72 results about "Data diversity" patented technology

The invention discloses a face privacy protection-oriented generated image detection method and a face privacy protection-oriented generated image detection system. The method comprises the following steps of: firstly, preparing face and text pairing data, and finely adjusting a diffusion model; secondly, on the basis of the diffusion model after fine tuning, potential vectors are extracted and clustered, and text prompts and center vectors obtained through clustering form a dictionary; and finally, based on the obtained dictionary, obtaining a pseudo image and a label through the fine-tuned diffusion model, and outputting a detection result through a classifier. According to the method, potential spatial clustering and conditional diffusion generation are combined, privacy protection and data diversity are taken into consideration, and the security and generalization ability of forged face image detection are remarkably improved.

The invention discloses a large-scale language model fine tuning method, device and equipment and a storage medium, and relates to the technical field of large-scale language model fine tuning. According to the method, a preset check point is firstly subjected to large-scale language model fine tuning based on training loss performance, sample embedding space distribution and sample reply score performance of a current model; and performing multi-dimensional self-reference diagnosis on the training data set, identifying a suboptimal sample which is not matched with the capability of the current model, and then processing the suboptimal sample through an adaptive optimization engine to complete dynamic evolution of the training data set. Therefore, dynamic evolution of the data set adaptation model can be realized, the problem of redundancy or insufficient adaptation of static data is avoided, and the training efficiency is improved; meanwhile, data waste is reduced, data diversity is reserved, the model generalization ability is promoted, manual intervention is not needed, and the data optimization cost is reduced.

The invention discloses a domain generalization personnel re-identification method and system based on multi-modal fusion and structure perception enhancement. According to the method, a grey-scale map mode is innovatively introduced to extract biological characteristic information irrelevant to dressing, so that the identification limitation of a visible light image under the conditions of uniform shielding and severe illumination is effectively made up; meanwhile, a structure perception data enhancement strategy is provided, a key identification area is protected through a semantic segmentation technology, and damage to effective features is avoided while data diversity is improved; besides, by introducing targeted alignment loss, uniformity loss and intra-domain uniformity loss, the distribution characteristics of the feature space are directly optimized and regularized, and the generalization performance in an unknown substation scene is significantly improved. According to the method, superior detection performance can be realized under the condition of limited annotation data, the dependence on large-scale annotation data is effectively reduced, and the deployment cost is reduced.

The invention provides a training data generation method and device, electronic equipment and a storage medium. The method comprises the following steps: constructing a three-dimensional simulation environment constrained by a natural language navigation instruction; executing the natural language navigation instruction in a simulation environment corresponding to the three-dimensional simulation environment to verify the construction correctness of the three-dimensional simulation environment; if the navigation route is correct, verifying whether the navigation route described by the natural language navigation instruction is reachable in the three-dimensional simulation environment or not based on a navigation grid corresponding to a ground walkable area in the three-dimensional simulation environment; and if yes, generating training data of a visual language navigation task for the preset intelligent agent according to the natural language navigation instruction and the three-dimensional simulation environment. According to the method and the device, the diversity of the VLN training data and the consistency of the three-dimensional space can be considered, and the accessibility of the path corresponding to the navigation instruction in the three-dimensional environment is ensured while the data diversity is improved.

An adaptable deep learning method is provided that delivers sound hepatic lesion identification in NETs, while significantly reducing human effort for data annotation and improving model generalizability for PET image quantification. A region-guided GAN (RGGAN) model conducts image-to-image translation between list-mode simulated PET images and real-world clinical data, while preserving semantic content of interest, e.g., lesions. The RG-GAN model is integrated with a lesion detection model into an end-to-end, unified framework for joint-task learning, such that the two models can benefit from each other. The RG-GAN translates the list-mode simulated data into real world-style images, which appear to be drawn from the real clinical PET image dataset, and feeds the translated images into the lesion detection model for training. In order to deal with the limited diversity of list mode-simulated PET image data, a specific data augmentation module is incorporated into the unified framework to improve model training.

This invention provides a method for processing real-device data of an intelligent agent and related equipment. The method includes: acquiring a target action acquisition task and a target action type; in response to the target action type being a long-term task, acquiring key continuous operation segment data in the long-term task; acquiring failure scenario data and corresponding recovery operation data during the execution of the target action acquisition task; identifying and deleting invalid static frames from the key continuous operation segment data and recovery operation data using a pre-trained invalid frame recognition model to obtain cleaned data; evaluating the quality of the cleaned data based on a preset data quality assessment model to obtain a quality assessment score; and training the target intelligent agent using the cleaned data in response to the quality assessment score meeting a preset quality score threshold. This application can effectively reduce the redundancy of real-device data for intelligent agents and enrich data diversity, thereby improving the generalization ability of visual language action models.

This application discloses a method and apparatus for use in nodes for wireless communication and artificial intelligence. The node receives a first information block configured for training data collection for a first cell set and a second cell set; the maximum buffer size used by the node for the collection is a first size; the sizes of the training data collection for the first cell set and the second cell set are a first target size and a second target size, respectively; the first target size plus the second target size is greater than the first size; the node reports training data for the first cell set and the second cell set with reference to a first reference size and a second reference size, respectively; the second reference size is equal to the smaller of the candidate size and the second target size; the first reference size is equal to the first size minus the second reference size. This application improves data diversity.

This invention provides a method for building extraction from remote sensing images that integrates convolutional neural networks (CNNs) and Transformers, relating to the intersection of remote sensing image segmentation and computer vision technologies. This method involves creating a remote sensing image dataset from acquired remote sensing images, labeling each image individually, and dividing the dataset into training, validation, and test sets. The preprocessed remote sensing images are then preprocessed to increase data diversity. Feature extraction is performed on the preprocessed images to collect feature maps containing building information and obtain global features of the image. This method can significantly reduce false positives and false negatives for small target buildings, improve the completeness of segmentation for large target buildings, and mitigate boundary blurring caused by insufficient extraction of target building edge information.

In the field of machine learning, there may be challenges associated with constructing a comprehensive and diverse training data set. For example, the data that is available may not be sufficiently diverse, which may cause issues such as overfitting in a model trained using the available data. A computer-implemented method and system are provided to use an embedding function as a tool in assessing the diversity of a data set. The embedding function may be employed in constructing a training data set having a high degree of data diversity for training a model.

This invention relates to a fault tracing and analysis method for garment production line equipment. Addressing the challenges of processing multi-source heterogeneous sensor data and the interpretability of causal models in garment production lines, it proposes a fault identification and causal attribution method based on operational condition semantics and disturbance analysis. This method includes multi-source sensor data normalization and operational condition semantic mask construction, generation of physically constrained causal disturbance samples, dual-channel feature encoding to decouple causal features, counterfactual reasoning to measure the contribution of the causal subspace to the diagnostic results, and finally, generation of a structured fault report containing a causal inference chain. Furthermore, it can dynamically optimize the model feature space partitioning based on new fault modes. This scheme achieves efficient alignment and utilization of data diversity and dynamism in complex industrial environments, improves the accuracy of fault diagnosis and the transparency of decision support, and significantly enhances the interpretability and adaptability of the model.

The invention provides a multi-modulation signal robust identification method based on a time-frequency diagram and data enhancement. Original baseband signals of the traditional modulation mode and the novel modulation mode are generated and preprocessed; performing data set division according to a double sampling mode; performing STFT on the data to generate a time-frequency graph; performing enhancement operations of time axis offset, frequency axis offset, noise injection and tiny rotation on the time-frequency graph of the training set; a ResNet-18 model adaptive to single channel input is constructed; and training the model and carrying out a performance test of unknown signal modulation mode identification. According to the method, the novel data set comprising a traditional modulation mode and an emerging modulation technology is constructed, the defect that most of the current modulation data sets are the traditional modulation mode is overcome, a data basis closer to a future communication scene is provided for modulation recognition in a complex scene, multi-dimensional enhancement is performed on data, data diversity is improved, and the method is suitable for being applied to modulation recognition in a complex scene. The model generalization ability is significantly improved, and the identification precision of the modulation signal is also effectively improved.

A data diversity augmentation method includes inputting a sentence into an encoder and extracting a feature vector for the sentence, inputting the extracted feature vector into an attribute classifier to form a decision boundary for the feature vector, and moving the extracted feature vector based on the decision boundary to generate a transformed feature vector and inputting the transformed feature vector into a decoder to restore a transformed sentence for the sentence.

The application provides a project recommendation method, device, system and storage medium, and belongs to the field of item recommendation. The method comprises the following steps: training original knowledge graph triples by using a TransH model to obtain a scoring function; performing parameter update analysis on the TransH model according to the scoring function to obtain an initial loss function and an updated TransH model; and performing initialization processing on the original knowledge graph triples by using the updated TransH model to obtain an initialized project node vector, an initialized user node vector and an initialized relationship vector. The application realizes more accurate, interpretable and diversified user preference recommendation, alleviates the conflict between data noise and data diversity, and improves the accuracy and robustness of recommendation.

The invention discloses a colorectal early cancer infiltration depth prediction model construction method based on mixed cutting enhancement, and the method comprises the steps: carrying out the preprocessing of a historical white light endoscopic image, increasing the sample size on the premise of highlighting a focus region, obtaining a preprocessed sample set, obtaining a focus region ROI in the preprocessed sample set through a doctor and / or employing a feature recognition algorithm, and carrying out the prediction of the colorectal early cancer infiltration depth. The method comprises the following steps: marking category labels of shallow infiltration and deep infiltration on a focus region ROI, enhancing and improving data diversity and model generalization by utilizing CutMix to obtain enhanced data with labels, taking a historical white light endoscopic image, the enhanced data with labels and category labels as training data, and performing enhanced training on a basic model by virtue of a mixed cutting enhancement strategy, so as to obtain an enhanced white light endoscopic image, an enhanced white light endoscopic image, an enhanced white light endoscopic image, an enhanced white light endoscopic image, an enhanced white light endoscopic image, an enhanced white light endoscopic image and an enhanced white light endoscopic image; after a white light endoscope image is input into the colorectal early cancer infiltration depth prediction model obtained through training, the shallow infiltration / deep infiltration type is predicted, and objective judgment of colorectal early cancer infiltration depth under a conventional white light endoscope is achieved.

This invention discloses a generative image detection method and system for protecting facial privacy. The method first prepares face-text paired data and fine-tunes the diffusion model. Secondly, based on the fine-tuned diffusion model, latent vectors are extracted and clustered, and a dictionary is formed by combining text prompts with the cluster center vectors. Finally, based on the obtained dictionary, pseudo-images and labels are obtained through the fine-tuned diffusion model, and the detection results are output after passing through a classifier. This invention combines latent space clustering with conditional diffusion generation, achieving a balance between privacy protection and data diversity, and significantly improving the security and generalization ability of fake face image detection.

The application discloses a point cloud semantic segmentation method based on sampling enhancement and multi-view, first, airborne laser radar point cloud data is collected, a triangular facet model is used to generate virtual samples to enhance data diversity, and a dataset is obtained after preprocessing; then, the dataset is sampled, a boundary point analysis module is used to optimize the sampling strategy, and boundary features of ground objects are mainly captured; then, a point cloud semantic segmentation model is constructed, three-dimensional geometric features are extracted by using a kernel point convolution network, two-dimensional image features generated by multi-view side projection are combined, multi-dimensional feature complementation is realized, the model is trained and optimized according to a loss function; finally, the point cloud semantic segmentation model is used for semantic segmentation of new airborne laser radar point cloud data. The application can improve the recognition accuracy of ground object boundaries in complex urban scenes and the classification performance of small-scale targets that are easy to confuse, effectively reduces the burden of relevant staff, and improves production efficiency.

The application provides a device selection and bandwidth allocation system and method for layered federated learning, relates to the field of layered federated learning, and comprises a device selection module and a bandwidth allocation module configured with a cloud server, an edge server and a terminal device; the device selection module is used for selecting a terminal device; the bandwidth allocation module allocates bandwidth for the selected terminal device; the total training time is divided into a local training time, a local update transmission time and a pre-aggregated result uploading time, and the device selection and bandwidth allocation are performed under the constraints of total training time minimization and data diversity. The device selection and bandwidth allocation problem in the original layered federated learning is effectively solved, and an innovative scheme is provided for improving training performance and guaranteeing data diversity and model accuracy.

This invention discloses a method for segmenting skin lesions, comprising: extracting an initial feature map from a target skin lesion image; then performing feature extraction using an encoder incorporating a wavelet convolutional layer to obtain a wavelet feature map; multiplying the wavelet feature map with a tensor p that has undergone depthwise separable convolution and bilinear interpolation to obtain a processed wavelet feature map, dividing it into four groups of feature maps for processing to obtain a preliminary output feature map; performing multi-scale feature fusion on the wavelet feature map and the preliminary output feature map respectively; inputting the preliminary output feature map into a decoder; and performing an add operation on the intermediate output of the decoder with multiple multi-scale fused feature maps. The final output of the decoder is the skin lesion segmentation result. This application, through grouped feature processing, the introduction of partial convolution, and multi-scale feature fusion, not only improves the accuracy of feature extraction but also reduces the number of model parameters, improves computational efficiency, and enhances the module's adaptability to data diversity.

The invention relates to a large-scale test data generation method and device based on a distributed computing framework, and the method comprises the steps: analyzing table structure information of a target data table, and obtaining field annotations of a plurality of fields of the target data table; judging whether each field in the plurality of fields is a special field which needs to follow a specific generation rule or not; the method comprises the following steps: generating test data of a field and generating test data of the field according with a specific generation rule; and summarizing the generated test data of the plurality of fields and writing the summarized test data into a target data table. The parallel capability of the distributed computing framework is combined with the complex rule understanding capability of the large language model, so that on the premise of ensuring data diversity and business rule conformity, the generation efficiency and automation level of mass test data are remarkably improved, and high operation and maintenance complexity and labor cost are effectively reduced.

This invention relates to the field of artificial intelligence technology, and in particular to a method and apparatus for detecting and locating multiple sound events based on a neural network model. The method includes: innovatively designing time-frequency multi-scale residual convolutional blocks, which, together with a Conformer module and a cross-stitch unit module, form a network model to extract features at multiple scales, enhance long sequence modeling, and promote task-based collaborative optimization, thereby improving performance and accuracy; in terms of data processing, pre-emphasis and frame-by-frame windowing improve feature quality, audio channel swapping and spectrum enhancement increase data diversity and reduce overfitting, and SALSA-Lite features are used to enhance feature representation; in terms of training strategy, a multivariate loss function is used to accelerate convergence while considering task requirements, and hyperparameters are flexibly adjusted using a validation set. This makes the method highly efficient in training, has excellent practical performance, strong generalization ability on unknown data, and can accurately cope with complex and ever-changing real-world scenarios, effectively overcoming the shortcomings of traditional methods.

This invention belongs to the field of multimodal data processing technology and discloses a multimodal data alignment and enhancement method based on semantic consistency, aiming to solve the problems of semantic gap and insufficient generalization of multimodal data models. The method includes: acquiring original multimodal data on the same semantic topic, performing noise and anomaly processing and format normalization on the multimodal dataset; extracting semantic features of each modality from the multimodal dataset and normalizing and adapting them to construct a shared semantic space; achieving semantic consistency alignment between modal feature vectors of each modality through a cross-modal alignment mechanism; accordingly, performing intra-modal and cross-modal enhancement on each modality and its shared feature vectors, and integrating and deduplicating to obtain an aligned and enhanced multimodal dataset. This invention achieves accurate semantic alignment of multimodal data, improves data diversity, effectively enhances the generalization ability of multimodal learning models, and is applicable to artificial intelligence scenarios that rely on multimodal data collaboration, demonstrating strong practicality.

The application discloses a digital city simulation method fusing structured semantic modeling and trajectory sequence generation, and aims to solve the problems of insufficient data diversity and low multi-agent reasoning efficiency in trajectory prediction. First, a simulation environment based on a graph is constructed, and dynamic models such as IDM and MOBIL are combined to generate diversified and scene-consistent trajectory data. Second, a deep learning model based on ConvLSTM and CVAE is constructed to uniformly represent the agent state and the scene in the bird's eye view. The state pooling operation is introduced to realize synchronous prediction and achieve constant-time reasoning efficiency. Finally, the CVAE is used to model the trajectory behavior type, and the multi-modal future trajectory is generated through conditional sampling. The application provides efficient support for digital city simulation and automatic driving decision planning.

The invention relates to the field of smart fishery and computer vision, in particular to a Unity-based cultured fish target tracking model training data generation method, and the technical scheme comprises the steps: through a data set generated through virtual simulation, the high cost of manual labeling is avoided, and the research and development cost of the culture industry is reduced, so that the labeling cost is reduced; by adjusting parameters such as water, illumination and fish behaviors, training data with high diversity is generated, the requirements of different algorithms are met, and therefore the data diversity is improved; the generated large-scale data set can be used for algorithm training and verification, the development efficiency of an intelligent breeding system is improved, and therefore algorithm development is accelerated; the fish modeling does not depend on a real shooting environment, so that the cost of equipment and site construction is saved, and the environmental dependence is reduced; the method has the advantages of being low in cost, high in efficiency, capable of expanding the data size according to needs and the like, the acquisition difficulty of the multi-target tracking training data can be remarkably reduced, and the accuracy and consistency of labeling are guaranteed.

A method includes segmenting a region of interest in an anatomical region in both a source imaging data and a destination imaging data, wherein different regions of the region of interest are labeled with different segmentation masks. The method includes selecting a region from the different regions from the source imaging data and spatially matching the region to a corresponding region in the destination imaging data. The method includes determining a spatial intersection between the region and the corresponding region. The method includes utilizing an intersection mask to crop a first portion of the region from the source imaging data and utilizing the intersection mask to remove a second portion of the destination imaging data in the corresponding region. The method includes adding the first portion of the region from the source imaging data into the destination imaging data where the second portion was removed.

The invention discloses a medical image classification method based on semi-supervised federated learning, and the method comprises the steps: constructing disease distribution and equipment heterogeneous information among multiple institutions through the priori knowledge in the field of medical images, and constructing a fairness-driven federated learning framework in combination with the data diversity, historical participation degree and labeling proportion of a client; performing knowledge representation on the unmarked medical image by applying an entropy maximization strategy, and reasonably increasing the distance between samples instead of simply regarding different samples as negative samples; and then multi-round iterative optimization is carried out on model parameters through a fairness score discrete sampling mechanism and a weighted aggregation strategy. According to the method, efficient and fair classification identification can be carried out on medical images which are not independently and identically distributed and are scarce in labels in multiple medical institutions, an identification mode depending on centralized decoupling training or full-annotation data is effectively eliminated, the problem that rare disease species and low-resource hospitals are forgotten by models is solved, the manual annotation cost is reduced, and the identification efficiency is improved. And the classification fairness and generalization efficiency are improved.

The invention discloses a data diversity receiving method and device, computer equipment and a storage medium, and relates to the technical field of diversity, and the method comprises the steps: receiving first antenna data of an antenna of a first base station; receiving second antenna data sent by a second base station; and performing data diversity processing based on the first antenna data and the second antenna data. In the embodiment of the invention, the antenna data of a plurality of base stations are integrated in diversity processing, so that the system can utilize idle receiving resources of other base stations, and diversity of diversity receiving is expanded under the condition that the number of local antennas is not increased. According to the cooperation mechanism, the signal coverage range is obviously expanded, especially in a complex external field environment, the problem of signal shielding or interference of a single station can be effectively solved through multi-base-station cooperation, and therefore the stability of the overall receiving performance is improved. Meanwhile, the dependence on hardware resources of a single base station is reduced through cross-device data integration, the antenna deployment cost of the base station is reduced, and the application scene of the diversity technology is more flexible.

The adversarial learning-based point cloud denoising enhancement method of the present invention includes the following steps: during the training of a deep point cloud denoising network, a point cloud noise confidence estimation module is embedded into the original point cloud denoising network; noisy point clouds are input into the deep point cloud denoising network embedded with the point cloud noise confidence estimation module; adversarial loss is calculated based on the point cloud confidence vector estimated by the point cloud noise confidence estimation module and the denoised point cloud result predicted by the deep point cloud denoising network; in each iteration, the input point cloud is perturbed according to the adversarial loss, and adversarial examples are generated and added to the training. The adversarial learning-based point cloud denoising enhancement method of the present invention can dynamically generate adversarial examples that are difficult for the model to handle, improve the data diversity of point cloud denoising training, and the enhanced model of the present invention has certain improvements over the benchmark model in the tasks of generating Gaussian noise and simulating lidar sampling noise denoising.

The present application relates to the field of production management, in particular to a self-defined field and template low code system for unstructured contract order, and specifically comprises a contract self-defined module and an exported contract template module, wherein the contract self-defined module comprises a self-defined field structure module and a self-defined field value module for extracting and analyzing unstructured data; the exported contract template module comprises an order defined field module and a module style conversion module for matching the field structure and field value in the analyzed unstructured data, setting the contract order style, and then selecting a text format output. The entry mode of unstructured information can maximize the potential value of information, widen the information entry mode, improve the information collection and processing speed, better realize data mining, retain data diversity, and provide more accurate information processing mode for enterprises.

The invention relates to the technical field of communication, discloses a CNN-based time-frequency synchronization blind detection method and system, and aims at a complex channel environment and high-precision synchronization requirements of a 5G / 6G system to generate a real training data set by simulating multipath fading and noise so as to ensure data diversity and reliability. PSS signal features are automatically extracted by using CNN, complexity of manual feature design is avoided, and detection accuracy and robustness are significantly improved; compared with a traditional algorithm, the method is low in calculation complexity and short in detection time delay, PSS signal blind detection can be rapidly completed, and downlink time-frequency synchronization of the base station and the UE is achieved; according to the method, the high requirement of a 5G / 6G system on a synchronous detection algorithm is effectively met, and powerful support is provided for efficient operation and digital transformation of a network.

The invention discloses a text data balancing method based on adversarial training, and belongs to the field of data balancing, and the method comprises the steps: determining a target text sub-data set from each text sub-data set of a text data set to be subjected to data balancing; establishing a text generation cooperation strategy corresponding to the target text sub-data set based on the target text sub-data set, and performing cooperation adjustment on a text generator and a generated text discriminator respectively corresponding to the target text sub-data set based on the text generation cooperation strategy; and performing data expansion on the target text sub-data set based on the adjusted text generator to obtain a text data set after data balance. According to the method, a collaborative adjustment mechanism based on an adversarial thought is introduced, so that the balance between data diversity and simulation degree is realized while the category distribution deviation is effectively eliminated, and the problem that the data balance and the data capacity cannot be considered at the same time is solved.