197 results about "Auto encoders" patented technology

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 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.

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.

In various examples, self-supervised learning may be used to pre-train an encoder network of a masked prediction model to reconstruct masked regions of an input representation of 3D detections such as LiDAR point cloud(s). Spatial and / or temporal masking may be applied to a projected representation of 3D detections (e.g., a two-dimensional (2D) projection image), and the masked prediction model (e.g., a masked auto-encoder or joint-embedding predictive architecture) may be used to reconstruct a representation of the masked regions (e.g., reflection characteristic(s) stored in corresponding pixels or cells of the projected representation, a latent representation of the reflection characteristic(s)) during iterations of self-supervised learning. As such, the pre-trained encoder network of the masked prediction model may be used as a foundation model and fine-tuned with a task-specific output head or its pre-trained weights may be used to initialize a task-specific model.

A system is provided for dynamic analysis and verification of smart contracts. The system includes an autoencoder configured to preprocess smart contract code to reduce noise and highlight critical features; a capsule network configured to analyze the preprocessed smart contract code, capturing hierarchical relationships and dependencies within the code; a generative adversarial network (GAN) configured to generate optimal routing coefficients for the capsule network, enhancing the efficiency and accuracy of the analysis; and a blockchain-based platform for deploying and executing smart contracts, wherein the platform utilizes the capsule network to continuously monitor the smart contracts for anomalies during execution.

The invention provides a construction and use method of a potential diffusion model for SAR image super-resolution. The construction and use method comprises the steps of obtaining an original SAR image and inputting the original SAR image into a real degradation model to generate a degraded SAR image; inputting the degraded SAR image into an automatic encoder to generate a structure-enhanced submerged space feature map; and inputting the structure-enhanced latent space feature map and the degraded SAR image into a potential diffusion model to generate an SAR super-resolution image. The method has the beneficial effects that a two-stage training strategy is adopted, different optimization targets are focused in stages, the training efficiency is improved, and meanwhile, the learning ability of the model to SAR image features is enhanced; sAR imaging key degradation factors are comprehensively covered, so that a generated low-resolution sample is closer to a real scene, high-quality data support is provided for model training, and model learning is prevented from being separated from an actual degradation rule; sAR specific interference such as speckle noise is effectively simulated, and the anti-noise training effect of the model is enhanced.

The invention discloses a GRU-based vehicle-mounted CAN intrusion detection method and system, a storage medium and a computer system. According to the method, an automatic encoder (AE) is introduced to deepen the understanding of a model on input sequence characteristics, a sliding window is used for selecting batch CAN data to be preprocessed to obtain 13-dimensional time sequence data, and a scalar value within the range of [0, 1] is obtained through processing of the encoder, a GRU, a decoder, a full connection layer and a sigmoid activation function and used for classification of abnormal data. The Conv1D is used as a hidden layer, and compared with two-dimensional convolution, the one-dimensional convolution parameter quantity is smaller, and the calculation is simpler and more convenient. An attack message and a normal message can be completely distinguished, the precision and the accuracy rate reach 100%, and the precision in Fuzz detection is 0.9983; compared with the prior art, the method has high accuracy and reliability in the aspect of intrusion behavior detection, can effectively identify most intrusion events, and can keep a relatively low overall error rate, so that good balance between safety and availability is realized.

The invention relates to an aero-engine fault diagnosis method and device based on tensor decomposition and a medium, and the method comprises the steps: collecting the gas path parameter time sequence data of a plurality of parts of an aero-engine, and carrying out the time alignment and working condition label labeling; for the multi-parameter data of each component, a kernel principal component analysis and automatic encoder fused feature extraction method is adopted, and parameters of different components are unified into fixed dimension features; the feature matrixes obtained after feature extraction of all the components are combined, and a time-component-feature third-order tensor is constructed; training a high-order singular value decomposition model based on the engine health monitoring data; fault diagnosis is carried out by calculating the reconstruction error of the test data and the monitoring data; and fault positioning is realized by combining core tensor difference analysis. According to the method, the problem of feature fusion of engine multi-source heterogeneous data under variable working conditions is solved, and the engine fault detection sensitivity and positioning precision are improved.

The invention discloses a high polymer material performance prediction method and system based on deep learning, and the method comprises the steps: constructing an automatic encoder, and carrying out the dimension reduction of various data of a high polymer material through unsupervised learning; optimizing a multi-mode encoder structure by adopting an automatic design mechanism; extracting a fusion characteristic value of the high polymer material by using a multi-modal data encoder; dynamic attention fusion: introducing a dynamic gating weight to adaptively distribute modal weights to input data; introducing physical constraint, and embedding molecular dynamics into back propagation; performing quantum circuit acceleration graph convolution; and predicting and outputting, mapping the fusion characteristic value to the tensile strength and elastic modulus performance indexes of the high polymer material, and realizing nonlinear regression through a multi-layer perceptron. According to the method, a material molecular dynamics equation is converted into a forward propagation kernel from a posterior constraint, the dynamic behaviors of molecules can be simulated and predicted more accurately, and the calculation efficiency is improved while the precision is ensured.

The invention provides an intelligent abnormal value detection and processing method based on deep learning product quality data. The method comprises the following steps: collecting multi-dimensional product quality data from a production line; an encoder module is constructed, and an encoder adopts a multi-layer neural network structure and is used for compressing and mapping input high-dimensional quality data into low-dimensional hidden layer feature representation; constructing a decoder module, receiving low-dimensional representation features output by an encoder, and reconstructing original data through a reverse neural network structure; training the automatic encoder by adopting an unsupervised learning mode, and learning a distribution mode and feature representation of normal data; and based on the trained automatic encoder, performing reconstruction error calculation on the quality data input in real time, judging an abnormal value through a preset threshold value, and triggering an alarm. Through the steps, the problem that high-precision detection of product quality data in production cannot be achieved in the prior art is solved, the alarm function is achieved, and therefore the quality problem of products in the future is avoided.

The invention relates to a small sample training technology of a neural network. A neural network is trained to identify one or more features of the image. A neural network is trained using a small number of original images to obtain a plurality of additional images therefrom. Additional images are generated by embedding the rotated and decoded images in the potential space generated by the autoencoder. The image generated by rotation and decoding shows a change to the feature that is proportional to the amount of rotation.

An operator learning model generator is provided for training a smoothed operator learning model for predicting airflow dynamics in a room used by a controller connected to a heating, ventilation, and air conditioning (HVAC) system. The operator learning model generator includes an interface circuit configured to receive a training dataset via a network connected to a simulation computer, wherein the training dataset includes un-trajectories of airflow in the room for various time series of control actions applied to the HVAC system; a memory configured to store the smoothed operator learning model including an autoencoder and a neural ordinary differential equation, the training dataset, and training instructions for the smoothed operator learning model; and a processor configured to train the smoothed operator learning model stored in the memory, wherein the training instructions include jerk regularization that constrains smoothness of dynamics predicted by the smoothed operator learning model.

The application provides a multi-view clustering method and system based on a graph attention automatic encoder, relates to the technical field of multi-view clustering, and specifically includes the following steps: selecting a view with the largest information quantity from different views of the same group of nodes; learning a graph structure and node content by using a trained graph attention encoder based on the view with the largest information quantity and node content information, so as to obtain a node feature representation; performing specific constraint on the node feature representation by using an l1,2-norm penalty, so as to obtain a constrained node feature representation; inputting the constrained node feature representation into a self-optimizing clustering module to perform clustering, so as to obtain a final clustering result; and the application applies the graph attention network to multi-view graph clustering, simultaneously reconstructs the graph structure and the node content, and makes the latent representation well preserve the graph structure and the content information of the nodes, so that the application is more suitable for cluster tasks.

The application discloses an AMC scene migration method based on an adversarial domain adaptation, and belongs to the field of signal processing. Specifically, first, according to the flight characteristics of a UAV, a common sea and shore water area air-ground channel scene signal dataset of the UAV is simulated and constructed, wherein the sea scene is taken as a source domain, and the shore water area scene is taken as a target domain. Then, an adversarial domain adaptation model is established, signal features are extracted from generated signals via a feature extractor of the model, modulation features are obtained by an automatic encoder, and the modulation features are aligned by weighting two different features and using an adversarial training mode, so that the feature extractor extracts domain-invariant modulation features, and modulation mode recognition of the target domain signal under cross-scenarios is realized. The method of the application aims at the problem of performance decline of modulation mode recognition caused by domain difference, is based on the theory of transfer learning, minimizes the difference between domains by using the adversarial domain adaptation, and enhances the robustness of the modulation signal recognition method.

The invention provides a federal continuous learning method and system based on a variational automatic encoder. The federal continuous learning method based on the variational automatic encoder comprises the following steps: constructing a CNN-VAE-MLP model; the federal learning participation client sets initial parameters for the model, initializes a training data set and generates memory data; the federal learning participation client uses the data set and the memory data training model, and sends a training result to a server; the server updates a global CNN-VAE-MLP model according to the training result, and sends the updated global model to each federal learning participation client; the server updates the training information of the federal learning participating client according to the training result; the server judges whether the federal learning participating client should stop training according to the training information; and the federated learning participation client stores the global model and deletes the training data. According to the invention, through combination of the variational automatic encoder and federal learning, federal continuous learning is realized.

A system and a method for updating a Machine Learning (ML) model are described The method involves capturing reconstruction errors associated with reconstruction of images by a pre-trained autoencoder. Data points representing the reconstruction errors are clustered using affinity propagation. A preference value used by the affinity propagation for determining similarity between the data points is dynamically set through linear regression. Outliers and data drifts are determined from clusters of the data points. Classification output of the ML model is associated with the outliers and the data drift, for refinement of the ML model over a device hosting a training environment.

The embodiment of the present disclosure provides a kind of automatic driving test scene extraction method based on deep embedding clustering and medium, belong to data processing technical field, specifically include: to real traffic accident is deep accident deconstruction and obtains accident data, reconstructs the accident data collected, and extracts the pre-collision trajectory sequence of accident participant;Pre-collision trajectory sequence and static environment information are combined, and pre-collision trajectory matrix is created;With pre-collision trajectory matrix as input, dimension reduction is processed using heap auto-encoder, and low-dimensional embedding feature is obtained;Latent feature is used as input variable, clustering is carried out using k-mean algorithm, and multiple class feature cluster is obtained;According to the label in each feature cluster, backtrack pre-collision trajectory matrix, obtain the scene information of each cluster and summarize the scene commonality of each cluster, obtain the scene description corresponding to each typical accident class.By the scheme of the present disclosure, the implantability and adaptability of typical scene are increased.

The invention relates to the technical field of network information security, and particularly discloses a multi-domain network attack-oriented unsupervised integrated detection method and system. In the scheme, a multi-channel automatic encoder of an integrated network can design a differentiated structure according to feature dimensions and distribution rules of different attack domain data; the multi-channel design can accurately extract key features of data of each domain, representation deviation caused by inter-domain feature distribution conflicts is avoided, hidden space representation better fits essential features of attacks of each domain, and robustness is improved. The channel automatic encoder can realize effective information focusing through parallel dimension reduction and feature fusion: on one hand, each channel can adopt different dimension reduction paths to mine effective information from high-dimensional data in a multi-view manner, and on the other hand, the integrated network performs fusion processing on multi-channel dimension reduction results, local noise of each channel can be filtered, and the multi-channel dimension reduction result is obtained. The optimized hidden space representation has the characteristics of high information density and low noise interference at the same time, and the robustness and the detection precision can be improved.

The invention relates to an anticancer drug combination synergistic effect prediction method. Comprising the following steps: learning low-dimensional feature representation of drugs and cell lines through an automatic encoder of a self-attention mechanism; inputting the low-dimensional feature representation into a hybrid expert network, and constructing interactive feature representation of the drug and the cell line from multiple angles; through a self-attention gating network, multiple interactive feature representations constructed by a hybrid expert layer are efficiently fused, and enhanced interactive feature representations are generated; inputting the enhanced interaction feature representation into a prediction layer, introducing a weight coefficient to adjust a loss function, and realizing drug synergy and single-drug treatment sensitivity prediction; and through a self-attention gating network, feature representation is efficiently fused, and the internal relationship between the drug and cell line features is captured, so that the prediction precision of the synergistic effect of the anti-cancer drug combination is improved.

The invention discloses a community discovery method of a graph neural network based on multi-view information fusion, and belongs to the technical field of data mining. The method comprises the following steps: preprocessing data, performing PCA dimension reduction on an attribute matrix, and normalizing an adjacent matrix; inputting the correlation matrix combination into an automatic encoder and a graph attention automatic encoder to obtain a new matrix representation and fusing the new matrix representation into a final node representation matrix; then back-propagating the optimization model by using various loss functions; and finally, community division is realized by using a k-means clustering algorithm. According to the method, the structure and attribute information of the original data are fully utilized, multi-view information is effectively fused, the accuracy of community discovery is improved, and the method has great significance in the fields of research on citation networks, recommendation systems and the like.

Provided are a device and method for designing a material using deep learning. The method includes training a decoder which derives wave properties from given information on a material in advance, training an autoencoder including the decoder and an encoder which will be trained to derive information on a material from given wave properties, and inputting targeted wave properties to the trained encoder to acquire information on a material satisfying the input wave properties.

Systems and methods for property guided molecular optimization using artificial intelligence diffusion models. An equivariant continuous denoising diffusion implicit model autoencoder framework (DDIM-AE) can be trained (510) on a conformational dataset to predict raw data from data corrupted by a time-dependent noise to obtain a trained DDIM-AE that ensures controlled generation of threedimensional (3D) molecules. Linear optimization of semantic embeddings of 3D molecules can be performed (520) with a linear classifier to achieve a target property value from desired properties and obtain an optimized embedding. An optimized 3D molecule that includes molecular conformation with the desired properties while preserving interactions with biochemical molecules can be generated (530) from the optimized embedding with the trained DDIM-AE.

Embodiments of the present specification disclose a training method of a bounding box distribution model to solve the uncertainty problem of network prediction of an existing target detection model and improve the accuracy of target detection. The method comprises: obtaining a sample anchor box image set, each sample anchor box image comprising an initial anchor box of a target object region in a sample image and a corresponding real bounding box; inputting the sample anchor box image set into an initial bounding box distribution model for model iterative training until a convergence condition of the bounding box distribution model is met, to obtain a trained bounding box distribution model; the initial bounding box distribution model comprises a variational autoencoder, and each model iterative training of the bounding box distribution model comprises: obtaining a sample prediction image comprising a predicted bounding box corresponding to the initial anchor box by using the variational autoencoder; and adjusting model parameters of the bounding box distribution model according to the initial anchor box, the predicted bounding box, the real bounding box and a preset loss function of the bounding box distribution model.

The application relates to a network congestion link diagnosis method based on an adversarial automatic encoder and belongs to the technical field of network communication, which comprises the following steps: generating a data set containing link states and path states according to the congestion prior probability of a link; feeding the data set into an automatic encoder, superimposing a generative adversarial network on the encoder or the decoder of the automatic encoder, inputting the path states or the link states into the encoder, and reconstructing the link states or the path states; training the generative adversarial network, enabling the generative adversarial network to learn the congestion prior probability information of the link, and outputting a link state solution with the maximum congestion posterior probability; and reversely propagating the reconstruction loss and the generative adversarial loss, and iteratively updating network parameters. The application applies the adversarial automatic encoder, learns the unique mapping relationship from the link to the path through the link states and the path states, does not need to measure the network topology, and meanwhile, the adversarial automatic encoder is data-driven, has strong adaptability and has certain attack resistance.

The invention belongs to the field of deep learning and remote sensing image processing, and particularly discloses a hyperspectral few-sample classification network construction method and a hyperspectral ground feature classification method, and the method comprises the steps: obtaining a hyperspectral remote sensing image and sample label data thereof; performing cross-domain reconstruction and spectral curve extraction, and filtering out common information among categories through eigenvalue decomposition to obtain a discretized spectral curve; extracting physical invariance features, and taking the physical invariance features as constraint rules to generate virtual samples; the reconstructed hyperspectral data and the spectrum self-supervision auxiliary information are input into a double-branch variational automatic encoder network, multi-loss constraint of cross reconstruction is carried out, and the hyperspectral data and the spectrum self-supervision auxiliary information of the same category show greater similarity in a potential space; and outputting a final surface feature prediction result based on the multinomial logistic regression classifier, and completing the construction of the hyperspectral few-sample classification network. According to the invention, high-precision and high-robustness hyperspectral ground feature classification can be realized under the condition of sample scarcity.

The invention provides a cross-site scripting attack detection method, device and equipment, a medium and a product, which are applied to the technical field of computers, and are used for detecting a cross-site scripting attack by acquiring a pre-trained dynamic depth automatic encoder model and performing threshold selection on the pre-trained dynamic depth automatic encoder model. Determining a target error threshold value for cross-site scripting attack detection; obtaining to-be-detected current webpage feature data, and determining a current reconstruction error of the current webpage feature data by adopting a pre-trained dynamic depth automatic encoder model; and comparing the current reconstruction error with a target error threshold, and determining a cross-site scripting attack detection result of the current webpage feature data according to a comparison result. The target error threshold value determined based on the dynamic depth automatic encoder model is combined with the current reconstruction error to determine the cross-site scripting attack detection result, so that the detection precision of the cross-site scripting attack in a complex scene is improved, and the false alarm rate and the missing report rate are effectively balanced and reduced.

The application discloses a trajectory data preparation method based on federated learning, which utilizes a unified privacy protection framework designed by federated learning to realize trajectory data preparation in a federated environment by using the function of a large language model. Meanwhile, the application designs a trajectory privacy automatic encoder to ensure data transmission security and protect privacy, and introduces a trajectory knowledge enhancer to improve model learning of knowledge related to trajectory data preparation, so that development of the large language model for trajectory data preparation is realized. In addition, the application also proposes federated parallel optimization to improve training efficiency by reducing data transmission and realizing parallel model training.

The application provides an IoT data online prediction system based on a memory replay variational autoencoder, which comprises a prediction module and a training module; the prediction module is used for inputting IoT data to be predicted into a trained memory replay VAE to obtain a prediction result; the training module is used for training the memory replay VAE, wherein the memory replay VAE comprises an encoder and a generator; the training process of the memory replay VAE is as follows: first sample data are input into the encoder to obtain first sample latent factors and a first sample prediction result; the generator obtains first sample replay data based on the first sample latent factors; second sample data and the first sample replay data are input into the encoder to obtain fusion sample latent factors and a corresponding prediction result; a loss function is calculated based on labels and the obtained prediction result; and the training is completed when the loss is minimum. The application combines an OLVAE with an attention mechanism and a brain replay mechanism, alleviates the forgetting of old knowledge by the encoder, and realizes efficient prediction of IoT data.

A system for improving a machine learning model is described. In some cases, the system improves a potential diffusion machine learning model trained to receive text as input and output an image based on the received text by identifying an autoencoder for such model. The system identifies a plurality of channels in a decoder of the autoencoder, the decoder configured to receive the potential feature as input and output an image. The system also identifies a performance feature of the decoder and changes a node topology of the decoder based on the performance feature to generate an updated decoder. The system retrains the potential diffusion machine learning model by inputting potential features to an updated decoder, receiving an output image from the updated decoder, and updating one or more weights of the decoder based on an evaluation of the output image to use the updated decoder.