16 results about "Function learning" patented technology

The application provides a large model code translation method and device fusing code functions and styles, and relates to the technical field of natural language processing. The method comprises the following steps: obtaining a code pair composed of a source code and a target code from an online programming platform, processing the code pair according to similarity retrieval, fine-grained scoring and differential testing, and constructing a function consistency dataset; performing function learning training on a large model according to the function consistency dataset and an instruction fine-tuning method to obtain the large model trained through the function learning training; obtaining the source code, generating positive sample translation and negative sample translation of the source code, and constructing a style guide dataset; and performing style learning training on the large model trained through the function learning training according to the style guide dataset to obtain a trained code translation large model. The application develops a low-cost and high-efficiency code translation model around a large-scale language model, and enhances the correctness and readability of translated codes.

The utility model discloses a multi-functional learning tower, comprising a tower base, a tower top, two sets of first flat-mouth fastener assemblies and two sets of second flat-mouth fastener assemblies. The tower base comprises two first side plates, a seat plate, and a first desk board. The tower top comprises two second side plates, anti-falling bars, and a second desk board. The tower top is arranged right above the tower base, and the tower base is hinged to the tower top. A height of the first side plate is the same as that of the second side plate. According to the utility model, the tower base and the tower top are connected by the four sets of fastener assemblies, realizing efficient and convenient transformation between a learning tower and a desk and a secure connection.

The invention provides an I / O efficient approximate nearest neighbor search method and device based on function learning, equipment and a medium, and the method comprises the steps: obtaining three billion-scale data sets, carrying out the deduplication and random sampling, and constructing a basic data set and a query set; dividing the data set into a plurality of clusters by adopting a clustering algorithm, training a clustering selection model, and predicting the cluster with the highest probability to which the query point belongs; constructing a Hash model, and mapping high-dimensional data into low-dimensional Hash codes by fusing a composite loss function based on ranking and distance; constructing an index to realize directional retrieval of a target cluster, and obtaining a candidate point set through a priority queue mechanism; training a hash code reconstruction model, reconstructing the hash code of the candidate point set into a vector close to original data, calculating the distance between the reconstructed vector and a query point, and obtaining a top-K approximate nearest neighbor result, so as to solve the problems that index structure optimization is not combined in a hash function learning process, and a large amount of expensive random I / O cost exists.

The invention provides a power load prediction method and device based on multi-scale adaptive feature function learning, and aims to significantly improve the accuracy and stability of power load prediction through deep learning and a feature adaptive mechanism. The method is based on the following core technical means: constructing an adaptive feature function by using historical power load data and multi-time scale features; by learning the feature modes of the load data under different time scales, the data dimension is expanded, and the expression ability of model prediction is improved; by combining characteristic function learning models of different power grid characteristics, various power grid load modes are automatically adapted, and a prediction result is optimized; and finally, through data set training and feedback optimization based on extended features, the prediction precision of the model is improved.

There are provided an apparatus and a method for reconstructing a 3D human object based on a monocular image through depth image-based implicit function learning. A 3D human object reconstruction method according to an embodiment includes: predicting a double-sided orthographic depth map from a front perspective color image of a human object; predicting a signed distance (SD) regarding points on a 3D space from the predicted double-sided orthographic depth map; and reconstructing a 3D human object by using the predicted SD. Accordingly, a human object and details can be naturally reconstructed with respect to not only an area visible through a front perspective color image of the human object but also an invisible area.

The objective is to provide a respondent response estimation system that allows the interviewer to conduct interviews in a natural manner, as if they were actually asking questions to the survey subjects. [Solution] The machine learning device 10 of the survey subject response estimation system 1 is configured to include the following functions: a learning model storage unit 11 that stores and holds a digital clone Ci of each survey subject Pi; a question input unit 13 that inputs question data indicating the content of the question given by the user Q to the required digital clone Cx (=C1, or C2, or ..., or CN) from among the digital clones C1 to CN; and a response output unit 14 that outputs the response obtained from the digital clone Cx according to the question data.

The invention discloses a non-invasive continuous blood pressure prediction method based on multi-dimensional data driving, which comprises the following steps: reading ECG and PPG original signals, and preprocessing the ECG and PPG original signals; wherein the preprocessing comprises filtering, polarity correction, time alignment and normalization preprocessing; time sequence features, amplitude features and waveform morphological features are obtained through calculation based on the preprocessed ECG and PPG signals, and a feature set is constructed in combination with physiological parameters; selecting a deep learning framework, constructing a Transform hybrid model, and designing a loss function; and selecting an optimization function amount learning rate attenuation strategy, setting hyper-parameters including the number of iterations and the batch size, inputting the feature set into the network for training, and testing the model. The electrocardiogram signals, the photoplethysmography signals and the physiological parameters are fused, blood pressure prediction is achieved through an improved Transform architecture, and the adaptability and stability of the model are enhanced while the prediction accuracy is improved.

The invention discloses a risk-aware automatic driving simulation test case sorting method, and relates to the technical field of software testing, and the method comprises the steps: collecting a label-free automatic driving test set, and the test case comprises a plurality of driving scene features; an optimal projection matrix is obtained by adopting objective function learning embedded with double sparse constraints; projecting the test set into a low-dimensional discriminant feature space by using the projection matrix; selecting two test cases with the maximum Euclidean distance in the discriminant feature space as initial elements of a sorting sequence; constructing a strategy scoring device based on a risk-driven component and a prospective exploration component, and generating scores for the candidate cases; the candidate case with the maximum score is selected to be added to the tail of the sorting sequence; repeatedly executing until all the test cases are sorted; and outputting the sorting sequence. Through utilization of dynamic balance risk driving and prospective exploration, an optimal test execution sequence is iteratively constructed, and the capability of covering diversified risk scenes is effectively enhanced.

The application relates to the technical field of data retrieval, and discloses a cross-modal hash retrieval method and system with consistency, which comprises the following steps: S1. acquiring heterogeneous data; S2. acquiring hash codes of the heterogeneous data, and performing hash code learning on the heterogeneous data; S3. performing hash function learning according to the obtained optimal hash code; and S4. mapping the heterogeneous data to the same low-rank Hamming space through the hash function after the function learning is completed, using an exclusive OR operation to the similarity of the heterogeneous data and a retrieval set, returning a result with high similarity, and completing cross-modal retrieval of the heterogeneous data to be retrieved. The application solves the problems of insufficient retrieval precision and complicated optimization process in the prior art, and has the characteristics of being capable of overcoming the heterogeneity of cross modalities.

The application discloses a risk-aware automatic driving simulation test case sequencing method, and relates to the technical field of software testing, and comprises the following steps: collecting a no-label automatic driving test set, wherein the test cases in the test set contain a plurality of driving scene features; obtaining an optimal projection matrix by using a target function learning with embedded double sparse constraints; projecting the test set into a low-dimensional discriminative feature space by using the projection matrix; selecting two test cases with the maximum Euclidean distance in the discriminative feature space as initial elements of a sequencing sequence; constructing a strategy scorer based on a risk-driven component and a forward-looking exploration component to generate scores for candidate test cases; selecting a candidate test case with the maximum score to add to the end of the sequencing sequence; repeating the execution until all test cases are sequenced; and outputting the sequencing sequence. By dynamically balancing the utilization of risk driving and the forward-looking exploration, an optimal test execution sequence is iteratively constructed, and the coverage capability for diversified risk scenarios is effectively enhanced.

The invention belongs to the technical field of electronic measuring instruments, and discloses an intelligent interaction method and system for a signal analysis instrument, and the method comprises the steps: collecting multi-dimensional context data in the operation process of the signal analysis instrument in real time; reasoning a user operation intention through a user intention reasoning model based on the multi-dimensional context data; dynamically screening out candidate function menus from a function menu set of the signal analysis instrument according to the inferred user operation intention, and generating a recommended menu set; and presenting the recommended menu set on a human-computer interaction interface of the signal analysis instrument. The technical problems of deep menu hierarchy, low operation efficiency, difficulty in finding associated functions by a user, high learning cost, poor user experience and the like due to the adoption of a static multi-layer menu in an existing signal analysis instrument are solved.

The invention relates to a belief propagation passive sonar array DOA tracking method based on deep learning likelihood modeling. A belief propagation framework of likelihood function learning fuses deep learning (DL) data driving and belief propagation (BP) multi-frame association capability; a data-driven convolutional neural network (CNN) efficiently maps an array sampling covariance matrix into a sparse power spectrum in a nonlinear manner; according to the likelihood function suitable for Bayes reasoning, a sparse power spectrum is converted into a likelihood function with continuous values by using Gaussian smoothing and cubic spline interpolation technologies; efficiently solving a factor graph of a multi-target tracking problem, and reasoning confidence approximation of target state edge posteriori in combination with a BP algorithm and a constructed likelihood function; the maneuvering target stable tracking and clutter suppression capability is enhanced, and a multi-model interaction algorithm and a time correlation-based target new mechanism are fused. According to the method, the DOA of a plurality of maneuvering weak targets can be stably and continuously tracked at the same time under the conditions of complexity and low signal-to-noise ratio.

The invention provides a software and hardware combined artificial intelligence education learning system, which relates to the technical field of digital systems, and comprises a user layer used for providing corresponding application portals for different user groups; the AI front-end application platform is in communication connection with the user layer and comprises AI exploration learning terminal software which is used for providing an AI exploration type function, an AI learning type function and a special topic type function; the AI teacher teaching assistant platform software is used for assisting the teacher in completing lesson preparation, classroom teaching and after-class learning condition analysis; the AI comprehensive accomplishment evaluation system is used for generating a comprehensive evaluation report by collecting and analyzing student data; and the AI integrated management platform is in communication connection with the AI front-end application platform and is used for realizing a system management function. The interests of students can be stimulated, and the innovation and practical ability can be cultivated; the teaching efficiency and accuracy of teachers are improved; the school competitiveness is enhanced, and course and mode innovation is promoted; future talents can be cultivated for the society, scientific and technological quality of the whole people is improved, and social progress is driven.

The application discloses a kind of unmanned vehicle trajectory planning methods based on continuous action advantage function learning, belong to robot intelligence decision and control field;First, the Markov decision process model of unmanned vehicle trajectory planning is constructed, respectively the expression of state variable, control variable, transition model, loss function of unmanned vehicle trajectory planning is obtained;Then strategy network and evaluation network are established;Again, after every step of unmanned vehicle, strategy network and evaluation network are trained and updated by continuous action advantage function learning, until convergence;Finally, the strategy network for unmanned vehicle trajectory planning is obtained.The application realizes the trajectory planning of unmanned vehicle under the condition that the dynamics model of unmanned vehicle and the environment it is in are completely unknown, so that it reaches the predetermined target in the shortest time, and has very high practical value.

This invention relates to the field of power electronics technology, specifically to a dynamic hybrid modulation method and system for a multi-level active midpoint clamp converter. By constructing a hierarchical cost function that integrates efficiency, thermal balance, and power quality, and combining real-time operating condition perception and online learning mechanisms, the system can dynamically select the optimal modulation strategy under complex scenarios such as load fluctuations, high-temperature aging, and deteriorating heat dissipation. This overcomes the adaptability bottleneck of traditional fixed-rule or offline lookup table methods, achieving autonomous performance optimization across the entire operating domain. A two-layer decision-making mechanism is employed to rigorously verify key safety boundaries such as junction temperature exceeding limits and midpoint voltage imbalance before strategy selection, ensuring that device safety limits are not violated under any operating condition. Simultaneously, a confidence-gated online learning mechanism is introduced to avoid model erroneous updates due to noise or abnormal data, ensuring long-term robustness and stability.

A position estimation system with low power consumption is provided. The position estimation system includes a comparison unit, a learning unit, a data acquisition unit, an inference unit, a data conversion unit, and an evaluation unit. The comparison unit has a function of calculating a first parallel movement amount and a first rotation amount on the basis of machine learning data representing geographic information. The learning unit has a function of generating a machine learning model through learning using the machine learning data, the first parallel movement amount, and the first rotation amount. The data acquisition unit has a function of acquiring acquisition data representing environmental information on the vicinity of a position estimation device. The inference unit has a function of inferring a second parallel movement amount and a second rotation amount, with use of the machine learning model, on the basis of the acquisition data and the machine learning data. The data conversion unit has a function of converting the machine learning data to evaluation data on the basis of the second parallel movement amount and the second rotation amount. The evaluation unit has a function of evaluating the degree of correspondence between the acquisition data and the evaluation data.