135 results about "Autocode" patented technology

A system for supervised automatic code generation and tuning for natural language interaction applications, comprising a build environment comprising a developer user interface, automated coding tools, automated testing tools, and automated optimization tools, and an analytics framework software module. Text samples are imported into the build environment and automated clustering is performed to assign them to a plurality of input groups, each input group comprising a plurality of semantically related inputs. Language recognition rules are generated by automated coding tools. Automated testing tools carry out automated testing of language recognition rules and generate recommendations for tuning language recognition rules. The analytics framework performs analysis of interaction log files to identify problems in a candidate natural language interaction application. Optimizations to the candidate natural language interaction application are carried out and an optimized natural language interaction application is deployed into production and stored in the solution data repository.

The invention discloses a recommendation method based on deep social relationship, which comprises the steps of 1, constructing a rating matrix of users for an object and a social relationship matrixamong the users; 2, constructing an input layer in a one-hot coding mode; 3, obtaining a social embedded matrix by automatic coding according to the social relationship matrix among the users; 4, constructing an embedded layer through the input layer and the social embedded matrix; 5, constructing a collaboration layer through the embedded layer; and 6, inputting information of the collaboration layer into a hidden layer to obtain corresponding predicted score information, and thus performing object recommendation. According to the invention, the social relationship among the users can be fully utilized to solve a problem of data sparsity, excellent accuracy is ensured at the same time through establishing a complex deep neural network structure between users and objects, and thus accuraterecommendation is carried out for the users.

The invention relates to a method and a terminal for increasing data transmission safety by using a multi-encryption method. The data to be transmitted is encrypted; the identity authentication is carried out and the data transmission error is checked; the data is processed by using the terminal before being sent; when the data is received, the data is analyzed by using the terminal and then sent to a receiver. The method provides a multidimensional key concept, a key structure needed by encryption is more complicated, and the breaking difficulty is higher. The data processing method and theterminal not only further strengthen data safety, but also use an XML (Extensible Markup Language) organization to transmit data, have distinct levels and are convenient to analyze. A terminal moduleprovides a user interface, is convenient for a user to use an autocode function module and has the advantages of secondary development performance, expansibility and high practicability.

The invention relates to a distributed outlier detection method and system based on an automatic coding machine. The method includes the steps that a training data set and a testing data set are defined; training data of the training data set are distributed to a plurality of calculation units randomly; the calculation units conduct parallel execution, and each calculation unit solves coding and decoding parameters; the coding and decoding parameters of each calculation unit are summarized to obtain a final coding and decoding parameter, and a self-duplication model is built; the self-duplication model is applied to the testing data set, and concurrent computation is conducted on reconstruction errors of all testing data; the testing data are arranged according to a descending order of the reconstruction errors, and the testing data with the reconstruction errors larger than a predetermined threshold value are outliers. According to the method, the total time required for processing and the number of processed samples are independent, and the total time and the number only depend on the required accuracy of parameter solution. The distributed outlier detection method and system based on the automatic coding machine are very suitable for detecting outliers on large-scale data sets on the basis of MapReduce frameworks, and have good flexibility and good expansibility.

The invention described here is intended for enhancing the technology domain of real-time and high-performance distributed computing. This invention provides a connotative and intuitive grammar that allows users to define how data is to be automatically encoded / decoded for transport between computing systems. This capability eliminates the need for hand-crafting custom solutions for every combination of platform and transport medium. This is a software framework that can serve as a basis for real-time capture, distribution, and analysis of large volumes and variety of data moving at rapid or real-time velocity. It can be configured as-is or can be extended as a framework to filter-and-extract data from a system for distribution to other systems (including other instances of the framework). Users control all features for capture, filtering, distribution, analysis, and visualization by configuration files (as opposed to software programming) that are read at program startup. It enables large scalable computation of high velocity data over distributed heterogeneous platforms. As compared with conventional approaches to data capture which extract data in proprietary formats and rely upon post-run standalone analysis programs in non-real-time, this invention also allows data streaming in real-time to an open range of analysis and visualization tools. Data treatment options are specified via end-user configuration files as opposed to hard-coding software revisions.

The invention discloses a building near-zero energy consumption controller based on renewable energy big data deep learning. According to the building near-zero energy consumption controller, featurelearning adopts stacking noise reduction automatic coding in deep learning, depth features in original data are learned in an unsupervised manner, a building renewable energy prediction model is established by utilizing XGboost in machine learning, and finally a linear regression model is constructed by regarding a renewable energy predicted value as an input layer and operation history energy consumption data as an input layer of the energy consumption, thus energy consumption weights of portions corresponding to the operation history energy consumption are obtained, the energy consumption ofenergy-consuming equipment of the building is controlled according to the weights, and the total operation energy consumption approaches the renewable energy predicted value, thereby achieving the control purpose of building near-zero energy consumption.

A machine learning method operates on training instances from a plurality of domains including one or more source domains and a target domain. Each training instance is represented by values for a set of features. Domain adaptation is performed using stacked marginalized denoising autoencoding (mSDA) operating on the training instances to generate a stack of domain adaptation transform layers. Each iteration of the domain adaptation includes corrupting the training instances in accord with feature corruption probabilities that are non-uniform over at least one of the set of features and the domains. A classifier is learned on the training instances transformed using the stack of domain adaptation transform layers. Thereafter, a label prediction is generated for an input instance from the target domain represented by values for the set of features by applying the classifier to the input instance transformed using the stack of domain adaptation transform domains.

A machine learning method operates on training instances from a plurality of domains including one or more source domains and a target domain. Each training instance is represented by values for a set of features. Domain adaptation is performed using stacked marginalized denoising autoencoding (mSDA) operating on the training instances to generate a stack of domain adaptation transform layers. Each iteration of the domain adaptation includes corrupting the training instances in accord with feature corruption probabilities that are non-uniform over at least one of the set of features and the domains. A classifier is learned on the training instances transformed using the stack of domain adaptation transform layers. Thereafter, a label prediction is generated for an input instance from the target domain represented by values for the set of features by applying the classifier to the input instance transformed using the stack of domain adaptation transform domains.

The invention discloses an automatic coding method and device, and the method comprises the steps: carrying out the structural processing of a dictionary file, so as to obtain a standard word; performing structured processing on the obtained original words by utilizing a preset attribute word dictionary to generate structured information; performing structured attribute matching on the structuredinformation and the standard words to obtain candidate results corresponding to the structured information; carrying out encoding post-processing on the candidate result to obtain a processed standardword; and outputting the processed standard words to realize automatic coding of the original words. By means of the technical scheme, free texts in any form can be coded, medical terms of reports and a MedDRA dictionary have no similar or same requirements, a synonym library does not need to be accumulated and maintained, manual input is greatly reduced, and coding efficiency is improved.

The invention discloses a wind power gear box state monitoring method based on a long-short-term neural network and an automatic encoder, and the method comprises the steps: segmenting a sample through employing a sliding window in combination with a bidirectional long-short-term neural network, capturing a time dependence relation in multi-dimensional time series data, and carrying out the reconstruction mapping of multi-dimensional nonlinear time series data to a low-dimensional space through learning a large number of normal samples; on the basis, comparing and analyzing the difference between a reconstructed sample and an original sample to achieve state monitoring and fault diagnosis of the wind power gear box. The method provided by the invention can effectively process high-dimensional time series data, and can be better applied to a variable working condition operation environment of the wind power gear box.

The invention discloses a contact network insulator detection method based on reconstruction and a classification convolution self-encoding network. The method specifically comprises the following steps: 1, imaging a high-speed railway contact network supporting and suspending device; 2, establishing a sample data set of the insulator, and carrying out insulator target detection and segmentation;3, adjusting the insulator to be horizontal by utilizing coordinate transformation, removing noise by utilizing outlier detection, carrying out edge detection on the insulator, then carrying out quadratic function fitting, cutting the acquired insulator images one by one, and finally establishing an insulator piece data set; 4, building a reconstruction and classification convolution automatic coding network, judging whether there is an insulator misclassification or not, and extracting an insulator fault region; 5, clustering the separated foreground images, and establishing a fault judgmentcriterion according to a clustering result; whether the insulator fails or not is judged by setting a threshold value, and the fault level is further evaluated. The detection result is objective, realand accurate, and the defects of a traditional detection method are overcome.

The invention belongs to the field of cutting processing, and particularly discloses a main shaft and workpiece vibration prediction method based on a stack sparse automatic coding network, which comprises the following steps: S1, obtaining main shaft current signals, cutting force signals and main shaft and workpiece actual vibration signals under different cutting processing parameters; S2, inputting the main shaft current signal, the cutting force signal and the cutting machining parameter into a sparse automatic coding network layer for training to obtain a deep time sequence characteristic, inputting the deep time sequence characteristic into a full connection layer, and training the whole network on the basis of a pre-training parameter to obtain a main shaft and workpiece predictionvibration signal; S3, adjusting the stack sparse automatic coding network according to the main shaft and workpiece prediction and actual vibration signals, and completing training to obtain a prediction model; main shaft and workpiece vibration signal prediction in cutting machining is achieved through the prediction model, a dynamic frequency response function can be replaced, a good predictioneffect is achieved in the time domain and the frequency domain, the prediction model can adapt to working conditions of various machining parameter combinations, and the generalization capacity is high.

The invention relates to a day-ahead power load prediction method based on double-end automatic coding, and the method comprises the following steps: obtaining influence factor features which need tobe considered and have an influence on a power load, and forming a historical power load and influence factor feature database; constructing an auto-encoder 1, training the auto-encoder 1, and inputting historical power load characteristic data into the trained auto-encoder 1 to obtain a fused data set; constructing a neural network, training the neural network, and inputting the obtained fused data set into the trained neural network for power load prediction; constructing an auto-encoder 2, and training the auto-encoder 2; inputting the obtained power load prediction data into the trained auto-encoder 2; performing reverse normalization on the obtained decoded power load prediction data to obtain final day-ahead power load data. According to the method, the dimension problem in day-aheadload prediction is solved, and the precision of day-ahead load prediction is improved.

The invention discloses an ICD automatic coding method for electronic medical records based on deep learning. The method comprises the following steps: S1, carrying out vectorization on electronic medical records and medical codes respectively, and acquiring medical record feature vectors and medical code feature vectors; S2, learning information of the electronic medical records to obtain text vectors, and learning the information of the medical codes to obtain medical code vectors; S3, calculating a target function; and S4, reducing differences between the electronic medical records and themedical codes according to the target function so as to complete the ICD automatic coding of the electronic medical records. According to the coding method of the invention, coding candidates are provided for encoders, manual intervention is reduced, and coding efficiency is improved. Through coding, the electronic medical records are well applied secondarily, and statistics and analysis of medical data are better facilitated. Compared with the prior art, the method of the invention has the following advantages: all the electronic medical records come from real ward records of intensive care units and have the characteristics of high authenticity and feasibility, and the method has the advantages of high accuracy and universality.

The invention discloses an automatic coding method for a battery thermal runaway detection system. Included are an intelligent instrument, a background monitoring center and thermal runaway detectors arranged in a battery box. The thermal runaway detection system is additionally provided with a data concentrator. The data concentrator can carry out communication with and power control over the thermal runaway detectors. Each of the thermal runaway detectors is additionally provided with a coding resistor and an AD code collector. The thermal runaway detectors are provided with a coding function. The two ends of each coding resistor are connected with connecting terminals. The plurality of thermal runaway detectors are connected to a coding control end of the data concentrator by a coding line sequentially through the connecting terminals at both ends of the coding resistors to form a series loop. According to the invention, an ID code can be configured with one key, the operation is simple, the error rate is low, the cost is saved, and mass production and packaging of the product is facilitated.

The invention provides an ICD coding method and device, electronic equipment and a storage medium. The method comprises the steps of determining a plurality of candidate ICD codes corresponding to a to-be-coded medical record text; and determining an ICD code corresponding to the medical record text based on the medical record text and the disease knowledge text of each candidate ICD code. According to the method, the device, the electronic equipment and the storage medium provided by the invention, the disease knowledge text of each candidate ICD code is combined in the ICD coding process, sothat the ICD codes obtain the diagnosis and treatment knowledge of the disease corresponding to each candidate ICD code as support while considering the information contained in the medical record text, the problem of overfitting is avoided while automatic ICD coding is achieved, generalization and accuracy of ICD coding are improved, and timeliness of ICD coding is guaranteed.

Disclosed in the invention is a method for realizing file transfer protocol (FTP) access based on automatic code conversion in a heterogeneous environment. According to the method, an interface provided by a FUSE is used for realizing a self-defined user-state file system Ftp_FS; a FTP space unit is mounted to a local file system based on the Ftp_FS; and the access of a mounting point directory by the user is oriented to the FTP space unit again automatically. In the heterogeneous environment, a coding sub module of a Ftp_FS file system is used for mounting a FTP space unit to a local file system based on a specified code format; and an FTP client is used for carrying out accessing. According to the invention, because the FTP is mounted to the local file system, the terminal user can operate a file at an FTP server like the way of operating the local file, thereby improving the user experience. On the basis of intelligent transcoding, a problem of messy code caused by different coding in a heterogeneous environment can be solved.

The invention discloses a bearing anomaly detection method and system. The method comprises the following steps: acquiring bearing acceleration sample data; preprocessing the acceleration sample data;mining feature data of the preprocessed acceleration sample data by adopting an automatic coding machine; carrying out clustering analysis on the characteristic data corresponding to each acceleration sample data; determining normal feature data and abnormal feature data according to a clustering analysis result; training the neural network by taking the feature data as a sample and taking normality or abnormality of the feature data as a label to obtain a neural network model; acquiring acceleration data of a to-be-detected bearing; preprocessing the acceleration data of the to-be-detected bearing; mining feature data of the preprocessed acceleration data of the to-be-detected bearing by adopting an automatic coding machine; and inputting the characteristic data of the acceleration dataof the to-be-detected bearing into the neural network model to obtain a result of whether the acceleration data of the to-be-detected bearing is abnormal. According to the invention, whether the bearing is abnormal or not can be directly identified.

The invention discloses an industrial system anomaly detection method based on a graph attention network and an LSTM automatic coding model. The method comprises the following steps: 1) sample division and standardization: dividing original industrial system data into samples by adopting a sliding window; 2) building an anomaly detection model: building the anomaly detection model by adopting a graph attention network and an LSTM (Long Short Term Memory) automatic coding machine; and 3) performing real-time anomaly detection: calculating an anomaly degree score based on a reconstruction error, and performing anomaly state judgment on the basis. According to the method, the automatic coding machine is adopted, the anomaly detection model is trained in an unsupervised mode, and an anomaly annotation sample does not need to be provided; and the graph attention network is adopted to mine the association between different dimensions of the industrial system, and the anomaly detection accuracy in the complex industrial system is improved.

The invention discloses a method for automatically setting an address code between lamps and realizing coexistence with RDM, and a circuit for realizing the method. After the method is utilized, an automatic coding function is added on the condition of being compatible with a traditional signal wire, and workload in setting the address code of the lamp by a field engineer can be greatly reduced.

The invention discloses an industrial system production anomaly detection and diagnosis method based on multi-dimensional sensing data, and the method comprises the steps: carrying out the preprocessing of a multi-dimensional sensing data sample, and dividing the preprocessed multi-dimensional sensing data sample into a plurality of sub-samples through a sliding window; an automatic coding machine is adopted, and an anomaly detection model is obtained based on normal subsample training in an unsupervised training mode; training a classification model according to the anomaly detection model; and carrying out real-time detection and diagnosis on the production abnormity of the industrial system based on the abnormity detection model and the classification model. The problem that abnormality diagnosis is difficult to carry out under the condition that a black box model is used for multi-dimensional sensing data abnormality detection at present is solved.

The invention provides an ICD coding method and device based on multi-task learning and a graph attention network. The method comprises the following steps: acquiring a medical text corresponding to at least one coding prediction task; enabling the at least one coding prediction task to comprise an ICD coding task; according to the medical text corresponding to the at least one coding prediction task, constructing a corresponding text graph; inputting the text graph into a coding prediction model to obtain a coding prediction result; and enabling the coding prediction result to comprise an ICD coding result, and obtaining the coding prediction model by training a graph attention network based on training data. According to the ICD coding method and device based on multi-task learning and the graph attention network provided by the invention, the medical text is converted into the text graph before coding prediction, so that the problem of data imbalance can be relieved, the automatic coding effect of the medical text is effectively improved, and the coding result is more accurate and reliable.

The invention discloses an automatic serial port testing method, which does not depend on other testing tools, directly calls an interface in an application, achieves the purpose of controllability of input parameters and output parameters, and meanwhile, compilation of a test case is more comprehensive and effective and is not limited by testing a finite function of DEMO any more. According to the method, the interface testing process is greatly simplified, automatic use case execution and result representation are achieved, the testing quality is ensured on the previous basis, and the testing period is shortened; automatic coding and address arrangement are carried out on all connected equipment, so that the dispatching uniformity of the automation of the upper computer instrument at the PC end is further enhanced; when instruction issuing is carried out on the serial port, certain safety and compatibility are ensured through a self-defined instruction structure.

The invention discloses a distributed energy management algorithm based on short-term wind power forecasting results applied to micro-grid energy management. Firstly, a short-term wind power forecasting algorithm in combination with an automatic coding technology, a back propagation algorithm and a genetic algorithm is proposed, and compared with a traditional wind power forecasting algorithm, the accuracy of wind power forecasting can be effectively increased by using the short-term wind power forecasting algorithm; and secondly, based on the short-term wind power forecasting, the energy management problem is modeled as a three-stage Stackelberg game, the dynamic relationship among grid companies, energy storage companies, micro-grids and users is described, and an optimal management solution can be obtained by analyzing a game model through a reverse recursive method.

The invention discloses a self-adaptive host intrusion detection sequence feature extraction method, which comprises the following steps of: extracting a fixed-length feature subsequence and a variable-length feature subsequence to obtain a fixed-length corpus and a variable-length corpus, obtaining a union set to obtain a feature corpus, counting the occurrence frequency of the subsequences in the feature corpus in a to-be-tested system call sequence to obtain a feature vector, and carrying out dimension reduction on the feature vectors by utilizing an automatic coding machine, inputting the feature vectors subjected to dimension reduction into a classifier for classification, and obtaining a classification result. The invention further discloses a self-adaptive host intrusion detection sequence feature extraction system which comprises a fixed-length feature extraction module, a variable-length feature extraction module, a feature fusion module, an automatic coding machine and a classifier. According to the method an system, host program behaviors are described in combination with fixed-length and variable-length features, better adaptivity is achieved, given program behaviors can be better described through variable-length feature extraction, and features highly contributing to classification can be further extracted through a TF-IDF-based fixed-length feature selection method.

Embodiments of the invention are directed to systems, methods, and computer program products for cross-technology code analysis for redundancy identification and functionality recognition. In particular, the novel present invention provides a unique platform for analyzing software code across multiple coding language using a unique approach involving the use of denoising autoencoders. Embodiments of the inventions are configured to leverage a marginalized stacked denoising autoencoder approach to analyze software code, identify code redundancies, and improve efficiency for code storage and query ability by the use of a trained autoencoding module to autoencode software code attributes into vectorized data that can be compared to determine cross-platform functionality and redundancy within a software library.

The invention discloses a traffic data anomaly detection method and device, electronic equipment and a storage medium, and the method comprises the steps: inputting preprocessed traffic data into an automatic coding network model, so as to carry out the automatic coding and decoding of the traffic data, and obtaining a reconstruction feature vector; inputting the reconstructed feature vector intoa recurrent neural network model to obtain internal feature information, wherein the internal feature information is used for representing the front-back correlation of the internal features of the flow data; and finally, classifying the internal feature information through a Sigmoid function to obtain a dichotomy result used for representing whether the flow data is abnormal or not, so that the anomaly detection accuracy can be improved.

The invention discloses an equipment measurement data processing method and system based on a deep neural network, and a terminal, and relates to the technical field of power station equipment data processing, and the key points of the technical scheme are as follows: carrying out entity recognition on target equipment measurement data through a recognition model established based on a bidirectional long-short-term memory neural network and a conditional random field; obtaining a short text sequence which is labeled by a label and jointly expressed by a character vector and a word vector; expanding the short text sequence, inputting the expanded short text sequence into a convolutional neural network, obtaining short text deep semantics by learning deep features in the short text, and performing clustering processing according to the short text deep semantics to obtain clustering equipment measurement data; obtaining a mapping relation between historical equipment measurement data and a standard code through training of a pre-constructed training model, and after clustering equipment measurement data are input into the training model, obtaining a new measurement data prediction code label in combination with the mapping relation. According to the method, unified and standardized automatic coding processing can be carried out on different devices.

The invention relates to a zero-sample image recognition method, in particular to a generalized zero-sample image recognition method based on a conditional adversarial automatic encoder. The inventiondiscloses the generalized zero sample image identification method based on a conditional adversarial automatic encoder, and the method employs a decoder in the conditional adversarial automatic coding machine to generate a pseudo sample through combining with prior distribution and class attribute conditions, and solves a zero sample problem. The invention has the beneficial effects that the sampling conditional adversarial automatic encoder matches integrated posteriori distribution of data potential semantic representation into prior distribution, such as standard normal distribution, so that the stability of a generation model is improved; class attributes are used as conditions to train an adversarial automatic coding machine, the discrimination capability of the model is improved, and generated pseudo samples are closer to real distribution of data; visible and invisible pseudo samples are generated at the same time, and only the pseudo samples are used to train the SVM classifier to eliminate the influence of data imbalance.

The invention provides a slave control module address automatic coding system of a battery management system. The slave control module address automatic coding system comprises a master control module and N slave control modules, the main control module comprises a parameter detection unit and a main control DO control unit; the parameter detection unit is used for detecting whether the on-line slave control modules finish address coding or not and whether the number of the on-line slave control modules is consistent with the number of the slave control modules set by the battery management system or not; each slave control module comprises a slave control DI detection unit and a slave control DO control unit which are respectively used for receiving a DI signal and sending a DO signal; the master control DO control unit is connected with the slave control DI detection unit of one slave control module, each slave control DO control unit is sequentially connected with the slave control DI detection unit of the next slave control module in series, and the DO node of the last slave control module is in empty connection. According to the invention, the address of the slave control module can be automatically coded, material normalization is realized, and the production and maintenance cost is reduced.