54 results about "Function learning" patented technology

A Mapreduced short period load prediction method of a multinucleated function learning SVM realizing multi-source heterogeneous data fusion is provided; the method uses multinucleated function to effectively process multi-source heterogeneous data affecting load prediction, and the data comprises history load data, meteorology factors, date types, holiday information, electricity price information and traffic information; the multi-source heterogeneous data can be effectively fused so as to improve nucleus machine performance, thus better using information from different data source; a multinucleated support vector machine is Mapreduced so as to have better speed-up ratio and scalability, thus applying to large scale data analysis.

The invention discloses a reinforcement learning reward self-learning method in a discrete manufacturing scene. The method comprises the following steps: 1, refining the process of the current production line, wherein g belongs to G = {g1, g2,..., gN}, and the intelligent agent reaches a preset target g and is recorded as an interaction sequence episode; according to the initial parameters, obtaining multiple sections of episodes corresponding to the g1 as a target, taking state actions in the episodes and a state difference value delta as a training data set to be input into a GPR module, andobtaining a system state transition model based on state difference; enabling the intelligent Agent to continue to interact with the environment to obtain a new state st, and enabling a Reward network to output r (st), enabling an Actor network to output a (st), enabling a Critic network to output V (st) and enabling a GPR module to output value function Vg as the updating direction of the whole;when the absolute value of Vg-V (st) is smaller than epsilon, considering that award function learning under the current procedure is completed, and carrying out parameter storage of the Reward network; continuously carrying out interaction, and generating the following sub-target g < n + 1 > as the episodes of the updating direction for updating the GPR; and when the set target G = {g1, g2,...,gN} is all realized in sequence, finishing the process learning of the production line.

The invention discloses a pedestrian comparison method based on multi-scale feature fusion, and belongs to the technical field of computer video processing. Multiple pedestrian features are fused, and stability and uniqueness of the comparison features in a multi-camera environment are enhanced; meanwhile, according to expressions of the features on different image scales, the different pedestrian features are compared on different scales; firstly, the pedestrian features are compared and filtered on the small scale, then screened pedestrians are matched on the large scale, and on the premise that the comparison performance of the features is guaranteed, the complexity of the method is lowered; existing texture features are improved, and the novel comparison method based on marked feature points is adopted. According to the method, distance function learning is conducted by the introduction of a semi-supervised distance learning method, the complexity of the training and calibration processes of a traditional distance learning algorithm is lowered, and matching accuracy is improved.

The invention relates to a steering wheel key function learning method which comprises the following steps: S1, determining pull-up resistors matched with the steering wheel key resistors; S2, sampling the sample voltages of different key resistors of the steering wheel; and S3, setting corresponding functions on the steering wheel keys according to the sampled voltages, wherein the step S1 comprises the following steps: S11, sequentially acquiring the sampled voltages of the keys of the steering wheel when different pull-up resistors are used to obtain the acquisition result; and S12, determining the pull-up resistors matched with the steering wheel key resistors according to the acquisition result. The invention also relates to a steering wheel key function learning system and a corresponding vehicular information system. Through the invention, the pull-up resistors can be automatically adjusted according to the steering wheel key resistors of different automobile types, and the shortcoming that good key function setting cannot be realized by the existing steering wheel key function learning method for the steering wheel keys of different automobile types due to relatively largedifferences in the steering key resistors among different automobile types is avoided.

The invention relates to a software defect prediction method based on few sample data learning, and belongs to the field of software engineering. The method comprises the following steps of S1, constructing an SDNN based on a twin network, i.e., a twin full connection network; S2, inputting positive sample data and negative sample data, performing few-sample learning through an SDNN network, and extracting high-level depth features of the samples on the data; S3, performing comparative learning and probability output on the high-level deep features extracted in the step S2 by adopting a metriclearning function, adjusting the proportion of positive and negative samples, and setting function learning parameters, so that the metric learning function more pays more attention to learning of defective data features; S4, obtaining a prediction result. Compared with the prior art, the method adopted by the invention has the advantages that a better prediction effect can be obtained on a limited, high-dimensional and unbalanced data set, and the performance is more stable under different unbalance rates; and a better prediction result can be obtained under the conditions of less data, lesstime and the like.

The invention discloses a short text classification method based on a deep neural mapping support vector machine, and belongs to the field of text classification and deep learning. A short text classification algorithm combining a convolutional neural network and the deep neural mapping support vector machine (DNMSVM) is provided for solving the problems that Softmax is adopted as a classifier forthe convolutional neural network, the generalization capability is insufficient, feature extraction and kernel function learning are needed for directly using the support vector machine for classification and the optimal solution is often difficult to achieve, and thus the classification effect on short text is improved. By means of the method, complex preprocessing of the text is not needed, theaccuracy is high, and the reliability and robustness are improved.

The invention provides a multi-channel playing system. The multi-channel playing system comprises an OTT controller 100, display equipment 200, information source equipment 300 and a remote controller 400, wherein the OTT controller 100 is used for providing a first information source; the display equipment 200 is communicated with a first channel 101 of the OTT controller 100; the information source equipment 300 is communicated with a second channel 102 of the OTT controller 100 and used for providing a second information source; the remote controller 400 is used for sending a control signal; equipment is switched through a mode selection key on the remote controller 400; control coding of different equipment is learned through a function learning key on the remote controller 400; the plurality of information sources are converged by utilizing the OTT controller 100; the required information source is selected through an information source selection switch in the OTT controller 100 and output to the display equipment 200 to play; the switching process can be realized by utilizing one remote controller 400; the disadvantage that the plurality of remote controllers 400 are used in the switching process is avoided; the switching operation is simple and easy to do; and the user experience is improved.

The invention relates to a non-contact remote control learning method. The method comprises the steps as follows: firstly, a first remote controller emits a preset learning function order to a remote control function learning device; secondly, the remote control function learning device receives and recognizes the preset learning function order emitted by the first remote controller and learns the function corresponding to the learning function order, and enables the state of the remote control function learning device to be set as a learning state; thirdly, a user presses one of the buttons on a second remote controller, and the second remote controller then sends out the value of the button to the remote control function learning device; fourthly, the remote control function learning device receives the value of the button on the second remote control device, judges if the remote control function learning device is in the learning state at present, and binds and storages the value of the button on the second remote controller as well as the received learning function order.

The invention belongs to the technical field of artificial intelligence. The invention discloses an artificial intelligence method and system for machine simulation of learning and working of a to-be-simulated target. The artificial intelligence system comprises a data acquisition module, a data identification module, a data analysis module, a central control module, a training module, a knowledgelearning module, an execution module, a path planning module, a data storage module, a terminal module and a display module. The thinking mode of a to-be-simulated target is permanently stored through a training module; the method has important application values for research and behavior analysis of a to-be-simulated target and thinking mode recording of an important to-be-simulated target; meanwhile, a knowledge learning module simulates a cognitive model of an objective object and an intelligent mechanism for logical reasoning based on the cognitive model to a computer system through intelligent calculation and judgment of a to-be-simulated target by a manual method. Intelligence function learning knowledge of a to-be-simulated target is simulated by a machine, and a brain-like artificial intelligence service platform is formed.

The invention discloses a steering wheel key function learning system in one embodiment. The steering wheel key function learning system comprises a control module for providing pull-up resistance values matched with the steering wheels of different vehicle models, a key input port for inputting resistance values generated when different keys of the steering wheel of the original vehicle are pressed down, a pull-up resistance matching module for a control module to provide different pull-up resistance values according to the corresponding resistance values input by the key input port, an A / D conversion module for collecting voltage values when the pull-up resistance matching module provides the different pull-up resistance values, converting the voltage values into digital signals and transmitting the digital signals to the control module, and an I / O interface for providing the A / D conversion modules with different pull-up control ports. The steering wheel key function learning system is capable of matching the operations of the steering wheel keys of automobiles of different types, and therefore, the compatibility of various vehicle-mounted systems in the use process is greatly improved.

The invention provides a multi-task Triplet loss function learning method based on a semantic hierarchy. The multi-task Triplet loss function learning method comprises the following steps: constructing a semantic hierarchy network for a database, performing triplets sampling on the semantic hierarchy, training the multi-task Triplet network, and performing multi-task classification by using a treeclassifier. The multi-task Triplet loss function learning method proposes a loss function combining a semantic hierarchy network with Triplet for solving the problem of multi-level training of the Triplet network, utilizes semantic knowledge to guide a network to differentiate sample structures hierarchically, learns a Triplet feature which contains semantic hierarchy information and is higher ingeneralization, effectively applies the Triplet feature to multi-task learning, and improves feature separability under different semantic hierarchies. Meanwhile, a new hierarchical sampling method is studied, so that the network can mine more effective hard triplets, and finally, the performance of the network is improved.

The invention provides a use case decomposition and function learning-based system automatic testing method. The method comprises steps of transversely decomposing all test use cases to be automatically executed to obtain a factor set for all test use cases, conducting longitudinal summary and conclusion, extracting universal factors for test use cases, conducting function learning, configuring anIP and a route rule of a test computer according to a network plan of a to-be-tested system, orderly reading and saving parameters of each test use case, configuring a system scene according to the current use parameters, checking and judging a test entry condition, mapping key words of the current use case parameters to a function module layer by layer and starting tests, recording at the same time, differentiating business types, test results and determination conditions, judging results, integrating a test result of each use case, outputting a test report and producing a test record file.By the use of the use case decomposition and function learning-based system automatic testing method, quick and expandable system test automation can be achieved; and system test can be accurately andhighly effective conducted.

The invention provides a multimode heterogeneous association entity recognition method based on cross-network representation learning. The method comprises the following steps: giving two multimode heterogeneous information networks, wherein EA and EB are an entity set, RA and RB are an entity relationship set, and TA and the TB are entity type sets; cA and CB are an entity relationship type set;setting two entities EAi belonging to EA and EBj belonging to EB, establishing a multimode relationship transition probability Mij between the EAi and the EBj through an iterative method based on a random walk path set between the EAi and the EBj, and obtaining multimode heterogeneous feature vectors of the EAi and the EBj through Mij by utilizing target function learning; and when it is judged that the EAi and the EBj have multimode heterogeneous consistency, attribute consistency and environment consistency at the same time, determining that the EAi and the EBj are associated entities. According to the method, multimode heterogeneous characteristics of the multimode heterogeneous information network are fully analyzed, and a multimode heterogeneous information network formalized description method and a multimode heterogeneous associated entity recognition model and method based on cross-network representation learning are formed.

The invention discloses an AI optimization data governance method. The method comprises the steps of AI data acquisition and processing, AI metadata optimization and intelligent data quality evaluation management. The AI data acquisition processing comprises data access, data conversion, data loading, strategy template storage and data quality evaluation management; wherein the AI optimization metadata comprises technical metadata and service metadata; and the intelligent data quality evaluation management adopts AI to define a conversion rule, and extracts a data quality evaluation dimension.According to the scheme of the invention, the AI technology is introduced into data management; improvement of data quality, the mining of the association relationship and the blood relationship between the data is improved; a unified strategy template library is provided, strategy templates for data governance of all industries are enriched through AI learning, the technologies of classificationlearning, function learning, regression and the like are innovatively introduced, the conversion rule of the data quality evaluation standard and the weight of each dimension are dynamically adjusted, and the problem of overweight interference of human experience is avoided.

The invention relates to a cold start recommendation algorithm based on commodity attribute information, and aims to solve the problem of data loss in new commodity recommendation by using a multi-core weighted fusion collaborative filtering algorithm. According to the algorithm, the incidence relation of the commodities in the attribute space is determined in a multi-core weighting mode, and therefore new projects are recommended to the user. Wherein the multi-kernel learning algorithm is based on an existing kernel function learning algorithm and is used for carrying out weighted summation on all kernel functions, so that the accuracy of the algorithm in a complex data environment is improved; Wherein the attribute similarity is obtained by calculating the similarity of attributes amongthe commodities, so that the calculated preference score of the user for commodity prediction is more interpretable; Wherein the weight is optimized through a learning method of random gradient descent. According to the method and the device, the item similarity measurement for describing user preferences can be learned according to the attribute information of the commodities, so that the new item recommendation accuracy is effectively improved.

The invention discloses a state distribution perception sampling-based deep-value-function learning method of an agent. The method is used for more quickly learning a value function by the agent underfewer samples, and specifically includes the following steps: 1) obtaining empirical data used for learning the value function by the agent, and defining an algorithm target; 2) using a convolutionalneural network to preprocess the empirical data to obtain a more expressive feature set; 3) using an unsupervised method to cluster the empirical data set in feature space of the empirical data set;4) using a uniform sampling and cluster equal-probability sampling interpolation-based sample state distribution perception sampling method to carry out sampling according to state distribution of theempirical data set; and 5) using sampled samples by the agent for learning of the value function. The method is suitable for use in the game problem of the reinforcement learning field, and can quickly achieve a better result in a case of lesser sample quantity.

The invention belongs to the technical field of personalized recommendation, and particularly relates to a personalized recommendation method and system based on meta-path network representation learning, and the method comprises the steps: building a heterogeneous information network through employing the social relation of a user, the scoring relation of the user for commodities, and the type relation between the commodities; for the heterogeneous information network, extracting a user feature vector matrix and a commodity feature vector matrix through a preset meta path; inputting the user feature representation vector and the commodity feature representation vector as a score prediction model, and completing training learning of the prediction model by optimizing a connection matrix in a target function learning model; and performing commodity prediction scoring on an unknown target customer by using the trained score prediction model, and performing personalized recommendation according to a commodity prediction score. The matrix decomposition performance can be improved by fusing the heterogeneous data information under the condition that the score data is sparse, the personalized recommendation performance is optimized, and the invention has a good application prospect.

The invention relates to a combined multifunctional learning machine multimedia device with a pick-up head, which relates to a learning machine device. The combined multifunctional learning machine multimedia device comprises a multifunctional learning machine, wherein the multifunctional learning machine is mechanically connected with the pick-up head, a fixed hanging wall is used to fix the whole frame on the wall surface through rivets, a multifunctional learning machine circuit is electrically connected with a pick-up head circuit, an interactive electronic whiteboard circuit is electrically connected with a telecommunication wireless transmitting device circuit and is electrically connected with a multifunctional learning machine circuit, and the telecommunication wireless transmitting device circuit is electrically connected with the multifunctional learning machine circuit. An interactive electronic whiteboard, the multifunctional learning machine and the pick-up head are used in a connected and combined mode, all written operations of teachers on the interactive electronic whiteboard can be fed back to the multifunctional learning machine of students through the wireless transmitting technique of a telecommunication multifunctional learning machine in time, which is convenient for students to learn and summarize.

The invention relates to an internet-based enterprise management function teaching platform, including an enterprise management function teacher APP and an enterprise management function learning APP.An enterprise management function platform includes an enterprise leader registration unit, an enterprise management function scene learning module, an enterprise management function network teachingmodule, an enterprise management function teacher invitation module, an enterprise management function scene standard database, a data comparison module and a data processing center; the data processing center is connected with the enterprise leader registration unit, the enterprise management function scene learning module, the enterprise management function network teaching module, the enterprise management function teacher invitation module, the enterprise management function scene standard database and the data comparison module; the data comparison module is connected with the enterprisemanagement function scene standard database; the output of the data comparison module is connected with a data packaging module; the output of the data packaging module transmits data to enterprise leaders; the data processing center is connected with the two APPs. The internet-based enterprise management function teaching platform provides improved teaching efficiency.

The invention discloses a combined multifunctional learning machine using photovoltaic power, and relates to teaching equipment using photovoltaic power. The combined multifunctional learning machine using the photovoltaic power comprises an interactive whiteboard, a multifunctional learning machine and telecommunication wireless transmission equipment. The combined multifunctional learning machine is characterized in that a circuit of photovoltaic equipment is electrically connected with a circuit of a power plug, a circuit of a Chinese learning machine is electrically connected with a circuit of the multifunctional learning machine, a circuit of the interactive whiteboard is electrically connected with the circuit of the multifunctional learning machine, a wireless transmission equipment circuit of the multifunctional learning machine is electrically connected with the circuit of the multifunctional learning machine, and the circuit of the interactive whiteboard is electrically connected with the wireless transmission equipment circuit of the multifunctional learning machine. Photovoltaic self-generating equipment is arranged, so that the combined multifunctional learning machine can be charged as long as light is available; the interactive whiteboard is combined with the multifunctional learning machine, so that all writings of teachers made on the interactive whiteboard can be immediately fed back to multifunctional learning machines of students by means of the wireless transmission technology of the telecommunication multifunctional learning machine; consequently, the combined multifunctional learning machine is convenient for the students to learn and summarize the written contents.

The invention discloses a multifunctional intelligent learning robot. The multifunctional intelligent learning robot comprises an outer shell. The top of the outer shell is provided with an access hole. A moving plate is slidably arranged in the outer shell. A gear motor is arranged at the left end of the bottom of the outer shell. The output end of the gear motor is connected with a threaded rod.The end, far away from the gear motor, of the threaded rod penetrates through the moving plate. The multifunctional intelligent learning robot has the beneficial effects of being capable of storing and letting out a robot body to achieve the dust prevention effect, preventing the situation that electronic elements in the robot are invaded by dust due to the fact that the robot body is invaded bydust or sundries after being exposed to the outside for a long time, and prolonging the service life. A driving motor drives a gear I to rotate to cooperate an internal thread of a rotary seat, and rotating of the robot body can be realized. The gear I and a gear II are in engaged transmission to cooperate with an annular rack on the inner wall of a tube to form a planetary gear structure. The multifunctional intelligent learning robot has the advantages of being small in size, high in bearing capacity and stable in operation.

The invention discloses an IPv6 network service customized reliable routing system and method based on function learning, which belong to the technical field of network routing management. The systemcomprises a service customization layer and a data forwarding layer, wherein the service customization layer integrates the demand types of users, and customized services are microscopically providedfor each user; and the data forwarding layer updates the routing table in the routing node according to the calculated service path, so that the service data flow required by the user is transmitted along the service path. The method is oriented to huge user service demand, service customization based on function learning can effectively guarantee performance reliability and connection reliabilityof service transmission, and the method has great practical significance.

The invention discloses a method for solving highlight quadrant fringes through driver torque feed-forward compensation. The method comprises the following steps: 1) adjusting the axial gain of a machine tool on a controller to ensure that the inertia ratio of an axial motor is correct and vibration does not occur during movement; (2) conducting torsion sharp corner debugging on a human-computer interface of the controller; 3) setting a circle following plane, and starting a parameter compensation function of the controller; 4) ensuring correct setting of a workpiece coordinate system, and performing function learning; and 5) changing the operation mode into starting compensation, and checking a debugging result. According to the method for solving the highlight quadrant veins by utilizing the driver torque feedforward compensation, when a glass part of a 3C product is processed, the torque feedforward compensation can be carried out in the servo driver, the quadrant veins during glass processing can be effectively solved, the debugging process is simple, the operation is convenient, and the practicability is relatively high.

The invention relates to a method for realizing current calibration and hardware calibration of a body controller based on a remote control key, wherein the method includes setting the body controller and the remote control key to enter an initialization learning mode; in the learning mode, according to the current temperature The environment selects the corresponding temperature mode; in each temperature mode, the window and door lock actions are controlled by the remote control key to enter the corresponding function learning mode; the body controller performs the matching learning of the motor load to realize the hardware calibration. The present invention also relates to a corresponding apparatus, processor and computer-readable storage medium thereof. The method, device, processor and computer-readable storage medium thereof of the present invention can be adapted to various vehicle models and different working conditions, and have good compatibility. At the same time, because the learning and calibration operation is simple and convenient, and Has a wider range of applications.

A system for determining a prioritized instruction set for a user, the system comprising a computing device, wherein the computing device is configured to receive at least a physiological goal and provide a plurality of biological extraction data. Computing device may determine a user baseline profile using training data, wherein training data correlates biological extraction data and physiological goals to baseline profile elements, train a machine-learning model using the training data, and determine the user baseline profile as a function of the machine-learning model. Computing device may generate a differential action as a function of the user baseline profile and the physiological goal, receive a plurality of user preference data, and selecting the differential action from the plurality of candidate differential actions. Computing device may receive an updated biological extraction datum corresponding to the user and may modify the differential action as a function of the updated biological extraction datum.

The invention discloses an agent deep value function learning method based on state distribution perception sampling, which is used for the agent to quickly learn the value function with fewer samples. It specifically includes the following steps: 1) Obtain empirical data for the agent to learn the value function, and define the algorithm goal; 2) Use convolutional neural network to preprocess the empirical data to obtain a feature set with stronger expressive ability; 3) In the In the feature space of the empirical data set, an unsupervised method is used to cluster the empirical data set; 4) According to the state distribution of the empirical data set, the sample state distribution-aware sampling method based on uniform sampling and cluster equal probability sampling interpolation is used for sampling; 5 ) The agent uses the sampled samples to learn the value function. The invention is applicable to the game game problem in the reinforcement learning field, and can achieve better results quickly under the condition of less sample size.

A system performs a layout design of a circuit for a small area satisfying a design rule within a short period of time. In a layout design system which includes a processing portion and in which a circuit diagram and layout design information are input to the processing portion, the processing portion has a function of generating layout data from the circuit diagram and the layout design information by performing a Q learning, the processing portion has a function of outputting the layout data, the processing portion includes a first neural network, and the first neural network estimates an action value function in the Q learning.

The invention discloses a dental model point cloud segmentation method based on a cross-graph attention mechanism and cost function learning. The method comprises the steps: firstly constructing a dental model point cloud segmentation model, building an interaction graph network of heterogeneous geometric data in the model, and exploring local information in the same adjacent graph and between different adjacent graphs through the cross-graph attention mechanism. The dependency among the heterogeneous geometric data is learned, the identification capability of context sensing features is improved, and the problem that current heterogeneous geometric data analysis only analyzes each type of data or simple linear combination heterogeneous data neglects a semantic gap among the heterogeneous data is solved; according to the method, on the basis of NAS (NeuralArchitecture Search), a cost function is designed through automatic machine learning and an evolutionary algorithm, the cost function is formulated as an original mathematical operator of a tree structure, the cost function with the highest consistency with a metric function is solved, and the problem that the cost function is inconsistent with the metric function is solved. Compared with other advanced methods, the method has considerable competitiveness.