38results about How to "Reduce computing resource consumption" patented technology

Techniques for providing reliable transport of data across a network between a server on a server device and a client on a client device include a process on the client device which determines a number of packets to be sent by the server device without an acknowledgement message from the client device. The number of packets is based on considerations of network problems. A data packet sent to a server device includes a first field holding data indicating the number of packets to be sent by the server device without an acknowledgement message from the client device. Accordingly, a transport layer process on a client device to download large amounts of content from a stateless transport layer process on a server device, substantially reducing the computational resources consumed by the server device in serving thousands of clients.

The invention discloses an overlay convolutional network-based rolling bearing failure mode recognition method and device, and relates to the field of rolling bearing failure diagnosis. The method comprises the following steps of: extracting a time-frequency domain feature of a vibration signal of a state-known rolling bearing; normalizing the obtained time-frequency domain feature of the state-known rolling bearing into a feature pixel according to a CNN network input format; inputting the feature pixel into a CNN network, and adjusting a model parameter of the CNN network through carrying out forward self-learning and gradient descent-based counter-propagation on the CNN network so as to obtain a trained CNN network; and during the recognition of a practical rolling bearing failure mode, extracting high-order features capable of reflecting intrinsic information layer by layer by utilizing the trained CNN network by taking a time-frequency domain feature of a vibration signal of a state-unknown rolling bearing, and inputting results of the feature self-learning into a top classifier layer by layer, so as to realize failure mode recognition of the rolling bearings under multiple working conditions and strong noises.

The invention discloses a height measurement method and device of a highway height permitted frame for a vehicle based on binocular vision. Two wide-angle cameras and a binocular vision system are used for shooting, tracking and locating a height permitted beam target in real time, the binocular vision system is used for measuring the distance from the vehicle to the height permitted frame and therelative height of a vehicle roof in real time, early warning information is provided for drivers, driving and road traffic safety are ensured, a space geometry relation is used for calculating the height of a road frame and the distance between the vehicle and the road frame, the height measurement method and device have the advantages of being high in computing speed and less in calculation resource consumption, the real-time requirements of vehicle safety early warning are effectively met, a limit constraint binocular correction method is adopted to reduce the matching search space, two-dimensional search is reduced to one-dimensional search, the advantage of high matching speed is realized, a checkerboard calibration method is adopted, the internal parameters of the camera can be obtained effectively, and the accuracy of image depth data calculation can be ensured.

The invention provides a tumor mutational load detecting method, a device and a storage medium. The method comprises the steps of acquiring mutation frequency data of a single testing sample, whereinthe mutation frequency data comprise the site mutation frequency of a sample target area, comparing the site mutation frequency with a preset threshold for obtaining a mononucleotide mutation which ishigher than the threshold, and eliminating meaningless mutation in mononucleotide mutation for obtaining the number of effective mononucleotide mutations; acquiring Indel mutation frequency data of the single testing sample, wherein the Indel mutation frequency data comprises the Indel mutation frequency of the sample target area, comparing the Indel mutation frequency with a preset threshold forobtaining the number of the Indel mutations which are higher than the threshold; calculating the number of the tumor mutational load according to a number evaluation formula of the tumor mutational load, wherein the number estimation formula comprises the weight item of the number of the effective mononucleotide mutations and the weight item of the number of the Indel mutations. The method according to the invention accurately detects the TMB index of the lung cancer sample under a precondition of no dependence to a common database and a normal pairing comparison sample.

The invention discloses a key frame extraction method based on inter-frame difference and color histogram difference. The method comprises the steps: in the first stage, detecting a moving target in amonitoring video based on inter-frame difference, dividing the monitoring video into a video sequence only containing the moving target, and obtaining candidate key frames; and in the second stage, calculating the color histogram difference value of two adjacent frames in the candidate key frames, setting a threshold value, removing similar image frames, and obtaining a final key frame set. Compared with other key frame extraction technologies, the method has the advantages that the time required for key frame extraction is remarkably shortened, the real-time requirement is met and redundantinformation in the set is reduced as much as possible while relatively high representativeness of the content of the key frame set is ensured, so that the number of key frame frames is relatively small, and thus, the time complexity of task algorithms such as subsequent target identification and pedestrian quantity statistics is reduced.

The invention discloses a method and a device for real-time simulation of a steady-state sound field space in a marine environment. The method comprises: receiving original sound source data input bythe user; establishing the background model of the sound field space in the marine environment; determining a steady-state sound field space matrix in the marine environment according to the background model of the sound field space in the marine environment, the original sound source data, a sound trajectory equation under a non-coplanar condition of the sound ray and the sound velocity gradientvector, a eikonal equation of a discrete process, and a group march method GMM; outputting the steady-state sound field space matrix in the marine environment. The method utilizes the sound ray tracking technology of the steady-state sound field space to obtain the sound field information in the whole marine environment of the region of interest under the requirement of high precision, and the amount of calculation required is low.

The invention discloses an online game data updating method. The online game data updating method includes steps of establishing a corresponding table of a dependency item in a calculation function and a unique mark for all data affected by the dependency item according to the calculation function preset in advance and the unique mark of the data corresponding to the calculation function. Further, if the dependency item is varied, modified marks of all the data, affected by the dependency item, are real. When in needs of acquiring the data, if the modified marks of the data are real, the calculation function can be searched out according to the unique mark of the data and values of the data can be calculated and stored again, and recalculated values of the data can be fed back, and then the modified marks of the data are marked as false; if the modified marks are false, the values of the data calculated and stored are fed back directly. The online game data updating method can reduce resource consumption of data calculation and improve calculation efficiency.

The invention relates to a method for actively providing personalized service, a device, a cloud server and an automobile. The invention aims to realize a personalized service recommendation effect. The method comprises the following steps: receiving sensing information input by a multi-dimensional sensing module; inputting the sensing information into a preset rule model to judge whether one rulematched with the sensing information exists in the preset rule model or not; if so, outputting a first intermediate value; if not, inputting the sensinginformation into a locally stored user portraitmodel to obtain a second intermediate value; respectively carrying out similarity comparison on the first intermediate value or the second intermediate value and each intermediate value in a pre-stored standard recommendation instruction library, and determining one intermediate value with the highest similarity with the first intermediate value or the second intermediate value; determining a recommendation instruction corresponding to one intermediate value with the highest similarity in a standard recommendation instruction library as a target recommendation instruction; and according to the target recommendation instruction, controlling a related actuator on the vehicle to execute the target recommendation instruction.

The invention relates to artificial intelligence, and discloses a medical image picture analysis method, which comprises the steps of training a pre-constructed deep learning network model by utilizing a preset part disease historical picture set to obtain a disease recognition model; constructing a distillation loss function according to the disease identification model and a pre-constructed initial diagnosis model; performing distillation training on the initial diagnosis model according to the distillation loss function to obtain a first diagnosis model; performing training and output adjustment on the first diagnosis model according to a preset diagnosis target to obtain a target diagnosis model; and when the to-be-analyzed medical image picture is received, analyzing the to-be-analyzed medical image picture by utilizing the target diagnosis model to obtain an analysis result. The invention further relates to a block chain technology. Data of the training model can be stored in theblock chain. The invention further provides a medical image picture analysis device, electronic equipment and a computer readable storage medium. According to the invention, the consumption of modelcomputing resources for medical image picture analysis can be reduced.

The invention provides a voice print recognition method and device, a storage medium and electronic equipment, and relates to the technical field of artificial intelligence. The method comprises the following steps: determining a candidate unit structure; constructing and training a neural network model; adding a mask module into the neural network model, and constructing a generator; using the trained neural network model as a baseline network, and constructing and training a generative adversarial network model through the baseline network, the generator and a discriminator; according to the trained generative adversarial network model, determining a to-be-clipped intermediate state node and clipping the to-be-clipped intermediate state node; after retraining the cut neural network model, extracting voice print features of the to-be-recognized target; and determining a voice print recognition result based on the similarity. The problems that computing resources are consumed in the searching process of an existing neural network architecture, the number of network parameters in the searched network structure is large, and the calculation amount is large can be solved.

The embodiment of the invention provides an intelligent contract data processing method, a block chain network and a storage medium. The method comprises the following steps: explaining and executinga code of an intelligent contract through a consensus node of a blockchain network; tracking the intelligent contract data generated in the interpretation and execution process to obtain a tracking result of the internal transaction between contract accounts; and analyzing the tracking result to obtain information of the internal transaction.

The invention relates to a context uncertainty elimination system based on hierarchical integrated quality index QoX and a working method thereof. The context uncertainty elimination system based on the hierarchical integrated quality index QoX comprises a context acquisition layer, a context processing layer and an application layer. The invention adopts quality indexes such as quality of device,quality of context, quality of service, quality of experience and the like as a hierarchical integrated quality index QoX, and integrates the hierarchical comprehensive quality index QoX into a context sensing system to provide an effective method for context uncertainty elimination in context sensing computing; at the same time, a user and the system can actively adjust various preset thresholdsand rules according to the own needs and changes in environmental conditions, which makes the system more reliable and adaptable.

The invention provides a method and a system for rapidly eliminating false detection for the position detection patterns of a QR code. The method comprises the following steps of: dividing a first sampling area above the upper edge of each position detection pattern, dividing a second sampling area below the lower edge of each position detection pattern, dividing a third sampling area at the left side of the left edge of each position detection pattern, and dividing a fourth sampling area at the left side of the right edge of each position detection pattern; acquiring the chromatic value of each pixel point in each sampling area, then counting the number of the pixel points belonging to the same chromatic value, and obtaining four histograms by taking the chromatic value as a transversal coordinate and taking the number of the pixel points as a horizontal coordinate; filtering and deriving the four histograms respectively, judging whether two histograms are histograms presenting single-peak characteristics, if so, then correctly detecting; or else, the detection is false detection. The method and the system are capable of simply and rapidly judging whether the positioned position detection patterns are correct, and further has the advantages of being low in calculation resource consumption, and high in positioning result accuracy.

The invention provides a method for predicting stability of a new crown vaccine based on a quantum gating recurrent neural network, and belongs to the technical field of quantum computing. According to the method, a quantum gating circulating unit is used for resetting and updating a basic group in an RNA sequence and a corresponding hidden state, a new hidden state is generated, and the reaction rate and degradation rate of the RNA sequence at different positions are output, so that the stability of the new crown vaccine is predicted, and therefore, the stability of the new crown vaccine is predicted. The method meets the requirement for establishing the long-term and short-term dependency relationship of the sequence by a gated recurrent neural network algorithm, and is low in computing resource consumption, so that the method is widely applied to cooperative work of a quantum chip and an electronic chip.

The invention discloses a satellite component layout temperature field prediction method based on uncertainty. The method comprises the following steps: S1, constructing a deep neural network model; s2, generating a training data set and a pool data set; s3, training a deep neural network model by using the training data set; s4, repeatedly performing temperature field prediction on each pool of data for multiple times by using the model, and calculating the variance of multiple prediction results; s5, setting the pool data in a descending order according to the variance, selecting a preset number of pool data located in the front order, taking the selected pool data and the corresponding temperature field as new training data to be added into the training data set, and deleting the selected pool data from the pool data set; s6, continuing to train the model; s7, judging whether the prediction precision of the model meets a preset requirement or not; and if not, repeatedly performing temperature field prediction on the data of each pool for multiple times by using the model, calculating the variance of multiple prediction results, and returning to the step S5. According to the invention, the high-precision agent model can be obtained through a small amount of training data, and the calculation cost is reduced.

The invention relates to a mathematical formula positioning method. The method comprises the following steps of obtaining a row minimum bounding rectangle of each row of characters from a to-be-identified document; obtaining a character minimum bounding rectangle of each character; dividing the minimum enclosing rectangle of the character into the minimum enclosing rectangle of the row; verticallycovering and combining the minimum enclosing rectangles of the characters in each row to obtain a minimum enclosing rectangle of the combined characters; according to the minimum enclosing rectangleof the combined character, adopting an interval mode method, and counting and analyzing the general area of the Chinese character; and processing the minimum enclosing rectangle of the combined character according to the general area of the Chinese character to obtain the minimum enclosing rectangle of the mathematical formula. According to the method, the minimum enclosing rectangle of the mathematical formula is obtained by combining the characteristics of the mathematical formula for processing, the algorithm complexity of the positioning method is low, the operation resource consumption isreduced, and the positioning speed is increased. The invention further relates to a mathematical formula positioning system, a readable storage medium and computer equipment.

The invention discloses a convenient sun sensor in-orbit health monitoring method and system, and the method comprises the steps: judging whether a preset in-orbit health monitoring starting condition is satisfied or not according to a current working mode of a satellite; if it is determined that the preset on-orbit health monitoring starting condition is met, starting on-orbit health monitoring, and executing digital sun sensor on-orbit health monitoring, fixedly-installed simulation sun sensor on-orbit health monitoring and on-sailboard simulation sun sensor on-orbit health monitoring sequentially; and outputting the obtained on-orbit health monitoring result of the digital sun sensor, the on-orbit health monitoring result of the fixedly mounted simulated sun sensor and the on-sailboard simulated sun sensor as comprehensive monitoring results. The problems that the measurement and control arc section of a low-orbit remote sensing satellite is short, and fault diagnosis of the sun sensor is not timely are solved.

The invention, which relates to the technical field of communication, discloses a wireless positioning method and apparatus. According to one embodiment, the method comprises: a positioning device is deployed in a regional map as a positioning base point and positioning coordinates of the regional map are generated according to a physical relationship between a to-be-localized device and the positioning device; a to-be-detected coordinate of a to-be-detected point is calculated based on a relative relationship between the to-be-detected point and the positioning base point; and the to-be-detected coordinate is matched with the positioning coordinates and an actual coordinate corresponding to the positioning coordinate with the highest matching degree is selected as an actual coordinate of the to-be-detected point. Therefore, the cost of early-stage positioning preparation is lowered; and the calculation difficulty during positioning is reduced.

The invention provides an image processing method and device for a three-dimensional scene, electronic equipment and a computer readable storage medium. The method comprises the steps: carrying out image acquisition processing on the three-dimensional scene to obtain a scene image of the three-dimensional scene; storing the scene image in an image buffer area of a memory; carrying out coloring processing on the scene image in the image buffer area through at least one coloring component so as to update the scene image into a target scene image with target visual features; and carrying out rendering processing on the target scene image in the image buffer area so as to present the target scene image on a human-computer interaction interface. According to the method and the device, the target visual features can be accurately and quickly added to the three-dimensional scene.