86 results about "Discriminative model" patented technology

A training method and apparatus for a generative model. The training method comprises: acquiring a first training sample which comprises a first initial image and first information; inputting the first training sample into a generative model for model processing to obtain a first target image carrying steganographic information; inputting the first target image into a pre-trained discrimination model to obtain a first visibility score, wherein the first visibility score indicates the visibility of the steganographic information; and with the objective of reducing a first loss, adjusting parameters of the generative model, wherein the first loss is positively correlated to the first visibility score.

The invention relates to the technical field of voice processing, can be applied to business scenes such as financial science and technology, medical health and the like, and discloses a data tracing method, device, equipment and medium based on cooperative training, which comprises the following steps: obtaining a generative model and a discrimination model to output initial intermediate feature representation, generating voice data and inputting the voice data into the discrimination model to generate discrimination loss; updating the generation model based on the discrimination loss to obtain a traceable intermediate feature representation, generating an updated intermediate feature representation by using the updated generation model, generating training voice data, and inputting the training voice data and real data into the discrimination model for training to obtain an updated discrimination model; and when input voice data is received, the updated discrimination model is used for discriminating and outputting a detection result whether the input voice data is generated by the generation model or not. According to the invention, through cooperative training of the generative model and the discrimination model, the generative model outputs voice embedded with traceable features, the rejection capability of the discrimination model to unknown sources is improved, and waveform disturbance reduction and accurate source determination are realized.

The invention belongs to the technical field of compressor performance prediction, and particularly provides a method and system for determining a surge judgment critical value of a compressor based on multi-source data fusion. Comprising the steps that a compressor surge test system is built and operated, and a first surge critical value is calculated; obtaining multi-source historical data related to surge of the compressor, selecting characteristic variables to construct historical input vectors, and fitting the multiple regression model by using the historical input vectors to obtain a discrimination model; training the machine learning model to obtain a trained machine learning model; a real-time input vector is constructed based on data, collected in real time, of the to-be-monitored compressor, and then a second surge critical value and a third surge critical value are obtained; and fusing the first surge critical value, the second surge critical value and the third surge critical value to obtain a predicted surge critical value of the to-be-monitored compressor. According to the method, high-precision dynamic testing and historical data regression analysis are combined, and the actual value of the surge critical value is finally and quantitatively determined through multi-parameter coupling measurement and data mining.

The invention relates to the technical field of debris flow disaster monitoring, and discloses a debris flow type discrimination model training method and device based on RFECV multiple models, computer equipment, a computer readable storage medium and a computer program product, and the method comprises the steps: S101, data acquisition and preprocessing; s102, carrying out feature processing; s103, data division is carried out; s104, carrying out sample unbalance processing; s105, preliminarily training the model; s106, optimizing a feature subset; s107, carrying out hyper-parameter range convergence; s108, performing hyper-parameter assignment; s109, training a candidate type discrimination model; s110, performing loop optimization judgment; and S111, outputting a discrimination result. According to the automatic feature optimization method based on RFECV, redundant features are effectively eliminated, and the interpretability and stability of the model are improved; a hyper-parameter adaptive adjustment mechanism is introduced, dynamic model optimization is achieved, training efficiency and generalization ability are enhanced, the problems that traditional classification depends on subjective experience, feature selection is difficult, and the model is unstable are solved, and efficient and accurate classification of debris flow types in the cold and cold mountainous area is achieved.

The invention discloses a special equipment safety monitoring method based on the Internet of Things, and provides a multi-mode sensing data acquisition, noise removal, data standardization and label completion method. Rare abnormal samples are screened through unsupervised clustering and anomaly detection, and diverse pseudo-abnormal data are expanded through a conditional generative adversarial network. And deep representation learning, multi-modal feature mapping and migration fusion are further applied to construct a multi-modal collaborative abnormal feature set, and the multi-modal collaborative abnormal feature set is incorporated into an actively optimized security event discrimination model to realize dynamic rule adaptive evolution.

Method, system and computer-accessible medium can be provided for generating one or more drug conjugates of one or more small molecules. For example, with such exemplary method, system and computer-accessible medium, a multimodal discriminative model can be trained to predict at least one peptide-ligand binding for one or more DNA ligands, a generative nucleotide model can be trained to generate a plurality of compounds. Further, a feedback can be provided from the multimodal discriminative model to fine-tune the generative nucleotide model so as to facilitate the generation of the drug conjugate(s).

The invention discloses an incremental sample screening method for power grid transient stability discrimination model training. The method comprises the steps of obtaining a basic sample set based on a plurality of initial operation modes of a power grid; obtaining a to-be-selected sample based on the generated power grid operation mode; calculating the distance between the to-be-selected sample and each basic sample in the basic sample set; constructing a neighbor sample subset of the to-be-selected sample based on the distance; calculating a neighbor distance between the to-be-selected sample and the neighbor sample subset; judging whether the to-be-selected sample is an incremental sample based on the neighbor distance; and training the power grid transient stability discrimination model by adopting the incremental sample. The screening of the incremental samples only depends on time domain simulation and distance calculation, the identification difficulty of the incremental samples can be greatly reduced, and the data requirement of the artificial intelligence model training for the transient stability discrimination of the power grid can be effectively reduced by identifying the key samples with important value for stability discrimination.

The invention discloses a geothermal fluid source discrimination method based on geochemical index and machine learning fusion, and relates to the technical field of geothermal resource intelligent exploration, and the method comprises the steps: firstly collecting and calculating multi-dimensional geochemical indexes, and constructing a standardized feature sequence; establishing an end member feature library by using unsupervised clustering; identifying the source type of the fluid through a discrimination model fused with an attention mechanism; aiming at the mixed source fluid, constructing an optimization model embedded with physical and chemical constraints, and quantitatively inverting the contribution proportion of each end member; and finally, on the basis of the time sequence proportion data obtained by inversion, predicting a future evolution trend by adopting a time convolutional network-long and short-term memory network model fused with an attention mechanism. According to the geothermal fluid source discrimination method based on geochemical index and machine learning fusion provided by the invention, the whole-process intelligent analysis of the geothermal fluid source from qualitative identification to quantitative prediction is realized.

The invention relates to the technical field of natural language processing, in particular to a method and system for optimizing large model questions and answers based on multi-scale fine-grained feedback, a candidate reply set is generated through multi-model collaboration, and then the large model questions and answers are obtained through a fine-grained judgment model of a multi-dimensional evaluation system including loyalty, integrity, conciseness and the like. The method comprises the following steps: performing structured scoring on each reply, generating a detailed scoring report, and finally, dynamically optimizing target model parameters by adopting a direct preference optimization algorithm and combining user-defined dimension weight distribution to enable an output result to adapt to different scene requirements. According to the method, through multi-model collaborative generation, fine-grained feedback learning and dynamic preference adaptation, the accuracy, reliability and scene adaptability of question and answer reply are remarkably improved, and the method can be widely applied to the fields such as intelligent customer service, medical consultation and education assistance which have strict requirements on answer quality.

The invention provides a discrimination method, a model training method and device, a discrimination system and a related product, and belongs to the technical field of artificial intelligence. The judgment method comprises the steps of calling a first judgment model to perform first judgment on a target agent based on target execution data under the condition that a judgment request for the target agent is received, and obtaining a first judgment result; under the condition that the first judgment result meets a preset screening condition, calling a second judgment model to perform second judgment on the target agent based on the target execution data to obtain a second judgment result; wherein the target execution data is execution data which is carried in the discrimination request and is used for representing that the target agent executes the target instruction, and the preset screening condition is used for determining whether to call a second discrimination model to discriminate the target agent again according to the confirmation degree of the first discrimination result. According to the embodiment of the invention, resource overhead and time delay can be reduced.

Exemplary embodiments of the present application provide a service analysis method and apparatus for a network device, an electronic device, and a storage medium. The method comprises: acquiring an application data sequence and a network data sequence from a service data sequence of a network device; on the basis of the application data sequence and a pretrained discriminative model, determining one or more first application events of an application on a user terminal and corresponding first time information; on the basis of the network data sequence and a pre-established first event determination rule, determining a first network event occurring in the user terminal and corresponding second time information; associating the one or more first application events with the first network event on the basis of the first time information and the second time information; and generating a service analysis result of the network device on the basis of the result of the association.

The invention discloses a method for constructing a stable state judgment model of a medium-voltage direct-current integrated power system. The method comprises the following steps of: making a data set from original simulation data; the invention discloses a construction method of a convolutional neural network model containing a time attention mechanism (TA) and spectral normalization (SN). In the training process, the neural networks with the same initial parameters and the structure are constructed and are respectively a first sub-network and a second sub-network; in the training process, a labeled training set is injected into the first sub-network to calculate cross entropy loss, a non-labeled training set is respectively input into the first sub-network and the second sub-network, consistency loss is calculated, the weight of the first sub-network is updated through back propagation of combined loss obtained after weighting of the first sub-network and the second sub-network, and the cross entropy loss is calculated. And then updating the weight of the second sub-network through exponential moving average. According to the method, dependence on labeled samples can be reduced, and after training is completed, the stable state of the system can be rapidly and accurately judged according to the bus voltage after large disturbance.

The invention relates to a loyalty-driven large model preference optimization method, which comprises the following steps of: S1, acquiring question and answer data pairs containing question cues and reference text blocks; s2, inputting the question cue word into the to-be-optimized large model, and obtaining a model answer; s3, performing fine-grained splitting on the model answer to obtain a plurality of statement unit sets; s4, performing fact reasoning verification on the statement unit set by using a preset discrimination model and taking the reference text block as a basis; s5, calculating an answer loyalty score of the to-be-optimized large model according to a verification result of the statement unit; s6, judging whether the score is lower than a preset threshold value or not; if yes, a preference optimization data item is constructed; and S7, forming a training data set according to the preference optimization data items, and performing training update on the to-be-optimized large model. Through the automatic loyalty evaluation and preference data construction process, the illusion problem of the large model in the professional field is solved, and the reasoning credibility of the model can be improved without manual large-scale labeling.

The invention provides a text generation method, a training method and device of a generation model, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence, in particular to the technical field of deep learning, natural language processing and large models. According to the specific implementation scheme, the theme and the field of a to-be-generated text are determined according to input information; and in response to triggering of the first writing mode, generating a target text conforming to the first writing mode by using a generative model according to the theme and the field, the generative model being obtained by adversarial training with a discrimination model, and the discrimination model being used for discriminating the writing mode of the text generated by the generative model, the writing mode is one of a first writing mode and a second writing mode.

The invention discloses an identification method, device and equipment for predicting a drilling operation state by jointly using deep learning and a tree model, and a medium, and the method comprises the steps: carrying out the mathematical transformation and analysis of obtained time parameters related to drilling operation, and screening out a drilling parameter feature set closely related to the drilling operation state; obtaining an autoregression model and a multivariate classification model which are connected in series as an operation state discrimination model; the screened drilling parameter feature set is input into an operation state judgment model, and drilling operation parameters of the next time period are predicted through an autoregression model; the multivariate classification model is used for judging drilling operation states at current and future moments. The method can be widely applied to the field of petroleum drilling engineering and drilling operation sites.

A single-channel speech enhancement method and device based on a mean-reverting Schrodinger bridge. The method comprises: forming clean speech samples and noisy speech samples into a set of sample pairs; performing preprocessing and Fourier transform on speech samples to obtain spectral complex matrix pairs; constructing a discriminative model, estimating a spectral complex matrix of a clean speech, adjusting model parameters on the basis of a difference between the estimated spectral complex matrix of the clean speech and a spectral complex matrix of a real clean speech, and using the discriminative model corresponding to an optimal parameter as a target discriminative model; constructing a score-based model, estimating a reverse optimal shift score in a reverse generation process, adjusting model parameters on the basis of a difference between the estimated reverse optimal shift score and a real reverse optimal shift score, and using the score-based model corresponding to an optimal parameter as a target score-based model; and providing a noisy speech, using a trained target score-based model to parameterize a reverse optimal shift score of a mean-reverting Schrodinger bridge, and performing the reverse generation process to generate a clean speech. The method involves lower computation costs and achieves a better speech enhancement effect.

The invention discloses an educational scene-oriented generative model-based library reference consultation question and answer method, which comprises the following steps of: obtaining and analyzing a reference consultation request text to obtain a structured analysis result containing a task type, integrity intention clue and the like; constructing a learner portrait in combination with the course policy parameter set, calculating a comprehensive risk score and determining a risk level; then, according to the risk level, the course policy and the task completion degree, determining an allowable highest output level, constructing inference input containing constraint conditions, and generating candidate reply; and finally, obtaining a score through a submittability discrimination model, if the score reaches the standard, executing descending rewriting on the candidate reply, and generating a compliance reply for output. According to the method, the balance between compliance and teaching is realized, the academic disuse risk is effectively avoided, and the reference consultation service quality is improved.

The invention discloses a label determination method and device, equipment and a storage medium, and relates to the technical field of computers.The method comprises the steps that an identification text corresponding to comment information is generated according to the comment information of a current object and an identification instruction, and the identification text is input into a pre-trained large language model, the big language model outputs identification features of the comment information; inputting the comment information and the recognition features of the comment information into a pre-trained discrimination model, and determining a discrimination result of the recognition features of the comment information according to the output of the discrimination model; and determining the label of the current object for the identification feature of the correctly identified comment information according to the judgment result. According to the technical scheme, the large language model determines the more accurate recognition features of the comment information according to the recognition text containing the comment information and the recognition instruction for the comment information, and determines the tag which is more accurate and better matched with the current object according to the recognition features of the comment information with the higher matching degree with the recognition features of the comment information.

This application discloses a model training method, apparatus, device, storage medium, and program product, belonging to the field of machine learning technology. The method generates pseudo-samples based on a first generative model in an adversarial generative network (PGN). The predicted labels of the pseudo-samples are determined by a first discriminative model in the PGN. Based on the predicted and actual labels of the pseudo-samples, the weights of multiple loss functions of the first generative model are iteratively trained until the weights of the multiple loss functions satisfy a first convergence condition. The iterative training of the weights of the multiple loss functions ends, and a first target model is determined based on the weights of the multiple loss functions at the end of the iterative training. This method can learn the most suitable weights for each loss function during iterative training, ensuring the optimality of the weights for each loss function, thereby improving the performance of the final target model.

The application discloses a soil salinization discrimination method based on combination optimization of geochemical parameters, and belongs to the field of soil salinization monitoring. The method comprises the following steps: collecting soil samples in a research area and measuring the content of multi-dimensional geochemical elements; removing outliers and performing standardization processing on the geochemical data; dividing the samples by using a clustering algorithm; screening characteristic elements which are significantly related to salinization and have low collinearity in each division to form an optimal parameter combination; establishing a partial least squares regression model based on the combination, constructing a salinization response function and calculating a discrimination index value; training and optimizing the index by using a machine learning model to obtain a discrimination model; and finally, realizing rapid discrimination of the degree of soil salinization by using the model. The application considers regional spatial heterogeneity and multi-element coupling characteristics, has the advantages of high discrimination accuracy and good stability, and is suitable for regional scale salinization investigation, grading evaluation and early warning.

The invention relates to the technical field of computer vision and machine learning, in particular to a future factory high-precision mold detection method and system based on vision, and the method comprises the steps: collecting and preprocessing a mold image, and constructing a data set; establishing a mold discrimination model which comprises a defect detection module and a score regression module, extracting a defect area in the data set through the defect detection module, and analyzing the defect area based on the score regression module to obtain a mold score; and designing a joint loss function, and optimizing the mold discrimination model to obtain a defect detection result and a quality score value of the mold. And the mold discrimination model constructed by the defect detection module and the scoring regression module is cooperated to perform joint optimization, and the real-time performance and high precision requirements are considered, so that the method is adaptive to the automatic production process of future factories, the model parameter scale and the reasoning delay are effectively reduced, and the deployment of production line edge equipment is facilitated.

The application discloses a carbon footprint data quality detection method and device and a storage medium, and belongs to the technical field of data quality detection. The method comprises the following steps: acquiring to-be-detected data, searching for at least one field knowledge information associated with the to-be-detected data in a knowledge database, calling a data quality discrimination model, identifying data characteristic information of the to-be-detected data, determining data quality information of the to-be-detected data according to the data characteristic information, combining the to-be-detected data, the field knowledge information and the data quality information, generating enhanced prompt information, inputting the enhanced prompt information into a large language model, and obtaining a quality detection result of the to-be-detected data generated by the large language model based on the enhanced prompt information. According to the cooperative detection mechanism of the fusion of multi-source information and the artificial intelligence discrimination model, the accuracy and reliability of the multi-modal and heterogeneous carbon footprint data quality control are significantly improved.

Various aspects of the present disclosure relate to techniques for generative-discriminative artificial-intelligence framework for digital-twin discovery of thin-film materials. An apparatus is configured to generate, using a generative artificial intelligence model, one or more precursor molecular structures for forming a thin film; simulate, using a digital-twin model of a thin-film deposition process, formation of the thin film from the one or more precursor molecular structures; determine, from results of the simulation, one or more physical properties of the thin film; train a discriminative model using the one or more physical properties of the thin film, wherein the discriminative model learns to predict thin-film properties without requiring full simulation; and update the generative artificial intelligence model based on the one or more physical properties of the thin film or predictions generated by the discriminative model to iteratively improve discovery of materials having target dielectric characteristics.

A novel classification device is provided. The classification device includes a memory unit, a processing unit, and a classifier. A plurality of pieces of image data and a discriminative model are stored in the memory unit. Each of the plurality of pieces of image data is image data determined to contain a defect. The discriminative model includes an input layer, an intermediate layer, and an output layer. First to n-th (n is an integer greater than or equal to 2) image data of the plurality of pieces of image data are supplied to the processing unit. The processing unit has a function of outputting feature values of the first to the n-th image data (a first to an n-th feature value) on the basis of the discriminative model. A feature value output from the processing unit is a numerical value of a neuron included in the intermediate layer. The first to the n-th feature value output from the processing unit are supplied to the classifier. The classifier has a function of performing clustering of the first to the n-th image data on the basis of the first to the n-th feature value.

The present application belongs to the technical field of compressor performance prediction, and specifically proposes a compressor surge discrimination critical value determination method and system based on multi-source data fusion. It includes building and running a compressor surge test system, calculating the first surge critical value; obtaining multi-source historical data related to compressor surge, selecting characteristic variables to construct a historical input vector, fitting a multiple regression model using the historical input vector to obtain a discrimination model; training a machine learning model to obtain a trained machine learning model; constructing a real-time input vector based on real-time data collected from the compressor to be monitored, and then obtaining the second and third surge critical values; and fusing the first, second and third surge critical values to obtain the predicted surge critical value of the compressor to be monitored. The present application combines high-precision dynamic testing and historical data regression analysis, and ultimately quantitatively determines the actual value of the surge critical value through multi-parameter coupling measurement and data mining.

The present technology relates to a signal processing apparatus and a method that enable improvement in the robustness of emotion estimation against noise. A signal processing apparatus extracts, on the basis of a measured biological signal, a physiological measure contributing to an emotion as a feature amount, outputs, with respect to time-series data about the feature amount, time-series data about a prediction label of an emotion status by a discriminative model built in advance, and outputs an emotion estimation result on the basis of a result of performing weighted summation of the prediction label with prediction label reliability that is reliability of the prediction label. The present technology can be applied to an emotion estimation processing system.

The invention discloses a deep learning-based computing power performance dynamic allocation optimization method and system, and the method comprises the steps: constructing a combined architecture of a generative adversarial network and an improved Transform model, and enabling a generative model of the generative adversarial network to generate a practical initial computing power allocation scheme by means of a constraint loss function and a task dependence weight matrix; the judgment model adopts a gradient-based updating strategy to improve the judgment capability, and the improved Transform model is dynamically adjusted through position coding, optimized by introducing a regularization term and the like, and accurately processes a calculation power data generation strategy. All units in the system cooperate, the data acquisition and feedback unit acquires computing power data in real time, the model is driven to be periodically optimized, and the distribution instruction conversion unit is combined with a fluctuation compensation factor to implement a strategy. According to the method, the problems that a traditional allocation mode is low in resource utilization rate and slow in response are effectively solved, and the computing power performance and the system operation efficiency are remarkably improved.

The invention discloses an incremental SVM security updating method and system for power grid transient stability discrimination, and the method comprises the steps: obtaining typical and additional scene data based on electromechanical transient simulation, and respectively constructing a pre-training set and an incremental updating set; training the initial SVM model; and performing incremental learning and updating on the model by adopting a safety updating mechanism fusing incremental sample screening, negative sample hard constraint and double-threshold dynamic adjustment, and finally realizing online transient stability judgment. By introducing a safety updating mechanism fusing incremental sample screening, negative sample hard constraint and double-threshold dynamic adjustment, the problems that a traditional discrimination model is difficult to adapt to a complex and changeable power grid scene, an unstable sample is easy to misjudge, and conventional incremental learning efficiency and safety are difficult to consider at the same time are solved.

A waveform information inference device according to one mode of the present invention includes: a waveform extraction unit (31) configured to extract a partial waveform to be modeled from a signal waveform acquired based on actual measurement using a predetermined analysis device; and an adversarial learning unit (32) configured to acquire a model function corresponding to the partial waveform, or the model function and shape distribution information in the function by performing adversarial learning using two mutually adversarial models which are a generation model and a discriminative model using the partial waveform obtained by the waveform extraction unit as an input. The present invention can acquire a precision peak model function and its shape parameter distribution information.

This invention discloses a general method, apparatus, and medium for constructing indexes for unstructured data, applicable to multimodal data such as images, videos, and text. The method includes: employing adapted deep learning models for feature extraction based on different modalities; constructing multi-level cluster structures using a recursive clustering method based on automatic anchor point selection and hyperplane partitioning; automatically generating semantic summaries for each cluster using a large language model; efficiently organizing the index structure through information such as category, cluster, and summary, supporting multi-dimensional retrieval; and during query processing, utilizing a minimal number of existing annotations within each cluster to train a discriminative model at low cost, achieving efficient cluster filtering and multi-category combination queries without requiring users to specify positive or negative samples. This method possesses advantages such as high automation, low annotation cost, strong scalability, support for multimodal data, and unique annotation inheritance, significantly improving the efficiency of index construction and retrieval for large-scale unstructured data.