43 results about "Element by element" patented technology

A system and method for providing users with granular control over arbitrary information that allows for selective, real-time information sharing in a communications network such as the Internet is provided. In a network including a plurality of network devices operated by a plurality of users, a real-time information exchange system for sharing user profile information between respective users includes a database management system connected to the network. The database management system, which may be distributed across the network, stores the user profile information for a plurality of registered users of the information exchange system The user profile information includes a plurality of data elements, each data element having an associated one of the plurality of registered users. Each data element has an associated group of users to whom access to the data element has been granted, and users not included in the associated group of users are denied access to the data element. Each registered user may selectively control the granting and denying of access to each of its associated data elements by other respective user, on an element-by-element, and user-by-user basis. Further, each registered user may dynamically create its own data fields.

A method and apparatus for estimating interference and noise power in an orthogonal frequency division multiplexing/orthogonal frequency division multiple access/discrete multi-tone (OFDM/OFDMA/DMT) system is disclosed. A correlator correlates a plurality of sub-carriers with a preset reference sequence on an element-by-element basis and outputs a result of the correlation. A signal noise producer calculates a difference between a correlation value associated with each of the plurality of sub-carriers output from the correlator and a correlation value produced from at least one adjacent sub-carrier and outputs a result of the calculation. An interference and noise power producer produce interference and noise power from the difference between the correlation values calculated by the signal noise producer.

The invention discloses a multi-speaker speech separation method and system. The method comprises the following steps : acquiring mixed voice signals, obtaining multichannel multi-speaker mixed voicesignals, and scanning the multi-channel multi-speaker mixed voice signals to obtain MUSIC energy spectra; obtaining S peak values from the MUSIC energy spectra, wherein each peak value corresponds toa beam direction; respectively enhancing S beams to obtain mixed voice in S directions; performing short-time Fourier transform on the mixed voice corresponding to each direction to obtain short-timeFourier amplitude spectra of the S target speaker voice, and respectively inputting the short-time Fourier amplitude spectra into a depth neural network to estimate a phase sensing mask correspondingto each target speaker; and performing element-by-element multiplication between the phase sensing mask of each target speaker and amplitude spectra of the corresponding mixed speech to obtain the amplitude spectrum of the target speaker, and the time domain signal of the target speaker is recovered by inverse short-time Fourier transform using the phase spectrum of the corresponding mixed speech.

The invention provides an FPGA accelerator of an LSTM neural network and an acceleration method of the FPGA accelerator. The accelerator comprises a data distribution unit, an operation unit, a control unit and a storage unit; the operation unit comprises a sparse matrix vector multiplication module, a nonlinear activation function module and an element-by-element multiplication and addition calculation module; the control unit sends a control signal to the data distribution unit, and the data distribution unit reads an input excitation value and a neural network weight parameter from the storage unit and inputs the input excitation value and the neural network weight parameter to the operation unit for operation. The operation resources are uniformly distributed to each operation unit according to the number of the non-zero weight values, so that idling of operation resources is avoided, and the operation performance of the whole network is improved. Meanwhile, the pruned neural network is stored in the form of the sparse network, the weight value of each column is stored in the same address space, the neural network is coded according to the row index, and the operation performance and the data throughput rate are improved under the condition that the precision is guaranteed.

An apparatus and method for encrypting/decrypting information, and the encrypting/decrypting system using the same, and a computer readable recording medium storing programs for realizing the above are disclosed. A method for encrypting information on a basic element-by-basic element basis, includes the steps of: a) receiving information to be encrypted; b) generating random number maps; c) decomposing the information into a plurality of basic elements which are classified into multiple channels; d) encrypting the basic elements and generating encrypted information; and e) outputting the encrypted information.

A method of estimating a steering vector of a sensor array of M sensors according to one embodiment of the present disclosure includes estimating a steering vector of a noise source located at an angle 0 degrees from a look direction of the array using a least squares estimate of the gains of the sensors in the array, defining a steering vector of a desired sound source in the look direction of the array, and estimating the steering vector by performing element-by-element multiplication of the estimated noise vector and the complex conjugate of steering vector of the desired sound source. The sensors may be microphones.

An apparatus for and method of accommodating an XML document within a data base management system. The XML document is parsed on an element-by-element basis into an acceptable format for use by a data base management system by information contained in an XML mapping tree. Each element may be accompanied by appropriate attributes. The tree may be saved for future use. The XML document may be defined by a Document Type Definition (DTD).

One embodiment of the invention provides a receiver for use in a navigation system comprising multiple transmitters. Each transmitter transmits a positioning signal comprising a pseudo-random noise (PRN) sequence corresponding to the respective transmitter. The receiver comprises a code module for supplying multiple PRN sequences corresponding to the respective multiple transmitters; and a correlator for correlating the PRN sequences supplied by the code module with an incoming signal. The multiple PRN sequences are based on a single Sidelnikov/Lempel/Cohn/Eastman (SLCE) generative sequence u^SLCE, and each of said multiple PRN sequences, denoted uⁱ satisfies the equation: uⁱ=u^SLCE⊕Tⁱu^SLCE, where ⊕ indicates element by element binary XOR addition, and Tⁱ indicates a cyclic shift of i chips.

The invention belongs to an improvement scheme of a TDCS (Transform Domain Communication System) based on sequence design, introduces a good correlation characteristic sequence and a spread spectrum structure, and is especially applicable to a multi-access system needing to satisfy a large number of users. The invention provides a TDCS improvement method based on the sequence design and capable of improving a multi-access capability. Under the condition of keeping the advantages, the good correlation sequence which is directly optimized for a periodic correlation characteristic is named a Minimizer for Periodic Integrated Sidelobe Level (MIPSL), as an initial phase sequence, element-by-element multiplication is carried out on the initial phase sequence and the spread spectrum sequence, thereby expanding the initial phase sequence into a sequence set, and the sequence set is allocated to different users.

A system, method, and computer program product allows a web service consumer to create an example query which illustrates the web service desired. The example query is compared, element by element, to potentially relevant web services. The difference between each element in the example query and any element in the web service being compared is quantified as a distance metric, based on the relationship between the elements. The relationships might be classified as: exact match, direct synonym, fuzzy synonym, hyponym, hypernym, or no match. The distance metric for each element in the example query is summed to calculate a total distance between the example query and the web service being compared. This sum is an expression of the difference between the example query and the web service being compared.

Systems and methods for modeling workforce talent supply to enable dynamic creation of job specifications in response thereto. A sampling database is built of candidates using normalized form of elements derived from candidate resumes to provide an element level description of each candidate. The national or regional workforce supply of talent is estimated at an element level by statistically extrapolating the information in the sampling database in accordance with BLS data. A job specification is analyzed to determine a corresponding set of normalized elements. The corresponding set of normalized elements is compared to the estimated workforce supply to determine the degree of match at an element-by-element level. The degree of match is presented at an element-by-element level so the user can identify how elements in the job specification affect the degree of match with the estimated workforce supply and so that the user may alter the job specification if desired.

The invention discloses a CMOS (complementary metal-oxide semiconductor) infrared detector reading-out circuit capable of realizing element-by-elementdark current suppression. The CMOS infrared detector reading-out circuit comprises input circuit, an integration circuit and an output circuit, the input circuit is of a structure combining current storage units with current mirrors, the current storage units are distributed in each pixel of a linear array circuit, each current storage unit is composed of a transmission gate, a virtual switch pair and a capacitance coupling loop, and customized modulation of dark current can be realized; the current mirrors are arranged at the left end and right end of the linear array circuit, and integral suppression of dark current can be realized. The CMOS infrared detector reading-out circuit has the advantages that the current storage units at an input end are designed, so that detector signal heterogeneity can be lowered effectively; the current mirrors at the input end are provided with a rough tuning port and a fine tuning port, so that scope of application, to medium-long-wave infrared detector working current, of the CMOS infrared detector reading-out circuit is expanded, and working state of a system can be adjusted accurately; the CMOS infrared detector reading-out circuit is low in power consumption, and is manufactured by adopting a submicron CMOS process, thereby being high in repeatability.

An electronic receiver may generate a differential detection sequence based on a received symbol sequence and based on a m-symbol delayed version of the received symbol sequence, where m is an integer greater than 1. The particular differential detection sequence may be a result of an element-by-element multiplication of the particular received symbol sequence and the conjugate of an m-symbol delayed version of the particular received symbol sequence. The receiver may calculate differential decision metrics based on the differential detection sequence and based on a set of differential symbol sequences generated from the set of possible transmitted symbol sequences. The receiver may generate a decision as to which of a set of possible transmitted symbol sequences resulted in the received symbol sequence, where the decision is based on the differential decision metrics and the set of possible transmitted symbols sequences.

The invention discloses a pooling feature map processing method, an image target detection method and, a pooling feature map processing device and a medium. The pooling feature map processing method comprises the following steps: respectively carrying out convolution activation processing on a pooling feature map twice, carrying out channel splicing processing on the result of the convolution activation processing twice; and outputting the maximum value of the corresponding position of each channel in the channel splicing processing result. A; and performing activation processing on the output result, performing element-by-element multiplication calculation on the result of the activation processing and the pooling feature map, performing element-by-element addition calculation on the result of the element-by-element multiplication calculation and the pooling feature map, and the like. According to the method, the pixel value information of the important region in the pooling feature map can be amplified, so that the pixel loss generated by the Faster RCNN algorithm in the region pooling process is made up, and the target detection accuracy of the Faster RCNN algorithm is improved. The method is widely applied to the technical field of image recognition.

A method and system for optimization of a simplified engineering model is provided. A software tool efficiently simplifies an engineering model such as a water distribution system and preserves the hydraulic accuracy of the simplified model. The system includes a software program that employs a genetic algorithm to evolve solutions for reinstating the behavior of the original network into a simplified network. The genetic algorithm can be used for identifying the less sensitive hydraulic elements (links and nodes), and removing them or replacing them with the best-fit element parameters produced by the genetic algorithm module of the present invention. A element-by-element skeletonization approach generates the layout of a skeleton network and then identifies potential equivalent elements for replacing series pipes and loops. Yet another alternative includes taking a skeleton network model, and using the techniques of the present invention, optimizes the demand distribution and pipe conductance of the skeletonized model.

The invention discloses a shallow-sea water depth radar remote sensing method. The shallow-sea water depth radar remote sensing method comprises the following steps: acquiring a multi-view SAR remotesensing image of a to-be-detected area, wherein the multi-view SAR remote sensing image comprises q single-view SAR remote sensing images, and each single-view SAR remote sensing image contains imagecharacteristics of a sea wave transmitted from a deep sea area to a shallow sea area in the to-be-detected area; calculating each single-view SAR remote sensing image to obtain a shallow-sea water depth detection result in the shallow sea area; performing tidal correction on the shallow-sea water depth detection result obtained from each single-view SAR remote sensing image; according to a predetermined rule, element by element, forming a one-dimensional shallow-sea water depth array containing q elements based on each corrected shallow-sea water depth detection result, and according to a Kalman filter algorithm, filter each shallow-sea water depth array, wherein the numerical value of the last element in each filtered shallow-sea water depth array is used as the depth of the shallow sea area. By the shallow-sea water depth radar remote sensing method, the accuracy of shallow-sea water deep detection can be improved.

The invention discloses a pedestrian attribute identification method based on deep learning. The method comprises the following steps: taking a pedestrian image as the input of a Deeplab-v2 network to obtain a mask map; multiplying the mask image and the pedestrian image element by element to obtain a foreground image, negating the mask image, and multiplying the mask image and the pedestrian image element by element to obtain a background image; constructing a pedestrian attribute identification network, and combining the regional-level ternary loss function and the weighted cross entropy loss function as a loss function of the network; taking the pedestrian image, the foreground image and the background image as input of a pedestrian attribute recognition network, calculating a networkloss value by utilizing a loss function, optimizing the network through a random gradient descent method, and storing network parameters; and initializing a pedestrian attribute recognition network byutilizing the pedestrian attribute recognition network parameters, and inputting pedestrian images to obtain an attribute recognition result. The method is reasonable in design, so that the accuracyof pedestrian attribute recognition can be greatly improved.

The invention relates to a semantic segmented convolutional neural network with context information coding, which reconstructs the existing SegNet semantic segmented convolutional neural network. Thesteps are as follows: a semantic segmented convolutional neural network structure being built based on SegNet; selecting a position where global features are extracted and a position where global features are fused; determining a unified dimension of the global eigentensor; determining the construction of each global feature extractor and each global feature fusion; the new global features extracted by each global feature extractor being added with the existing global features element by element. SegNet with global information encoding is obtained by connecting the global feature output from each global feature fuser with the current local feature as new feature information. A semantic segmented convolutional neural network with global information coding is obtained.