3156results about How to "Save computing resources" patented technology

Described are systems, media, and methods for applying deep convolutional neural networks to medical images to generate a real-time or near real-time diagnosis.

A method for managing application modalities using dialogue states can include the step of asserting a set of activation conditions associated with a dialogue state of an application. Each of the activation conditions can be linked to at least one programmatic action, wherein different programmatic actions can be executed by different modality components. The application conditions can be monitored. An application event can be detected resulting in an associated application condition being run. At least one programmatic action linked to the application condition can be responsively initiated.

Some embodiments provide a managed forwarding element (MFE that includes a set of flow tables including a first set of flow entries for processing packets received by the MFE. The MFE includes an aggregate cache including a second set of flow entries for processing packets received by the MFE. Each of the flow entries of the second set is for processing packets of multiple data flows. At least a subset of packet header fields of the packets of the multiple data flows have a same set of packet header field values, and a same set of operations is applied to said packets. The MFE includes an exact-match cache including a third set of flow entries for processing packets received by the MFE. Each of the flow entries of the third set is for processing packets for a single data flow having a unique set of packet header field values.

The invention discloses a target detection method based on a dense connection characteristic pyramid network, and relates to the image processing and computer vision technology. The method comprises the steps of collecting an image data set labeled with a target bounding box and category information; constructing a dense connection feature pyramid network containing a Squeeze-Exciton structure capable of learning a dependency relationship between feature channels as a feature extraction backbone network; alternately training the RPN subnet and the R-FCN subnet to obtain a target detection model; and detecting a specific target in the image by using the model. The squeeze-Exciton structure and a dense connection structure are introduced into a feature extraction trunk network, characterization capability of the model is enhanced, adaptability of the model to targets of different sizes is enhanced through the feature pyramid structure, calculation sharing of the whole network model is achieved to the maximum degree through the R-FCN detection head, calculation resources are saved, and performance of the whole target detection model is improved. .

The present invention relates to a multi-microphone system and method adapted to determine phase angle differences between a first microphone and a second microphone signal to detect presence of wind noise.

The invention discloses a pedestrian detection method based on an end-to-end convolutional neural network in order to solve the problem that the existing pedestrian detection algorithm has the disadvantages of low detection precision, complex algorithm and difficult multi-module fusion. A novel end-to-end convolutional neural network is adopted, a training sample set with marks is constructed, and end-to-end training is performed to get a convolutional neural network model capable of predicting a pedestrian candidate box and the confidence of the corresponding box. During test, a test picture is input into a trained model, and a corresponding pedestrian detection box and the confidence thereof can be obtained. Finally, non-maximum suppression and threshold screening are performed to get an optimal pedestrian area. The invention has two advantages compared with previous inventions. First, through end-to-end training and testing, the whole model is very easy to train and test. Second, pedestrian scale and proportion problems are solved by constructing a candidate box regression network, the pyramid technology adopted in previous inventions is not needed, and a lot of computing resources are saved.

The present invention discloses parallel mass data transmitting middleware of Internet of things and a working method thereof. The invention includes: a data collection module, for forwarding data collected from different types of sensors to an application interaction module after analysis; the application interaction module, for achieving communication between the parallel mass data transmitting middleware of the Internet of things and an application system of the Internet of things, analyzing a control command of the application system of the Internet of things, transmitting to the data collection module, and further transmitting the analyzed sensor data to a data management module; the data management module, for performing data screening, fusion, conversion on the mass data analyzed by the application interaction module, and finally storing the calculated data into the application system of the Internet of things; and a developing regulation module, for setting out a protocol plug-in regulation, a communication channel plug-in regulation, a service field metadata plug-in regulation, etc. The present invention can meet high concurrency access and management of the large scale sensing apparatus of different application fields, various communication channels and various communication protocols.

The invention provides a human body behavior identification method based on an inertial sensor. The method comprises the steps of acquiring human body behavior data of testees by means of the inertial sensor, conducting sliding window segmentation on the acquired human body behavior data, conducting feature extraction on a triaxial accelerated speed subsequence and a triaxial angular speed subsequence which are obtained after sliding window segmentation, conducting feature fusion on a feature vector to form a sample set of the human body behaviors of the testees, conducting feature selection on the sample set by means of the least correlated maximum redundant algorithm and the Bayes regularization sparse polynomial logistic regression algorithm, obtaining the classification feature vectors of all human body behaviors of all the testees, obtaining a classifier model of each human body behavior by means of a fuzzy least square support vector machine, and obtaining a human body behavior identification result after the human body behavior data are tested by means of the fuzzy least square support vector machine. By means of the human body behavior identification method, self-adaptation and identification efficiency can be improved.

A computer has protected resources presenting threat vectors that malicious software can use to attack the computer. A security module has monitoring components that monitor the protected resources to detect malicious software. The security module detects if a protected resource enters a steady security state. In response to a protected resource entering a steady state, the security module selectively disables the components that monitor the protected resource, thereby conserving the computing resources utilized by the security module and freeing the computing resources for other tasks. If the resource exits the steady security state, the security module temporarily blocks access to the resource while it enables the monitoring components for that resource.

The invention relates to a track planning method of a redundant space mechanical arm for on-track catching. The catching operation of the space mechanical arm is divided into a rough catching stage and a fine catching stage, so that problems are simplified. By aiming at outstanding problems of different stages, different planning algorithms are adopted. In the rough catching stage, a track searching algorithm based on configuration space is adopted, and the algorithm is based on the test in the configuration space; and in the fine catching stage, a pseudo-inverse algorithm based on kinematics is adopted, and the final precise catching is guaranteed by compensating the relative movement of a target.

The invention discloses a simulation method for high dynamic multi-mode satellite navigation signal source and a device thereof, wherein, the high dynamic multi-modal satellite navigation signal source can simulate satellite signals of the following three systems: Global Position Satellite system, global navigation satellite system and second-generation Beidou satellite navigation system in space, and the satellite signals are received by a dynamic navigation receiver; the signal resource can simulate the moving state of a user receiver, imitate the error of satellite signals caused in the course of transmission and produce three-mode satellite signals in real time. In the embodiment, the device consists of a user interface module, a control module, a satellite parameter calculation module, a satellite signal producing module, an output device and a data receiving and analysis module. The device simulates simultaneously three-mode satellite signals to accelerate the research of multi-mode navigation receiver. By using an inner fumitoshi interpolation method, the device saves calculation resources and improves interpolation accuracy. By using a parametric design, the device can improve the reusable ability of modules and aim at different satellite navigation systems to implement convenient transformation.

The invention provides a method and a system for searching information, as well as an input method system, wherein, the method for searching the information comprises the following steps that: a searching keyword required by a user is determined; according to the searching keyword, the input method system generates a searching command, which is sent out to a searching engine; the searching enginereturns the searching result to the input method system; the searching result is displayed. The invention can greatly improve the searching efficiency, reduce the operation steps of the user, and avoid the invalid data exchange and the invalid data processing. For some simple searching results, the invention can also directly call the display of an affiliated window or candidate word window, thereby a browse is avoided to be called, thus saving the resources of the system and calculating the sources. The invention is applicable to various internet searching engines as well as various local searching engines, and is also applicable to various searching tools or searching websites.

Methods for optimizing data access in a row-oriented relational database containing data sets having attributes using a computer are presented the method including: causing a computer to analyze a database workload to determine an access frequency for each of the attributes; causing the computer to assign each of the attributes to a priority classes corresponding with the access frequency, where the priority classes include a higher priority class and a lower priority class, and where a higher priority class corresponds with a higher access frequency and a lower priority class corresponds with a lower access frequency; causing the computer to store the attributes in accordance with the classes, where the attributes assigned to the higher priority class are stored in a high priority storage medium, and where the attributes assigned to a lower priority class are stored in a low priority storage medium.

A computer based system and method of determining whether to re-fetch a previously retrieved document across a computer network is disclosed. The method utilizes a statistical model to determine whether the previously retrieved document likely changed since last accessed. The statistical model is continuously improving its accuracy by training internal probability distributions to reflect the actual experience with change rate patterns of the documents accessed. The decision of whether to access the document is based on the probability of change compared against a desired synchronization level, random selections, maximum limits on the amount of time since the document was last accessed, and other criterion. Once the decision to access is made, the document is checked for changes and this information is used to train the statistical model.

The invention discloses a travelling robot positioning system based on a single camera. In the system, a point model is adopted for representing an environmental model, an off-line map is calibrated, an action model and an observation model are established for carrying out position updating and road sign calibration, Monte Carto particle filter technology is adopted, particle swarm distribution with weights is adopted for representing robot pose estimation, the weights and the distribution of particles are updated through road sign matching processing, field particle distribution is rasterized, and a sub-region with maximal particle density is selected for positioning the robot. As an independent platform, the system greatly improves the map calibration technology on the environmental model, has great flexibility on the identification of ground line sections, and just needs re-calibration but not needs model re-establishment after environment change. The model reduces the matching complexity in visual processing and improves calculation efficiency. The Monte Carlo localization algorithm is expanded on the information integration processing technology, thus ensuring the real-time property and robustness of the system.

The invention relates to a system and a method for required real-time transcoding and self-adaptive code rate stream media playing. The system comprises a real-time transcoding subsystem, a stream media playing subsystem, a content distribution subsystem and a user terminal module, wherein the real-time transcoding subsystem is used for transcoding the original videos of a video source into video encoding data which is suitable for a user terminal to play, the stream media playing subsystem is used for packaging the video encoding data outputted by the real-time transcoding subsystem into video steams which are suitable for the stream media to transmit, and providing the stream media service for the user terminal, and the user terminal module is used for sending a request to the real-time transcoding subsystem or the stream media playing subsystem, and dynamically switching and transcoding in real time according to the network state information, so as to obtain the video streams with different code rates for playing. By adopting the structure, the system and the method for the required real-time transcoding and the self-adaptive code rate stream media playing have the advantages that the required real-time transcoding of the stream media is realized, the playing code rate is dynamically selected according to the change of the band width of the terminal network, different user terminals can be flexibly matched, and the application range is wider.

The invention discloses a method for detecting mutation information in a multiplex amplification sequencing product of a genome. The method comprises steps as follows: sequencing data are subjected to quality assessment and preprocessing; a recognizable sequencing sequence is selected for sequence assembling; the recognizable sequencing sequence or a sequence obtained through assembling is compared with a reference gene sequence, and preliminary variation information is obtained; fine calibration of sequence variation is performed according to different types of conditions; a calibrated sequencing fragment is obtained; the homozygosis or heterozygosis state of a target fragment is obtained according to the type of the sequencing fragment with the highest abundance; finally, the mutation information in the multiplex amplification sequencing product of the genome is obtained. By means of the method, the amplification product can be rapidly, efficiently and accurately recognized, and the calculation resources are saved; the sequence assembling process is compatible, and the problem of reduction of the quality value of basic groups produced in the sequencing process can be effectively solved; the homozygosis/heterozygosis state of variation information can be more effectively and stably judged, and random errors introduced in the PCR (polymerase chain reaction) process and the sequencing process are eliminated.

Access control to a networked peripheral device by a walk-up user, wherein the networked peripheral device is accessible by both the walk-up user and a remote user, based on centralized access management information. Access control comprises receiving authenticated information for the walk-up user from the networked peripheral device at a centralized location, determining at a server connected to the network a first level of access to the networked peripheral device, determining at the networked peripheral device a second level of access to the networked peripheral device by the walk-up user based on received access management information for the walk-up user, and allowing the walk-up user to access determined user-available features of the networked peripheral device based on the determined first and second levels of access.