847 results about "Auto encoders" patented technology

A system and method for multi-scale temporal attention processing in multimodal technology deep learning systems. This system processes time-series, textual, sentiment, and structured tabular data across three hierarchically-organized temporal streams—quarterly, weekly, and intraday levels—with bidirectional cross-temporal information flow. Scale-specific attention mechanisms are optimized for respective temporal granularities, while an adaptive controller dynamically weights each temporal level based on real-time market volatility indicators. A multi-scale fusion processor integrates attention-weighted representations to generate temporally unified representations preserving both short-term market dynamics and long-term trends. This approach enables superior forecasting and risk assessment by leveraging temporal correlations across multiple time scales while automatically adapting to changing market conditions. The system facilitates interpretable AI analysis through attention visualization and enables synthetic scenario generation for model testing.

A method is provided for enhanced data routing in neural networks using Self-Organizing Maps (SOM) integrated with Autoencoder-GAN. The method comprises training an autoencoder to encode input data into a latent space representation; applying a Self-Organizing Map (SOM) to organize the latent space representation into a topological map; refining the latent space representation using a Generative Adversarial Network (GAN), wherein the generator generates enhanced latent space representations and the discriminator evaluates their quality; using the refined latent space representations to update the SOM topology dynamically; generating routing coefficients based on the updated SOM topology to guide data routing in a capsule network; and dynamically adjusting routing within the capsule network using the generated routing coefficients to enhance performance based on the refined latent representations.

A system and method for adaptive real-time multi-modal compression with dynamic resource allocation provides intelligent compression optimization based on continuously monitored device conditions. The system monitors battery level, CPU utilization, and memory availability while classifying incoming multi-modal data streams comprising image, audio, text, and sensor data to determine processing priorities. Multi-objective optimization balances compression efficiency, reconstruction quality, and energy consumption using evolutionary algorithms that generate optimal parameters for an adaptive variational autoencoder. The autoencoder features dynamically selectable processing complexity, adjustable latent space dimensionality, and modality-specific processing layers. The system automatically switches between operational modes including emergency mode triggered by resource constraints, which applies maximum compression settings and intelligent data triage. Continuous learning adapts compression parameters based on observed performance outcomes, improving future optimization decisions. The system enables homomorphic operations on compressed data and provides enhanced compression performance under varying resource constraints across diverse edge computing applications.

A computer-implemented method of predicting quality of a food product sample after a mixing process, based on properties of the food product, comprises: building a hybrid model by: training an autoencoder in an unsupervised learning step using historical process data of food product samples; training a supervised model in a supervised learning step using the output of the autoencoder; and predicting the quality of the food product by inputting process data of current samples into the hybrid model and classifying the samples.

The invention discloses an intelligent evaluation method for real-time learning and teaching effect feedback, and the method comprises the steps: carrying out the preprocessing of collected multi-modal teaching scene data of a target classroom, and obtaining a multi-modal data stream which comprises the aligned target video data, target audio data and a timestamp; pre-training the visual basic model based on the ViT architecture by adopting a plurality of education scene data sets to obtain a first model, carrying out alignment fusion on the first model and a preset variational automatic encoder structure to obtain a preset target model, and carrying out dimension reduction processing on the extracted spatial-temporal features and feature vectors by utilizing the model to obtain target features; and determining a classroom atmosphere index, a student participation index and a teaching interaction effect index based on teaching behavior characteristics, student behavior characteristics and teaching interaction behavior characteristics identified from the target characteristics, and synchronously generating a real-time teaching quality evaluation report. According to the method, the classroom information can be accurately and comprehensively analyzed, and a real-time and objective evaluation result is synchronously generated.

A neural text-to-speech system may be configured to emphasize words. Applying emphasis where appropriate enables the TTS system to better reproduce prosodic characteristics of human speech. Emphasis may make the resulting synthesized speech more understandable and engaging than synthesized speech lacking emphasis. Emphasis may be manually annotated to, and / or predicted from, a source text (e.g., a book). In some implementations, the system may use a generative model such as a variational autoencoder to generate word acoustic embeddings indicating how emphasis is to be reflected in the synthesized speech. A phoneme encoder of the TTS system may process phonemes to generate phoneme embeddings. A decoder may process the word acoustic embeddings and the phoneme embeddings to generate spectrogram data representing the synthesized speech.

Providing channel state information, CSI, in a wireless communications network by a first node to a second node. The first node has access to one or more trained Neural Network, NN,-based Auto Encoder, AE,-encoder models for encoding the CSI and the second node has access to one or more trained NN-based AE-decoder models for decoding the CSI provided by the first node. The method includes receiving, from the second node, an indication of an NN-based AE-decoder model out of the one or more trained models; selecting a trained model out of the one or more trained models to use for the NN-based AE-encoder based on the received indication of the NN-based AE-decoder model such that the selected trained model is compatible with the indicated NN-based AE-decoder model; and transmitting (604) the CSI to the second node based on output from the selected trained model.

A neural network system is provided. The system includes an autoencoder configured to encode input data into a latent space representation; a generator neural network configured to receive a noise vector and the latent space representation and output a set of routing coefficients; a discriminator neural network configured to evaluate the effectiveness of the routing coefficients by measuring the performance of a capsule network utilizing said routing coefficients; and a capsule network comprising a first capsule layer and a second capsule layer, wherein the routing coefficients are used to dynamically route outputs from the first capsule layer to the second capsule layer.

The invention relates to the technical field of data processing, and discloses an anti-oxidation barrier layer performance degradation prediction method and system based on multi-source data. The method comprises the following steps: collecting and preprocessing multivariate data of a barrier layer environment, and constructing three-dimensional tensor data; indexes such as oxygen blocking efficiency and anti-permeability performance are calculated, and performance time sequence data are obtained; extracting multi-dimensional features and fusing the multi-dimensional features through an automatic encoder; applying a hybrid deep learning model to predict performance degradation; analyzing degradation curve characteristics, and executing mode clustering to obtain a risk matrix; and optimizing a maintenance decision scheme based on risk assessment. According to the invention, the hybrid deep learning model is utilized to capture the dependency relationship of time and space dimensions at the same time, and the prediction precision is significantly improved; based on degradation mode identification and risk level evaluation, accurate multi-scene maintenance decision is realized, environmental risk is reduced, and maintenance cost is optimized.

A system and methods for upsampling compressed data using a jointly trained Vector Quantized Variational Autoencoder (VQ-VAE) and neural upsampler. The system compresses input data into a discrete latent space using a VQ-VAE encoder, reconstructs the data using a VQ-VAE decoder, and enhances the reconstructed data using a neural upsampler. The VQ-VAE and neural upsampler are jointly trained using a combined loss function, enabling end-to-end optimization. The system allows for efficient compression and high-quality reconstruction of various data types, including financial time-series, images, audio, video, sensor data, and text. The learned discrete latent space can be explored and manipulated using techniques such as interpolation, extrapolation, and vector arithmetic to generate new or modified data samples. The system finds applications in data storage, transmission, analysis, and generation across multiple domains.

A system and method for deep learning using a large codeword model with homomorphically compressed and dyadically encrypted data is disclosed. The system preprocesses input data, applies homomorphic-dyadic compression and encryption, tokenizes the compressed data into sourceblocks, and assigns codewords using a codebook. These codewords are processed through a machine learning core, which can be either a conventional transformer-based architecture or a latent transformer core utilizing a variational autoencoder. The system enables secure operations on encrypted data, preserving privacy while allowing complex computations. The processed output is decrypted, decompressed, and translated to match the input modality. A neural upsampler may further enhance the output. The machine learning core is continuously trained using the processed data and additional training data, improving performance over time.

The invention relates to the technical field of RFID smart tags, and discloses an RFID smart tag positioning and tracking printing method and system, and the method comprises the steps: inputting an RFID tag radio frequency signal into a conditional variation automatic encoder for convolution processing and attention weighting, and obtaining an RFID tag global feature vector; establishing a layered sensor node network to perform multi-dimensional acquisition on environmental parameters to obtain an environmental spatio-temporal data set; inputting the global feature vector of the RFID tag and the environment spatio-temporal data set into a deep fusion network for feature fusion to obtain a multi-modal fusion feature; performing sequence modeling based on the multi-modal fusion features to obtain an article trajectory prediction model; the object track prediction model is applied to a target area grid coordinate system for density distribution calculation, a real-time object positioning heat map and a track database are obtained, an efficient object positioning heat map generation and track data management mechanism is established, and real-time position visualization and historical track query are facilitated for the system.

The invention discloses a space transcriptomics cell type deconvolution method based on a generative adversarial network and a comparative learning framework. The method comprises the following steps: firstly, enhancing a gene expression matrix by using a histological image and spatial position information; then, in combination with GraphSAGE and self-supervised comparative learning, optimizing feature representation of space transcriptomics data, and reconstructing high-quality space gene expression data; single-cell RNA sequencing data gene expression features are extracted by using a variational automatic encoder, and meanwhile, the quality of a generated sample is optimized by using a discriminator, so that efficient feature extraction and accurate gene expression reconstruction are realized; and finally, mapping single-cell RNA sequencing data into spatial points through comparative learning and mapping matrix learning based on low-rank representation, thereby predicting the cell type of each point. According to the model, multi-modal information in single-cell RNA sequencing and space transcriptomics data is comprehensively considered, and a mapping matrix is constructed and optimized, so that the cell type of each space point is predicted.

The invention belongs to the technical field of artificial intelligence, and discloses a PCB use fault early warning system based on artificial intelligence, which comprises a data acquisition module, a data processing module, an intelligent analysis module, a decision level fusion module, a self-adaptive modeling module, a multi-model cooperation module and a fault early warning module. The intelligent analysis module realizes double breakthrough of nonlinear feature capture and adaptive anomaly discrimination ability through a dynamic error threshold mechanism of an LSTM time sequence prediction engine and a depth autoencoder; an XGBoost-1DCNN hybrid classifier is constructed, a gradient boosting tree and multi-scale convolution features are fused in fault mode recognition, and the complex fault classification precision is remarkably improved; and the decision level fusion module constructs a multi-model decision conflict resolution mechanism based on an improved D-S evidence theory, and realizes great optimization of a false alarm rate through a confidence interval dynamic synthesis algorithm, thereby forming a closed-loop system with real-time response, multi-dimensional root cause analysis and intelligent hierarchical early warning.

The present disclosure provides a method for configuring a learning model for music generation and the corresponding learning model. The method includes training a masked autoencoder with training data comprising a combination of a reconstruction loss over time and frequency domains and a patch-based adversarial objective operating at different resolutions. An omnidirectional latent diffusion model is trained based on music data represented in a latent space to obtain a pretrained diffusion model. The pretrained diffusion model is fine-tuned based on text-guided music generation, bidirectional music in-painting, and unidirectional music continuation. The method enables high-fidelity music generation conditioned on text or music representations while maintaining computational efficiency.

The subject technology provides frequency-aware masked autoencoders for multimodal pretraining on frequency-based signals. An apparatus receives input data comprising frequency-based signal information associated with one or more modalities. The apparatus transforms the input data from a time domain to a frequency domain. The apparatus generates a frequency-embedded latent representation of the input data comprising time-domain and frequency-domain information. The apparatus also generates a masked frequency-embedded latent representation by masking one or more frequency components in the frequency-embedded latent representation. The apparatus produces a trained machine learning model by training a neural network to predict one or more masked frequency components of the frequency-embedded latent representation.

The invention relates to the technical field of fault detection, and particularly discloses a prediction reconstruction framework causal perception space-time network for explaining anomaly monitoring in a complex industrial process, comprising the following steps: S01, constructing graph data E (V) and a causal graph; automatically adjusting the fusion proportion of the time-frequency characteristics according to the data characteristics so as to ensure that the model can comprehensively capture the information of the data in the time domain and the frequency domain; secondly, introducing a residual image attention network (RGAT), and converting the image data E (V) into image structure data G (S (V), E (V)); and S03, reconstructing a prediction error by adopting a variational automatic encoder (VAE), learning an error mode of normal data, providing an anomaly judgment AD (V) for anomaly detection, analyzing a causal relationship between data in combination with a causal graph, and positioning an anomaly reason according to an anomaly score, so as to form a prediction result. The network solves the problem that a traditional monitoring network is high in false alarm rate.

The invention belongs to the field of network security of smart grids, and discloses a method for detecting an antagonistic network attack in a smart grid based on an iterative algorithm, and the method comprises the steps: generating an antagonistic attack sample through iteration, so as to avoid the protection of a detection system; performing feature extraction and reconstruction of input data by using the constructed automatic encoder network model, and capturing complex time sequence features through convolution, pooling and up-sampling operations; a dynamic parameter adjustment strategy is adopted to adaptively optimize the training optimization momentum coefficient and the training optimization target reconstruction error in the training process, and the system robustness is improved; and finally, verifying the concealment and stability of the resistance attack sample under different scaling conditions through multi-scale evaluation so as to optimize the automatic encoder network model. According to the invention, the security protection level of the smart grid system in a complex network attack scene is improved, and stable operation of the power system is ensured.

Communication devices and methods for determining channel state information (CSI) report based on artificial intelligence (AI) / machine learning (ML) are provided. The method for determining CSI report based on AI / MI performed by a communication device includes determining, by the communication device, one or more CSI reports according to an AI / ML based CSI feedback, wherein each of the one or more CSI reports contains an output of an auto-encoder, a compression ratio, a rank indicator, quantization levels, a ground truth of an enhanced CSI feedback, and / or an ML model monitoring outcome, and determining, by the communication device, priority rules for the CSI reports according to the AI / ML based CSI feedback.

The invention belongs to the technical field of artificial intelligence and information security crossing, and discloses a multi-modal false news detection method based on decoupling and cross-modal clue mining, and the method comprises the steps: employing a pre-training comparison model to extract text and image features in news, and mapping the text and image features to a unified embedding space; designing a multi-modal decoupling automatic encoder, decomposing each modal feature into modal common and modal specific representation, and obtaining a pure decoupling feature through confrontation constraint and reconstruction loss; a cross-modal clue mining module is constructed, modeling complementation is carried out from non-unhooked features, and conflicting semantic clues are enhanced; a self-adaptive attention aggregation module is introduced, and the modal features and the semantic clues are dynamically weighted and fused; and inputting the aggregated representations into a classifier for false news identification. According to the method, through modal decoupling and multi-source clue modeling, modal redundancy and specific information are effectively separated, association and conflicts between images and texts are deeply mined, and the detection accuracy is remarkably improved.

The invention discloses a fishery resource dynamic trend prediction system and method based on time sequence analysis. The method comprises the following steps: S1, constructing a fishery resource time sequence data set; s2, forming a standardized fishery resource time sequence data set; s3, constructing a self-supervised learning model comprising an automatic encoder module and a contrast learning module by using the fishery resource time sequence data set; s4, screening high-correlation features to construct a feature space; s5, constructing a time sequence dynamic prediction model taking a self-attention mechanism as a core based on the feature space; and S6, generating prediction results of the spatial position of resource distribution, the resource abundance change trend and the time range of the peak period and the valley period. According to the method, accurate prediction of the dynamic trend of the fishery resources is realized, and powerful technical support is provided for sustainable management of the fishery resources.

The invention relates to the technical field of wire harness fault diagnosis, and discloses a vehicle-mounted wire harness fault diagnosis method and device and a storage medium. The method comprises the following steps: carrying out equivalent circuit modeling on a measurement node and a wire harness section of the vehicle-mounted wire harness to obtain a node connection relation matrix; performing feature analysis on the voltage data and the current data of the measurement nodes to obtain a fault feature vector and a fault type discrimination rule base; inputting the voltage data, the current data and the node connection relation matrix into a recursive variational graph automatic encoder for data reconstruction to obtain time sequence feature reconstruction data and topological relation reconstruction data; carrying out distributed sub-model division and model fusion to obtain a standardized feature vector; and performing matching analysis on the standardized feature vector and the fault type judgment rule base to obtain fault position information and fault type information, so that automatic optimization and screening of fault features are realized, and the reliability and interpretability of a diagnosis result are improved.

The invention relates to the technical field of network information security, discloses an information security management system based on big data, and aims to solve the defects of an existing system in the aspects of data acquisition, anomaly detection, threat prediction, risk assessment and the like. The system comprises a data acquisition module for collecting various types of data in real time; the abnormal detection model is used for identifying normal and abnormal behaviors by adopting an automatic encoder AE algorithm; the threat prediction module is used for predicting a future security threat type and probability by using a recurrent neural network (RNN) algorithm; the risk assessment module comprehensively assesses the system security risk through a random forest RF algorithm; and the response decision module is used for executing corresponding safety response measures according to the evaluation result. According to the system, the big data technology is utilized, comprehensive monitoring and accurate prediction of the network security state are achieved, the efficiency and accuracy of information security management are improved, and powerful support is provided for coping with novel and complex network attacks.

Provided is a method for token-based representation and learning of robotic perception data based on a graph neural network, comprising: obtaining a plurality of types of perception data of a robot; performing token-based representation according to types of the plurality of types of perception data; constructing an initial feature graph based on the plurality of types of perception data after the token-based representation; learning a compact representation of the initial feature graph based on an autoencoder and reconstructing a graph structure; after the autoencoder completes learning of the graph structure, fixing the graph structure; and converting the plurality of types of perception data into node feature vectors, constructing a feature graph based on the graph structure, and performing numerical encoding on each of the node feature vectors by utilizing the graph neural network to obtain a representation of high-dimensional feature vectors of the plurality of types of perception data.

The invention discloses an auxiliary film reading method and system based on artificial intelligence, and the method comprises the steps: 1, collecting a pathological WSI, an electronic medical record, detection data and equipment parameters, correcting the equipment difference through adaptive dyeing normalization, and constructing a structured data package associated with an ID-timestamp of a patient; 2, developing a dynamic branch CNN, migrating teacher model knowledge through knowledge distillation, and introducing federated learning; step 3, the edge generates a thermodynamic diagram to mark a suspicious area, and the cloud outputs a structured report; step 4, constructing a normal tissue feature space by the variational auto-encoder, detecting abnormal slices and triggering expert re-checking; a reverse automatic encoder generates a pseudo-health image to compare and position a pathological area, and dynamic weight adjustment balances the federal learning convergence speed; 5, integrating the thermodynamic diagram, the gene data and the clinical indexes by a three-dimensional platform, and supporting multi-dimensional superposition display; webGL realizes browser end rendering, and NLP automatically generates a report abstract marked with a key evidence chain and is in butt joint with an international diagnosis and treatment guide.

Autoencoder models may be used in the field of lithography to estimate, infer or predict a parameter of interest (e.g., metrology metrics). An autoencoder model is trained to predict a parameter by training it with measurement data (e.g., pupil images) of a substrate obtained from a measurement tool (e.g., optical metrology tool). Disclosed are methods and systems for synchronizing two or more autoencoder models for in-device metrology. Synchronizing two autoencoder models may configure the encoders of both autoencoder models to map from different signal spaces (e.g., measurement data obtained from different machines) to the same latent space, and the decoders to map from the same latent space to each autoencoder's respective signal space. Synchronizing may be performed for various purposes, including matching a measurement performance of one tool with another tool, and configuring a model to adapt to measurement process changes (e.g., changes in characteristics of the tool) over time.

The invention discloses an image editing method and system based on diffusion model inversion and attention optimization, and relates to the technical field of image editing, and the method comprises the steps: mapping a source image to a potential space through a pre-trained automatic encoder, and obtaining an initial noise feature; an EF noise space inversion algorithm is adopted to process the initial noise features, a high-variance noise graph is generated, and an intermediate result under each time step is obtained through a decoder; constructing an improved U-Net network based on edit-friendly feature reweighting and an edit-friendly attention mechanism, fusing the target prompt information into a denoising process of the improved U-Net network through a cross attention mechanism, and performing feature optimization based on the improved U-Net network; and reconstructing the potential features through a U-Net decoder, and generating an edited image conforming to the target prompt information. On the low-cost premise that complex model fine tuning and retraining are not needed, the method is superior in text-guided controllable editing tasks, and the image generation quality is improved.

The invention relates to the technical field of air target prediction, in particular to an air target trend prediction method, which comprises the following steps: S1, multi-source heterogeneous data adaptive fusion filtering processing; s2, manifold learning is constructed in the high-dimensional spatial-temporal feature space; s3, performing semantic modeling on the dynamic behavior pattern recognition intention; and S4, multi-dimensional threat situation assessment warfare area modeling is carried out. According to the method, high-precision space-time synchronization and noise suppression of radar sensor data, infrared sensor data and other sensor data are achieved through the multi-source heterogeneous data self-adaptive fusion filtering technology, target micro-Doppler features are effectively reserved, missing data are repaired, and through the combination of third-order Savitzky-Golay differential filtering and short-time Fourier transform, high-precision space-time synchronization and noise suppression of radar sensor data and infrared sensor data are achieved. An 18-dimensional compression feature space containing kinematics and electromagnetic characteristics is constructed, a nonlinear topological relation is reserved through t-SNE and an automatic encoder, the signal-to-noise ratio and feature expression capacity of original data are remarkably improved through the function, and a high-precision and low-redundancy input basis is provided for follow-up behavior recognition and prediction.

The invention provides a battery health state prediction method, system and device and a medium, and belongs to the field of battery state estimation, and the method comprises the following steps: obtaining original electrochemical impedance spectroscopy (EIS) data of a battery; inputting the original EIS data into an automatic encoder, compressing the original EIS data into a low-dimensional vector by using the automatic encoder, and segmenting the low-dimensional vector into a plurality of time sequences; and inputting the plurality of time sequences into a domain adaptive neural network DANN, performing feature extraction on the time sequences by using the DANN to obtain domain invariant features, and predicting the health state of the battery according to the domain invariant features. According to the method, dimension reduction processing is carried out on the original EIS data through the automatic encoder, the nonlinear features in the original EIS data are extracted, the domain invariant features can be extracted in combination with the domain adaptive neural network DANN model, and the generalization ability of the model is improved.

The invention relates to an offshore wind power prediction method, and aims to improve the accuracy and reliability of prediction. The method comprises the following steps: (1) a data preprocessing stage: detecting an abnormal value in data by using a DBSCAN clustering algorithm, reconstructing the abnormal value by using a KNN interpolation method, detecting time sequence abnormity through an LSTM automatic encoder, and performing regression reconstruction by using LSTM to ensure the retention of time sequence features; (2) a feature engineering stage: screening out key features through correlation analysis, generating a label column through K-means clustering and wind direction classification, and extracting features such as wind direction change rate, time periodicity and wind speed interaction; and (3) a model construction stage: constructing a composite model in combination with multi-scale convolution (Inception), a bidirectional long short-term memory neural network (BiLSTM) and a multi-head self-attention mechanism, extracting local features through a convolution layer, capturing a time dependency relationship through a bidirectional LSTM layer, enhancing the attention of key features by using the multi-head self-attention mechanism, and finally realizing high-precision wind power prediction.