135 results about "Latent variable" patented technology

The invention relates to the field of artificial intelligence and computer science, in particular to a causal structure and latent variable-based adaptive reinforcement learning reasoning migration method, which comprises the following steps of: constructing a causal world model fused with multi-modal observation and a decoupling latent variable space; establishing a hierarchical inference engine comprising an intuition layer, a conventional layer and a planning layer; pre-training a quick response and judicial planning dual-mode strategy and generating an interpretable fuzzy rule base; performing calculation level coarse tuning based on task identification and causal complexity; evaluating the real-time state criticality through an adaptive neural fuzzy system and dynamically switching a decision mode; after the action is executed, the threshold and the rule are subjected to closed-loop optimization, and cross-environment efficient migration is realized by utilizing a causal modularization characteristic. According to the technical scheme, consumption of computing resources is remarkably reduced on the premise that decision precision and safety are guaranteed, and the response speed and cross-scene adaptive capacity of a system on edge equipment are improved.

The invention discloses a multi-modal large model training data acquisition method and system, and relates to the technical field of data acquisition optimization, and the method comprises the steps: building a causal graph adjacency matrix based on a cleaning alignment data set, carrying out the anti-fact intervention after recognizing a prejudice variable, and generating an anti-fact sample set; combining the anti-fact sample set and the cleaning alignment data set into an enhanced data set; performing cross-modal anti-long-tail compensation on the enhanced data set to obtain a balanced data set; based on the image data and the text data in the balance data set, depth separable convolution feature extraction and BERT semantic coding are carried out respectively, cross-modal fusion is carried out, and fusion features are output; through the steps of depth separable convolution, BERT coding, quantum latent variable evolution and cross-modal semantic verification, the generalization ability, robustness and social adaptability of the multi-modal large model in a complex and real scene are significantly improved.

The invention belongs to the field of intelligent transportation, and relates to a future feature enhanced vehicle trajectory prediction generative adversarial method, which comprises the following steps: a conditional information learning step: completing joint modeling of historical and future features; a generative adversarial training step: performing multi-modal trajectory prediction by adopting a CVAE-GAN hybrid architecture; wherein the CVAE-GAN hybrid architecture comprises an encoder, a generator and a discriminator; the encoder comprises a regression encoder and a prior encoder, and the regression encoder is responsible for mapping a real track to a submerged space, learning compact representation of the track and only being used during model training; the priori encoder performs network sampling to obtain a latent variable; the generator reconstructs a multi-modal future trajectory according to latent variables and condition information; the discriminator receives the trajectory and the condition information, and evaluates the authenticity of the trajectory and the consistency of the trajectory and the condition information. By designing a double-stage strategy, stable generation and optimization of multi-modal trajectory distribution are realized, and the problems that traditional GAN training is unstable and CVAE output is too smooth are solved.

The invention provides a virtual sample generation method and system based on a non-stationary neural network Gaussian process, and the method comprises the steps: obtaining original data, judging the non-stationarity, and judging whether a statistical characteristic changes with an input position or not; if not, a hidden variable model fusing the neural network and the Gaussian process is constructed, and a hidden variable space representing non-stationary distribution is obtained through training; sampling based on hidden variable space probability distribution; using the model to generate = (, Z) (Z, Z)-1X through non-stationary high-dimensional mapping, which is a non-stationary kernel function, Z is a low-dimensional hidden variable, and X is an input variable; checking the consistency of the high-dimensional virtual samples and the original data, and screening a virtual sample set meeting statistical consistency; according to the method, high-quality and high-diversity virtual samples can be generated in small sample and non-stationary scenes.

The invention belongs to the technical field of physical ability analysis and diagnosis, and particularly relates to a latent variable fusion modeling method and system for old people weakness trajectory prediction based on biomechanics. According to the method, the plantar pressure, the joint angle and the electromyographic signal in the gait process of the elderly are collected through wearable sensing equipment, multi-dimensional structured feature parameters are extracted, and a key index group reflecting the weakness degree is generated based on preset weight fusion. And in combination with time sequence fitting and variation trend analysis, dynamic evaluation and risk level judgment of the weakness state of the old people are realized. The system comprises a data acquisition module, a feature extraction module, an index fusion module, a trend analysis module and a result output module, has the characteristics of high accuracy, strong real-time performance, good interpretability and the like, and is suitable for weakness risk screening and rehabilitation intervention guidance.

A method for generating a principal predictor model from multi-dimensional time series data, comprises: receiving, from a plurality of sensors, multi-dimensional time series data corresponding to a plurality of original variables; transforming the multi-dimensional time series data to a lower dimension to define reduced-dimensional time series data; extracting, by a controller, a plurality of latent variables from the reduced-dimensional time series data and determining values of the plurality of latent variables in a first time period; initializing, by the controller, a loadings matrix corresponding to a set of latent variables of the plurality of latent variables; and determining, by the controller, one or more principal predictor model parameters, by performing an iterative process. The iterative process comprises (a) predicting values of the plurality of latent variables based on the reduced-dimensional time series data and the loadings matrix, by using an estimation process which maximizes a covariance between the values of the plurality of latent variables and the predicted values of the plurality of latent variables; (b) calculating a new loadings matrix from the loadings matrix and the predicted values of the latent variables; (c) updating the loadings matrix based on the calculated new loadings matrix; and (d) iteratively repeating (a) to (c) until the one or more principal predictor model parameters reach convergence.

This invention discloses a stepwise data assimilation method for a set subspace in a nonlinear inverse problem, relating to the field of data processing technology. The invention constructs an initial prior physical set based on the physical state variables to be inverted and optimized, and historical observation data. It extracts the static subspace basis anomaly matrix, initializes the latent variable set and particle weights, calculates fractional-step incremental reweighting, evaluates the current likelihood mismatch penalty using predicted data, updates and normalizes the particle weights in the logarithmic domain, obtains the effective sample number, performs system resampling operations in conjunction with a preset resampling tolerance coefficient, eliminates low-weight particles and replicates high-weight particles, synchronously updates the latent variable set and predicted data, executes a dynamic MCMC mutation loop to generate proposed latent state vectors and affinely maps them to a high-dimensional physical space, and updates the latent variable set by evaluating the annealing target energy within the latent variable subspace until all fractional steps are traversed. Finally, it outputs the latent variable set and maps it back to the physical space.

The invention belongs to the technical field of intelligent shield construction, and particularly discloses a shield tunneling machine posture multi-step prediction and deviation correction control method based on latent variable enhanced Free Transform. Firstly, shield construction process data are obtained, and a control quantity set is established and preprocessed; error features, change rate features and unit mileage features are constructed based on the preprocessed data, and a feature sequence is formed after normalization; organizing the sequence into a token sequence with a single time step, and inputting the token sequence into a pre-trained Free Transform model after position and mileage coding; the model aggregates complete sequence information through a non-causal encoder dedicated block, calculates latent variable distribution parameters based on a hidden state sequence, samples to obtain latent variables, maps the latent variables into conditional vectors and broadcasts the conditional vectors to a decoding path; and finally, outputting a future shield attitude vector, an over-threshold probability and a recommended control quantity in parallel by a multi-task output head. According to the method, working condition self-adaptive multi-step prediction and deviation correction are realized, and the safety and intelligent level of shield construction can be improved.

The invention discloses a bridge bearing capacity dynamic evaluation method based on multi-source monitoring data driving, and aims to solve the problem that the bearing evaluation precision is difficult to guarantee under a sparse sensing coverage condition. According to the method, through cross-modal attention alignment guided by event anchor points, mask codes influencing line and propagation time delay constraints and geometric position coding, Green function kernels and time-varying boundary flexibility latent variables are introduced into a physical constraint neural operator model, symplectic integral propulsion is adopted, and block shape-preserving interval estimation and order-preserving calibration weighted according to working conditions are adopted. According to the method, the full-bridge space-time response and the bearing utilization coefficient are calculated, the risk level and the load limiting suggestion are output, and the technical effects of reliability improvement, physical consistency and long-time energy stability evaluation under the complex working conditions and the low signal-to-noise ratio are achieved.

The invention discloses a hydroelectric generating set fault diagnosis data enhancement method based on a diffusion model and generative adversarial training, and belongs to the field of industrial equipment fault diagnosis. According to the method, the data enhancement model combining the diffusion model and the generative adversarial network is constructed, a multi-modal non-Gaussian distribution mechanism and a latent variable control generation process are introduced, the limitation of a single Gaussian hypothesis of a traditional diffusion model is broken through, and richer and more real sample generation is realized. And a condition discriminator and a time sequence modeling mechanism are introduced, so that the training of the model under different noise levels is more stable, and the authenticity judgment capability of the discriminator on the generated sample is enhanced. The problems that an existing hydroelectric generating set fault diagnosis system is insufficient in sample, low in generated sample quality and unreal in sample distribution are solved, and the diversity and quality of data are remarkably improved.

A method performed by a computing device for a radio network for configuration of a network device on which network or data energy can be collected while preserving specified conditions in the radio network is provided. The method includes receiving inputs to a conditional generative model. The inputs include the specified conditions in the radio network including a value for a predicted key performance indicator, KPI, constraint parameter for a time period, a target value for a optimization parameter, and a latent variable. The method further includes outputting from the conditional generative model a configuration data for a future time period for the network node or the cell of the radio network. The configuration data is bounded by the specified conditions including the predicted KPI constraint parameter, the target value for the optimization parameter, and the latent variable.

The invention discloses a robot action generation method integrating multi-layer feature bridging and world knowledge prediction. The method comprises the following steps: extracting multi-layer middle layer visual features and action query hidden variables by utilizing a pre-trained visual language model; an initialization strategy based on robot body sensing state guidance is adopted, real-time pose priori is injected for action query, and traditional all-zero initialization is replaced; a spatial perception vector gating mechanism is introduced into the bridging attention module, and fine-grained selection of specific image region features is achieved; meanwhile, a world knowledge prediction task is integrated, and physical common knowledge is enhanced through explicit modeling environment depth, semantics and a dynamic region; and finally, replacing a traditional L1 regression action head with a diffusion model architecture, and generating an optimal action sequence under multi-modal distribution through an iterative denoising process. According to the method, the operation precision of the robot in a long-range complex task is improved, and the problem of control failure caused by an action'averaging effect 'is solved while the light weight of the model is kept.

The invention relates to the technical field of multi-objective optimization algorithm verification and intelligent decision making, in particular to an algorithm verification method and application system based on multi-objective optimization, and the method comprises the steps: obtaining data, and forming a training data set; performing joint representation learning and decoupling on a Pareto solution set based on the data set to obtain independent latent variables of three dimensions of business stability, performance expression and environmental sensitivity; constructing simulation environment associated environment sensitivity information and disturbance parameters, generating a training scene sequence to drive an intelligent body to explore, and outputting a robustness evaluation result; learning an implicit preference function in combination with an implicit interaction sequence of a decision maker, recommending a candidate solution set and adjusting network parameters according to feedback; and establishing a knowledge graph to store historical data, and performing joint optimization on system parameters by extracting experience. According to the method, interpretability representation of the Pareto solution set, robustness verification in a dynamic environment and adaptive learning of decision preference are realized, and algorithm verification efficiency and decision reliability are effectively improved.

Apparatuses and methods are disclosed for encoding and decoding data. Techniques disclosed provide for the encoding of a data region of a frame. The encoding includes training an INR network to produce the data region from latent variables. And, determining distributions of the latent variables using respective contexts, constructed based on latent variables located within a shaped context region. Then, coding into a bitstream the latent variables based on the determined distributions and further coding into the bitstream the parameters of the trained INR network. Techniques disclosed also provide for decoding the data region. The decoding includes decoding from the bitstream the parameters of the INR network, determining the distributions of the latent variables using the respective contexts, and decoding from the bitstream the latent variables based on the determined distributions. Using the parameters of the INR network, the INR network produces the data region from the latent variables.

Systems and methods for providing a neural network system for time series forecasting are described. A time series dataset that includes datapoints at a plurality of timestamps in an observed space is received. The neural network system is trained using the time series dataset. The training the neural network includes: generating, using an encoder of the neural network system, one or more estimated latent variables of a latent space for the time series dataset; generating, using an auxiliary predictor of the neural network system, a first latent-space prediction result based on the one or more estimated latent variables; transforming, using a decoder of the neural network system, the first latent-space prediction result to a first observed-space prediction result; and updating parameters of the neural network system based on a loss based on the first observed-space prediction result.

This invention discloses a method and system for determining the propagation conditions of wood-eating insect populations based on data modeling. Specifically, it relates to the field of computer data modeling and simulation optimization technology. The method addresses the problem that existing data modeling methods for establishing propagation conditions of wood-eating insect populations suffer from drift in model parameter mapping relationships and poor reproducibility of output propagation conditions in subsequent batches due to neglecting latent variables such as cross-interface microbial communities. The method establishes an initial mapping model by acquiring environmental parameters and population response data from multiple historical batches. For the data in the initial stage of the current batch, a residual sequence is calculated, and a non-random structure is determined based on the monotonic change trend of the residual fluctuation amplitude with the substrate fermentation process. When a non-random structure exists, the partial dependence relationship between the residual sequence and each environmental factor is extracted along the drift trajectory of the fermentation process as a characterization of the latent variable effect. After concentration screening, the initial mapping model is corrected, and the corrected mapping model is then used to simulate and optimize the output propagation conditions adapted to the latent variable environment.

A machine learning model for reinforcement learning uses parameterized families of Markov decision processes (MDP) with latent variables. The system uses latent variables to improve ability of models to transfer knowledge and generalize to new tasks. Accordingly, trained machine learning based models are able to work in unseen environments or combinations of conditions / factors that the machine learning model was never trained on. For example, robots or self-driving vehicles based on the machine learning based models are robust to changing goals and are able to adapt to novel reward functions or tasks flexibly while being able to transfer knowledge about environments and agents to new tasks.

The invention relates to the technical field of industrial process soft measurement and quality control, and discloses a paper quality prediction method based on multivariate statistical latent variable fusion and space-time transformation, which comprises the following steps: acquiring space-time sequence data of a multi-source sensor in a papermaking process, constructing a space-time diagram structure reflecting a topological relation of equipment, and preprocessing. Then, multi-view latent variables are extracted through non-negative matrix factorization, independent component analysis and robust principal component analysis, attention fusion is conducted on the latent variables through an LV fusion module, and fusion latent variables are obtained; and inputting the fusion latent variable and original node data into a multi-scale convolution auto-encoder to obtain spatial feature embedding, and inputting the spatial feature embedding and the fusion latent variable into a space-time Transform module together to realize joint modeling of space correlation and time dependence. And finally, outputting a paper quality predicted value through a linear regression module. The method can achieve the accurate prediction of the paper quality under a high-dimensional and multi-noise working condition, and is suitable for online monitoring and modeling updating.

The invention discloses a representation vector optimization method, system and device for sequence data prediction and a medium, and particularly relates to the technical field of representation vector optimization. According to the method, a hidden variable is introduced into each variable, meanwhile, hidden state coding is carried out on original observation variables full of uncertainty, probability distribution obeyed by the hidden variables is estimated by using a neural network, and parameters of the distribution serve as more robust representation vectors of the variables; the method comprises the following steps: adaptively screening out an invariant feature code which is most stable in association with a predicted target from a variable sequence by using a gating mechanism so as to reduce the uncertainty of the invariant feature code; meanwhile, through an asynchronous loop optimization strategy, information gains provided by the node variables and the information variables in the information spreading process are continuously obtained, and representation information of the node variables and the information variables under information coupling is enhanced.

The invention provides a personalized psychological stress adaptive identification method based on artificial intelligence, and relates to the field of personalized stress identification. According to the invention, a variational instance adaptive pressure VIASTress framework is designed: a multi-modal domain generalization strategy is integrated in a multi-modal encoder, so that features extracted by the encoder have certain domain invariant characteristics, and pressure and non-pressure state features of different individuals are easier to distinguish. A variational instance adaptive strategy is integrated in a classifier, the strategy uses variational reasoning, an individual physiological signal instance and variational distribution to approximate individual physiological feature actual distribution, and latent variables sampled from variational distribution are used as a classifier weight adaptive basis; in this way, the model can generate decision boundaries suitable for the current individual when facing different individuals. The decision boundary is adaptively adjusted through combination of multi-modal domain generalization and variational instances, and the performance of individual pressure identification under the condition of distribution offset can be improved.

The invention discloses a spatial multi-omics data prediction method and model based on spatial constraint and adversarial learning, and application. The prediction method comprises the steps of modal specific coding and potential variable inference; carrying out spatial dependency modeling based on a sparse variational Gaussian process; modeling potential variables of spatial dependency constraints; carrying out multi-modal potential representation alignment based on adversarial learning; modal specific decoding and cross-modal feature prediction are carried out; and combining objective function construction and model training. According to the embodiment of the invention, space coordinate information is fully utilized to restrain potential representation, an adversarial learning mechanism is introduced to realize effective alignment of potential features of different modals, cross-modal feature prediction is realized on the basis, and the precision, stability and application value of space multi-omics data analysis are improved.

This invention discloses a dual-domain evolution prediction method for carbon mechanics of civil structures based on generative physics field inference. The steps are as follows: 1. Set an initial boundary condition set B within [0,T]; 2. Construct a bidirectional coupled control equation system P for carbonization diffusion and mechanical response based on B; 3. Construct a generative physics field inference model and embed it into P, achieving coordinated iteration of the concentration field C(x,t) and stress field σ(x,t) through spatiotemporal feature encoding and a dual-domain generation kernel, followed by dynamic feedback correction to convergence via a physical consistency regulator; 4. Construct a joint loss function including data fitting, physical equation residuals, energy constraints, and constitutive consistency, and perform two-stage hybrid optimization training to convergence; 5. Discretize the time domain into N layers, performing dual-domain evolution recursion, latent variable residual self-repair, and multi-scale stability control at each step, outputting field distribution, carbonization depth, and stiffness degradation indices. This invention deeply integrates the generative model with the bidirectional coupled equations, achieving unified convergence of data-driven and physical constraints.

This invention discloses a lightweight modeling method, apparatus, system, and storage medium for high-dimensional data meteorological forecasting tasks, belonging to the technical field of computation, extrapolation, or counting. The method constructs prediction vectors and target variables from the original high-dimensional data required for the target meteorological forecasting task. Through an improved variational autoencoder strategy, it reconstructs latent predictor variables that influence or potentially influence the target variables from the prediction vectors. It then selects latent variables highly correlated with the original predictor variables to construct a candidate predictor variable dataset. After removing candidate variables that are significantly uncorrelated with the target variables and those with strong linear correlations from the candidate predictor variable dataset, it performs key predictor variable selection and constructs a lightweight prediction model based on the key predictor variables. This invention significantly reduces the dimensionality of the model input data while maintaining prediction accuracy, and enhances the model's prediction accuracy by constructing predictor variables that influence the target variables and have clear physical meaning.

One embodiment sets forth a technique for creating a generative model. The technique includes generating a trained generative model with a first component that converts data points in the training dataset into latent variable values, a second component that learns a distribution of the latent variable values, and a third component that converts the latent variable values into output distributions. The technique also includes training an energy-based model to learn an energy function based on values sampled from a first distribution associated with the training dataset and values sampled from a second distribution during operation of the trained generative model. The technique further includes creating a joint model that includes one or more portions of the trained generative model and the energy-based model, and that applies energy values from the energy-based model to samples from the second distribution to produce additional values used to generate a new data point.

The application provides a method for generating imperceptible three-dimensional point cloud adversarial samples, comprising the following iterative optimization process: based on the semantic gradient feedback of a target model to a current point cloud and the geometric attribute of the point cloud, the normalized saliency score of each point in the point cloud is dynamically determined through a learnable saliency prediction model; through a mapping rule based on a quantile threshold and a sigmoid gating function, a point-by-point continuous weight for modulating disturbance is adaptively assigned; a learnable disturbance latent variable is modulated by using the point-by-point continuous weight to generate an adversarial point cloud of the current iteration; the total loss function comprises a classification loss term for guiding error classification, a structure preservation loss term for limiting the amplitude of disturbance, and a distance loss term for maintaining geometric consistency; based on the total loss function, the parameters are synchronously optimized through gradient backpropagation to realize end-to-end differentiable training; the above process is repeated until a preset termination condition is met, and the final three-dimensional point cloud adversarial sample is output.

A meta continual-learning and inferring method uses an implementation of Bayes' theorem, and includes: calculating a likelihood of learning data for a given latent variable by a data distribution learner; performing a sequential Bayesian update and calculating a final posterior distribution of the latent variable by using prior distribution of the latent variable and the calculated likelihood by a Bayes' calculator; sampling the latent variable from the final posterior distribution; and inferring test output data based on the sampled latent variable and test input data by an inference engine, wherein respective meta parameters of a neural network of the data distribution learner, the prior distribution of the latent variable of the Bayes' calculator, and a neural network of the inference engine are trained by a meta learning.

This invention relates to the fields of medical data analysis and artificial intelligence, proposing a dual-latent variable decoupled data analysis method based on plasma Netrin-1 and clinical scale features. First, plasma Netrin-1 test data and clinical characteristics such as UPDRS scale scores and disease duration are acquired from the subjects. Batch effect correction is applied to the raw test values, and an input feature vector is constructed. Then, a variational autoencoder model is established. By setting orthogonal constraints, Netrin-1 regression constraints, and disease duration regression constraints, the latent space is structurally constrained, enabling the model to learn latent variables D representing disease progression and C representing adaptive changes during training. Based on this, an efficiency index η composed of latent variables C and D is calculated, and relevant trend indicator parameters are obtained by combining longitudinal follow-up data. This method can achieve decoupled representation of disease-related features and adaptive change features in multi-source clinical data, providing a new data analysis tool for statistical analysis and research of neurodegenerative disease-related data.

The invention provides a group epigenetic analysis method and system based on consensus peak recognition, and belongs to the technical field of bioinformatics, and the method comprises the steps: S1, recognizing a consensus peak from a multi-sample peak file based on a position offset threshold and a minimum sample proportion; s2, calculating the coverage depth of a consensus peak based on the segment / read segment to obtain an original peak intensity matrix; s3, eliminating GC preference through GC content inter-partition local correction; s4, performing cross-sample normalization by adopting methods such as DESeq2, TMM and the like; and S5, automatically removing a hidden batch effect by utilizing potential variable analysis. The system correspondingly comprises five functional modules. According to the method, the problems of inconsistent peak coordinates, incomparable signals and large technical deviation interference in multi-sample epigenetic data are solved through an integrated process, full-automatic generation from original data to a high-quality and high-comparability peak intensity matrix is realized, and the analysis efficiency and accuracy of researches on large-scale groups ChIP-seq, ATAC-seq and the like are remarkably improved.

The application provides a multi-task-oriented variational self-encoding low-rank adapter optimization method, which comprises the following steps: selecting a pre-trained model as a base model and initializing a weight matrix; explicitly defining a plurality of training tasks and determining training data sets of the training tasks; initializing parameters of a plurality of low-rank adapter modules, a variational self-encoder and a dynamic router; obtaining an input vector after preprocessing the training data sets; converting the input vector to obtain a fourth output vector; and constructing a loss function to optimize parameters of the low-rank adapter and the variational self-encoder. The application has the beneficial effect of effectively solving the problem of limited low-rank feature expression capability faced by the multi-task learning method based on LoRA at the present stage, thereby effectively improving the performance of multi-task learning of a language model; a normal latent distribution is constructed to constrain a latent variable, avoid feature disorder and improve the stability of coding, and the original task features of the first matrix are reserved through residual connection.