair vectors in the continuous space.

The present invention describes a method and apparatus for performing logarithmic arithmetic with real and/or complex numbers represented in a logarithmic format. In one exemplary embodiment, an ALU implements logarithmic arithmetic on complex numbers represented in a logpolar format. According to this embodiment, memory in the ALU stores a look-up table used to determine logarithms of complex numbers, while a processor in the ALU generates an output logarithm based on complex input operands represented in logpolar format using the stored look-up table. In another exemplary embodiment, the ALU performs logarithmic arithmetic on real and complex numbers represented in logarithmic format. In this embodiment, the memory stores two look-up tables, one for determining logarithms of real numbers and one for determining logarithms of complex numbers, while the processor generates an output logarithm based on real or complex input operands represented in logarithmic format using the real or complex look-up tables, respectively.

A generalized framework is disclosed in which a wide variety of propagation models can be cast in a matrix-based format using arbitrary matrix coefficients (e.g., real numbers, integers, etc.). Casting propagation models in the matrix-based framework enables efficient computer implementation and calculation, ease of tuning, admissibility (i.e., the tuned parameters of a linear matrix model are guaranteed to be the global optimum), and aggregating multiple propagation models into a single matrix-based model. Matrix-based propagation models based on transmitter-receiver azimuth orientation, transmitter antenna height, terrain elevation, and clutter are also disclosed. The propagation models can be used in conjunction with automated data acquisition from information sources such as topographic maps, clutter maps, etc.

A system for digital signal processing, configured as a system on chip (SoC), combines a microprocessor core and digital signal processor (DSP) core with floating-point data processing capability. The DSP core can perform operations on floating-point data in a complex domain and is capable of producing real and imaginary arithmetic results simultaneously. This capability allows a single-cycle execution of, for example, FFT butterflies, complex domain simultaneous addition and subtraction, complex multiply accumulate (MULACC), and real domain dual multiply-accumulators (MACs). The SoC may be programmed entirely from a microprocessor programming interface, using calls from a DSP library to execute DSP functions. The cores may also be programmed separately. Capability for programming and simulating the entire SoC are provided by a separate programming environment. The SoC may have heterogeneous processing cores in which either processing core may act as master or slave, or both cores may operate simultaneously and independently.

An improved method and system for performing dynamic online multi-parameter optimization for autonomic computing systems are provided. With the method and system of the present invention, a simplex, i.e. a set of points in the parameter space that has been directly sampled, is maintained. The system's performance with regard to a particular utility value is measured for the particular setting of configuration parameters associated with each point in the simplex. A new sample point is determined using the geometric transformations of the simplex. The method and system provide mechanisms for limiting the size of the simplex that is generated through these geometric transformations so that the present invention may be implemented in noisy environments in which the same configuration settings may lead to different results with regard to the utility value. In addition, mechanisms are provided for resampling a current best point in the simplex to determine if the environment has changed. If a sufficiently different utility value is obtained from a previously sampled utility value for the point in the simplex, then rather than contracting, the simplex is expanded. If the difference between utility values is not sufficient enough, then contraction of the simplex is performed. In addition, in order to allow for both real and integer valued parameters in the simplex, a mechanism is provided by which invalid valued parameters that are generated by geometric transformations being performed on the simplex are mapped to a nearest valid value. Similarly, parameter values that violate constraints are mapped to values that satisfy constraints taking care that the dimensionality of the simplex is not reduced.

A method, system, and computer program product that convert a real number from a floating point representation to a character string. Mantissa bits are extracted from the floating-point representation of a value into an integer format. The mantissa bits of the integer format may be shifted left by a number of positions that is equal to a base 2 exponent of a first representation of the value. Prior to converting the value, the floating-point representation of the value can be divided by 2ⁿ, where n is the number of the mantissa bits in the floating-point representation of the value. A set of operations can then be performed to convert a result of the division to a number greater than or equal to 1.0 and less than 10.0. The set of operations may include multiplication operations and/or retrieving a constant from a lookup table.

The invention discloses a Chinese hedge scope detection method based on a stacked neural network. The Chinese hedge scope detection method is characterized by comprising the following steps: carrying out word segmentation processing on sentences which contain hedges in a to-be-analyzed experimental corpus; carrying out syntactic parsing on the sentences after the word segmentation processing by employing a syntactic parser to obtain a phrase structure tree of the sentences; finding candidate phrases via a phrase-based candidate sample screening strategy, thereby determining boundary words of the candidate phrases, including left boundary words and right boundary words; respectively filtering the left and right boundary words as well as context information of the hedges by employing filtering windows; taking the left and right boundary words as well as the context information of the hedges as candidate sample word sequences and mapping to a real number vector space to convert into a word vector form; inputting a stacked learning model LSTM (Long Short Term Memory networks)-CNN (Convolutional Neural Network) based on a combination of the LSTM and the CNN for learning to obtain boundary classifiers; and carrying out classification on test data to obtain classification results of left and right boundaries.

The present invention provides an active data type for use in a computer program. The active data type has an identifier and at least one algorithm associated therewith. The identifier is utilized by the computer program to identify the instance of the active data type. The algorithm is configured to be automatically executed when an attempt to access a value associated with the active data type instance is made by a routine or otherwise. When a particular routine that uses an instance of the active data type attempts to access the value associated with the active data type, the algorithm determines the value associated with the active data type before the routine obtains access to the value. The active data type may be a real, an integer, or a string, for example. The algorithm automatically determines the current value associated with the active data type instance. Preferably, the active data type has an identifier, a first algorithm and a second algorithm associated therewith. The first algorithm preferably automatically determines the current value of the instance of the active data type when a routine that utilizes the value of the active data type instance attempts to access the value. When the value of the instance of the active data type is set, the second algorithm preferably automatically post-processes the value to which the active data type instance has been set. A locking/unlocking mechanism sets the value of the active data type instance prior to the first algorithm invoking the particular routine, suspends active data type algorithm processing while the routine executes, and processes the value of the active data type instance using the second algorithm once the routine has returned in order to post-process any changes to the value of the active data type instance.

The multi-mode Multiplier-And-Accumulator of the present invention is used with the double-precision Complex-Valued Multiplier-And Accumulator as a main configuration, and the different precisions and digital modes make it more flexible, compared to the traditional real number Multiplier-And-Accumulator. In addition, it does not have a data alignment problem which occurs in the traditional application of different precision Subword Parallel processors. This kind of Multiplier-And-Accumulator takes a double-precision Complex-Valued Multiplier-And- Accumulator as the main configuration, with four double-precision real-valued multipliers and several groups of accumulators to assist in different modes ofoperation. Each double-precision real- valued multiplier can be segmented into four single-precision multipliers, and then double-precision multiplier products are obtained by means of displacement addition. If two real numbers which are continuous in time sequence are taken as the real number input and imaginary number input for the original complex-valued multipliers, the accumulated products include not only the present output accumulated product summation but also the output accumulated product summation of the previous time and the next time.

A system and method for solving a decision problem having Boolean combinations of linear and non-linear operations includes translating the non-linear real operations using a COordinate Rotation DIgital Computer (CORDIC) method programmed on a computer device into linear operations maintaining a given accuracy. Linear and translated linear operations are combined into a formula. Satisfiablity of the formula is solved using a decision procedure for Boolean combinations of linear operations over integers and reals.

A first method to compute a phase map within an optical proximity correction simulation kernel utilizes simulated wavefront information from randomly generated data. A second method uses measured data from optical tools. A phase map is created by analytically embedding a randomly generated two-dimensional array of complex numbers of wavefront information, and performing an inverse Fourier Transform on the resultant array. A filtering function requires the amplitude of each element of the array to be multiplied by a Gaussian function. A power law is then applied to the array. The elements of the array are shuffled, and converted from the phasor form to real/imaginary form. A two-dimensional Fast Fourier Transform is applied. The array is then unshuffled, and converted back to phasor form.

The invention discloses a verifiable fully homomorphic encryption method based on matrix operation. The method comprises the following steps: preprocessing and converting plaintext data into vectors to obtain a vector set; Then selecting a non-zero random real number and each vector to construct a triangular matrix, and encrypting each triangular matrix to obtain an encrypted matrix set; Generating verification evidences by utilizing the triangular matrix set according to the calculation function type; Performing matrix operation on each encryption matrix according to a certain rule to obtainan encrypted operation result; And finally, decrypting the encrypted operation result to obtain a plaintext of the result, and verifying the correctness of the result by comparing whether the verification evidence is equal to the value of the calculation result. The method has privacy security and result verifiability; Simulation experiment results show that the method has high efficiency in the stages of secret key generation, decryption and verifiability, and the effectiveness and feasibility of the scheme are shown.

A method and apparatus for complex multiplication includes steps of: (a) receiving a complex multiplicand having a real value and an imaginary value (704); (b) generating a negation of the real value of the complex multiplicand (706); (c) generating a negation of the imaginary value of the complex multiplicand (708); (d) receiving a complex multiplier (710); and (e) selecting a phasor constant having a value wherein a complex product of the complex multiplicand times the complex multiplier times the phasor constant has a real value equal to one of the real value of the complex multiplicand, the imaginary value of the complex multiplicand, the negation of the real value of the complex multiplicand, and the negation of the imaginary value of the complex multiplicand (712).

The invention discloses a large-scale image data similarity searching method based on an EMD (earth mover's distance). The method comprises the following steps that an image data mapping function f used for mapping to a one-dimension real number key value space Omega(phi) is designed; an operation MR1 is started, and a load of each key value in the Omega(phi) is estimated; the operation MR2 is started, the cutting is carried out on the Omega(phi) through a Map task on the basis of the estimated key value load, and data corresponding to the cutting region are sent to a Reduce task in a segmented way; image data received by each Reduce task is mapped to the key values in the Omega(phi) on the basis of f, and an index structure oriented to the EMD is built on the basis of the key values; the similarity searching based on the EMD is executed on the basis of the index structure; execution results of each Reduce task based on EMD similarity searching in the MR2 are subjected to union set taking and output. The large-scale image data similarity searching method has the advantages that the network transmission data quantity is lower, the calculation load distribution is more balanced, the similarity searching efficiency is higher, and the big data set analysis and processing expandability is better.

Some embodiments provide an LP method that identifies route propagations. In some embodiments, this method is used by a router that hierarchically defines routes for nets within a region of a design layout. The router (1) partitions the region into a first set of sub-regions, and (2) for each particular net, identifies a route that traverses a set of the first-set sub-regions. In some embodiments, the invention's method partitions the first set of sub-regions into a second set of smaller sub-regions. It then identifies a plurality of propagation possibilities for propagating each route into the second set of smaller sub-regions of the first set sub-regions. The method next formulates a linear-programming (“LP”) problem based on the identified propagation possibilities. The method then solves the LP problem. In some embodiments, the formulated LP problem is an integer-linear-programming (“ILP”) problem, and solving the ILP problem returns integer solutions that specify one propagation permutation for each route in each first-set sub-region traversed by the route. In other embodiments, solving the LP problem returns real-numbered solutions. In some of these embodiments, the method converts the real-number solutions into integer solutions that specify one identified propagation permutation for each route in each first-set sub-region traversed by the route.