53 results about "Sequence memory" patented technology

The invention provides a novel method of transmit beamforming, which allows compact analog implementation of complex digital algorithms without compromising their features. It is aimed to support envelope shaping, apodization, and phase rotation per channel and per firing. Each of three embodiments represents a complete transmit channel driven by pulse-width modulated (PWM) waveforms stored in a conventional sequence memory. PWM signals controls the transmit pulse envelope (shape) by changing the duty cycle of the carrier. Beamformation data are loaded prior to a firing via serial interface. Under the direction of a controller, the circuitry allows high precision (beyond sampling rate) phase rotation of the carrier. It also provides transmit apodization (aperture weighting), which maintains an optimal trade-off among low sidelobe level and widening of the mainlobe. Implementing such an IC, the manufacturing cost of a high-end ultrasound system can be reduced. Equally, the proposed solution makes the benefits of digital transmit beamformers available to midrange and entry-level machines since it merely requires a modified programming of the sequence memory.

The invention provides a method for judging memory leak, comprising the following steps of: when a function application memory is judged aiming at each function in a program, accumulating the memory capacity and the application times of the named function application; when a function release memory is judged, reducing the released memory capacity from the memory capacity of the named function accumulated application and reducing release memory times from accumulated application times; sequencing the memory capacity and the times of the function application in different names in a descending mode by using the memory capacity as priority; ensuring and outputting a level of a memory leak degree according to the memory capacity and the times of the function accumulated application in different names; in addition, the invention provides a system for realizing the method in order to sequence memory distribution and information release according to the memory application times and liberate a programmer from large fussy analytical work.

One embodiment of the present invention provides a system that facilitates implementing a memory mechanism within an asynchronous switch fabric. The system includes a memory device, which does not preserve first-in, first-out semantics such as a random access memory or a stack. The system also includes a data destination horn, for routing data from a trunk line to a plurality of destinations. The memory device is one destination of the plurality of destinations. The system further includes a data source funnel, for routing data from a plurality of sources into the trunk line. The memory device is a source of the plurality of sources.

A resonate power generator is provided. The resonate power generator may include a waveform and sequence memory to record a predetermined waveform and a predetermined sequence, a delta-sigma modulator to delta-sigma modulate an output signal of the waveform and sequence memory, and a waveform recovery unit to receive, as an input, an output radio frequency (RF) signal of the delta-sigma modulator, and to convert the input into a waveform in a form of an impulse.

An intelligent eddy current array probe comprising a plurality of coil elements and an embedded non-volatile memory element is disclosed. Prior to coupling the intelligent eddy current array probe to an NDI system, a data table describing a desired firing sequence for the array probe within a given inspection operation is created. This data table is then stored within the embedded non-volatile memory element of the intelligent eddy current array probe such that when the array probe is coupled to the NDI system, the elements of the NDI system can load and execute the stored firing sequence without operator intervention. In this way, a plurality of intelligent eddy current array probes, each with its own firing sequence, can be used interchangeably within a single NDI system without the need for mechanical adjustments to the NDI system.

The invention relates to a neural network-based scholarship user portrait information extraction method, which comprises the following steps of: carrying out text preprocessing on a webpage, and carrying out structure adjustment and entity labeling on the webpage; constructing Word library tables for all the text nodes and characters thereof of the identified webpage; Extracting context features of the text nodes and the front and rear nodes, and then training a text node sequence to obtain a word vector h containing node sequence context information; And performing decoding calculation on theword vector output, correspondingly obtaining the score of each word vector on the target tag, calculating the tag probability distribution of the node sequence, decoding a model output result, obtaining an optimal prediction tag sequence when the target function is minimum, completing model construction, and performing model training. According to the method, end-to-end training of the model canbe realized, and the problem of long-term dependency between target extraction entities is effectively solved by utilizing the sequence memory characteristic of the LSTM network.

The invention discloses a sensitive content identification method, device and terminal based on voice analysis and a medium, wherein the terminal acquires an audio segment to be detected and a currentvoice scene and acquires a voice characteristic vector of each voice frame in the audio segment to be detected, according to a pre-established feedforward sequence memory network model, semantic textcorresponding to the voice feature vector is obtained by using the voice characteristic vector as an input, and a sensitive information base corresponding to the current voice scene is determined, different voice scenes correspond to different sensitive information bases, the accuracy of judging sensitive contents is improved, a sensitivity index of the semantic text is determined according to abidirectional neural network model established in advance and the determined sensitive information bases, and the semantic text is compared with the semantics of the sensitive information bases to further improve the accuracy, and the audio segment to be detected is marked as sensitive content when the sensitivity index is larger than a preset sensitivity threshold.

The invention discloses a nonvolatile sequence module memory which is divided into a plurality of memory blocks. The data information of every memory block is assigned to a uniform address, a random writing and reading mode is adopted among all memory blocks and a sequence memory mode is adopted in the memory block. Aiming at a great deal of data memory, the memory of the invention changes the disadvantages of an RAM memory mode and proposes a set of novel nonvolatile sequence module memory and a memory and writing and reading method, thereby being capable of improving the writing and reading speed of a phase-change memory chip and writing and reading repetition times and reducing power consumption.

The invention discloses a neural network training and voice endpoint detection method and device. The method comprises the steps: randomly mixing voice audio data and non-voice audio data to form mixed audio data; extracting acoustic features of the mixed audio data; and inputting the acoustic features into an FSMN model, and training the FSMN model to enable the classification of the voice audiodata and the non-voice audio data output by the FSMN model to be basically equal to the classification of the voice audio data and the non-voice audio data in the mixed audio data. According to the method, the non-voice audio data and the voice audio data are mixed and then the mixed data are used for inputting of a feedforward sequence memory neural network to train the neural network, so that the neural network can output information that whether the audio data units belong to voice audio data or non-voice audio data and then the information can be used for voice endpoint detection, and theresult of voice endpoint detection is more accurate.

A code converting apparatus is designed to estimate the incoming time and the like of a signal modulated by a spreading sequence composed of m elements in a set A having n kinds of symbols as elements accurately. A symbol sequence memory section 102 stores a converted sequence of symbols obtained by concatenating consecutive 2d symbols at the center of each sequence pattern p_i composed of 4d elements in the set A. The converted sequence of symbols is structured so that d symbol immediately preceding the 2d symbols and d symbol immediately following the 2d symbols are identical to d symbol immediately preceding corresponding 2d symbols and d symbol immediately following the corresponding 2d symbols of a corresponding sequence pattern. A symbol sequence specifying section 103 specifies a matching sequence pattern for every consecutive 2d symbols in a symbol sequence obtained by assigning a symbol to the signal in a coding cycle of the spreading sequence. A signal generating section 104 adds a portion to which the 2d symbols are assigned of the signal. A signal converting section 105 concatenates each pattern signal added for each sequence pattern, in an order equivalent to an order of the 2d symbols at the center of each sequence pattern concatenated in the converted sequence of symbols.

An electric drive for a vehicle uses an inverter circuit configured to provide a multi-phase AC drive signal to an electric machine according to a PWM drive signal. A PWM controller monitors an operating point of the electric drive and is configured to identify an operating region of the electric drive from a plurality of predetermined regions in response to the monitored operating point. A sequence memory stores a plurality of sequences, each sequence optimized for a respective one of the predetermined regions to reduce audible noise and switching losses. The PWM controller generates the PWM drive signal according to a respective sequence corresponding to the identified operating region. Thus, the reduction of audible noise and switching losses is optimized based on torque and/or speed of the motor.

A method is presented to address quantitative assessment of spatial sequence memory in a subject, where the method comprises the steps of: (1) presenting at least one scene to the subject on a display, the scene comprising a plurality of elements and a background; (2) modulating the saliency of a predetermined section of said scene(3) receiving feedback from the subject via at least one input device; (4) modifying the saliency of the predetermined section; (5) adjusting a functional assessment parameter relative to the scene; (6) moving the predetermined section relative to the scene; (7) receiving refined feedback from the subject via the input device; (8) quantitatively refining the refined feedback; (9) calculating a critical threshold parameter for the subject; and (10) recording the critical threshold parameter onto a tangible computer readable medium. An apparatus for quantitative assessment of spatial sequence memory of a subject comprising a display device, an input device, a control device, and a tangible computer readable medium. In its simplest sense, a quantitative profile of spatial sequence memory by psychophysical responses is generated on a tangible computer readable medium.

A communication device test signal generating apparatus wherein a sequence memory stores an order in which to read unit data including I and Q waveform data; read addresses; first sequence information including a desired signal level to be set in the unit data; and second sequence information including a frequency offset, and wherein the first and second sequence information stored in the sequence memory are sequentially used to apply a plurality of steps of frequency offsets to the I and Q waveform data in a digital stage preceding a D/A converter at predetermined frequency intervals with a predetermined carrier frequency being referenced, thereby outputting the frequency hopping test signal.

A system for the early detection and treatment of catheter-induced ischemic strokes includes a magnetic resonance system (20) and a processor (36). The magnetic resonance system includes a sequence controller for performing each of a plurality of imaging sequences and a sequence memory (32) which stores at least a magnetic resonance angiography (MRA) sequence, a diffusion-weighted imaging (DWI) sequence, and a perfusion-weighted imaging (PWI) sequence. The processor is programmed to control the magnetic resonance system to perform the steps of: performing the MRA sequence to generate a baseline MRA image; performing a catheter-tracking procedure to track a catheter; performing the DWI sequence after the catheter procedure to generate a diffusion-weighted image; performing the PWI sequence to generate a perfusion-weighted image; and combining the diffusion-weighted image and the perfusion-weighted image to generate a combined image for evaluating the ischemic stroke.

Techniques are disclosed relating to memory consistency in a memory hierarchy with relaxed ordering. In some embodiments, an apparatus includes a first level cache that is shared by a plurality of shader processing elements and a second level cache that is shared by the shader processing elements and at least a texture processing unit. In some embodiments, the apparatus is configured to execute operations specified by graphics instructions that include (1) an attribute of the operation that specifies a type of memory consistency to be imposed for the operation and (2) scope information for the attribute that specifies whether the memory consistency specified by the attribute should be enforced at the first level cache or the second level cache. In some embodiments, the apparatus is configured to determine whether to sequence memory accesses at the first level cache and the second level cache based on the attribute and the scope.

It is an object to properly control timing of opening and closing a switch by a circuit of simple constitution. A switch control apparatus for controlling a switch is provided, the switch control apparatus including a sequence memory for recording a sequence pattern includes open/close instruction data which instruct to open/close the switch thereon; an address control module for sequentially retrieving each of the open/close instruction data of the sequence pattern from the sequence memory; and an open/close state storage module for storing an open/close state instructed by changed open/close instruction data when the open/close instruction data retrieved by the address control module is changed, wherein the switch opens or closes in response to the open/close state stored by the open/close state storage module.

The invention relates to a real-time data flow anomaly detection method and system. The method comprises the following steps: firstly, for a real-time data stream, constructing a context relationship of the data stream through an HTM network, and on the basis of an abnormal condition of the analysis data, giving a preliminary evaluation result; secondly, giving error statistics for the preliminary evaluation result, and carrying out error probability model modeling on the related error statistics result; and finally, evaluating whether the data flow is normal or not by combining a preliminary evaluation result and an error statistical result, and constructing an abnormal probability detection model. According to the invention, real-time anomaly detection is carried out on the data stream based on an online sequence memory algorithm of hierarchical time memory, so that the problems of non-automation and low abnormal data detection accuracy in the existing data stream anomaly detection are solved.

The invention relates to an e-commerce trial evaluation and recommendation system based on user requirements. The system is characterized by being composed of a physical store trial terminal, a streetdisplay trial terminal, a marketing management center and a mobile phone terminal, and the physical store trial terminal and the street display trial terminal are respectively composed of a trial area, a keyboard input and display module, a gift area, a data processing and storage module and a wireless transmission module; the marketing management center is composed of a data preprocessor, a datamining PDM server, an FIFO sequence memory, a data exchange server, a marketing management server and a permanent memory. And the marketing management server is connected with a mobile phone terminalof a customer who participates in trial through a mobile communication network. The system solves the problem that the existing e-commerce enterprise has few means of attracting consumers and miningnew customers and is poor in efficiency, and compared with the existing telephone marketing means, the system takes the user requirements as a fundamental starting point, avoids the blindness, improves the return rate, and saves the manpower and material resources.