51 results about "Robust learning" patented technology

The invention discloses a heterogeneous sensing early warning system based on decoupling perception and adversarial robust learning, and the system comprises a feature extraction module which processes heterogeneous sensor original data collected in real time through a multi-layer decoupling encoder, separates target related features and environment interference features, and suppresses noise pollution from the source; the multi-dimensional collaborative fusion module adopts a cross-domain adversarial robustness learning framework to carry out space-time sequence alignment and deep fusion on decoupling features to generate high-robustness joint representation, and a data missing problem is processed through a cross-modal generative feature completion mechanism; and the cognitive enhancement closed-loop decision module constructs a cognitive heuristic confidence evaluation model based on joint representation, realizes graded early warning by combining real-time quality scoring and behavior prediction, and dynamically optimizes system parameters through a feedback mechanism. According to the method, the problems of poor target detection robustness, high delay and low accuracy in a complex dynamic environment are solved, the detection precision is remarkably improved, the false alarm rate is reduced, and the all-weather adaptive capacity is enhanced.

The invention discloses a built environment and traffic accident causal effect analysis method based on bi-robust learning, and belongs to the field of intelligent traffic safety and traffic accident risk assessment. By integrating satellite remote sensing, interest point distribution and traffic accident data, a spatial-temporal feature enhanced environment index system is constructed; realizing cross-regional data collaborative analysis by adopting federal learning, and guaranteeing data compliance in combination with differential privacy encryption; a generative adversarial network is introduced to generate a high-fidelity anti-fact scene, and a dual robust learning framework is coupled to accurately quantify a condition average processing effect of a built environment on accident risks. According to the scheme, a technical tool with both theoretical preciseness and practical feasibility is provided for urban traffic planning and safety management, and a scientific basis is provided for optimizing and building environment design and reducing traffic accident risks through a multi-source data fusion and privacy cooperation mechanism; and the blank in the aspects of causal robustness, insufficient anti-fact generation precision, privacy protection and multi-source data fusion in the prior art is filled.

Various embodiments provide a system, method, and device for applying a C2 machine learning-based detection framework. The method incudes (i) generating a fuzzing based on a C2 machine-learning detection model using a large learning model for performing profile-based seed generation; and (ii) detecting C2 traffic using the C2 machine learning detection model.

The invention discloses a medical image recognition system based on label noise robust learning, and provides a hard sample label refinement strategy based on a confidence perception weighted prototype and an effective noise sample joint correction method to process divided subsets in correction before training so as to obtain training data with higher quality; in progressive hard sample reinforcement learning, data are input according to the learning difficulty of samples for training, the ability of model learning discrimination feature representation is improved by integrating cross entropy loss, consistency loss and credible contrast loss, and the influence of medical tag noise is reduced. According to the method, by integrating label refinement and a progressive hard sample reinforcement learning technology, the accuracy and robustness of medical image recognition under the condition of noise label data are effectively improved, overfitting of the model to noise samples is relieved, and the method can be used for correctly building a disease auxiliary diagnosis model under the condition that the noise label samples exist in training data.

The invention discloses a robust federated learning method for processing heterogeneous noise and non-independent identically distributed data, and belongs to the technical field of federated learning. The method provides a robust learning framework of two-stage client quality perception. The method comprises the following steps of: 1, constructing a category-level loss vector and clustering by using a Gaussian mixture model, and accurately dividing a clean and noise client set; stage 2, performing differential training: performing standard training on the clean client; dual-network cooperative training, dynamic sample screening and exchange, and a self-distillation and entropy regularization mechanism are introduced to a noise client, so that robust learning is realized; in the global aggregation stage, a distance sensing weighting strategy is further adopted to dynamically suppress the influence of a noise client; according to the method, original data does not need to be shared, the robustness and generalization performance of the federated learning model in the coexistence environment of heterogeneous noise and non-independent identically distributed data can be effectively improved, and the method has wide application value in the fields of medical images, financial risk control and the like.

Embodiments described herein provide systems and methods for noise-robust contrastive learning. In view of the need for a noise-robust learning system, embodiments described herein provides a contrastive learning mechanism that combats noise by learning robust representations of the noisy data samples. Specifically, the training images are projected into a low-dimensional subspace, and the geometric structure of the subspace is regularized with: (1) a consistency contrastive loss that enforces images with perturbations to have similar embeddings; and (2) a prototypical contrastive loss augmented with a predetermined learning principle, which encourages the embedding for a linearly-interpolated input to have the same linear relationship with respect to the class prototypes. The low-dimensional embeddings are also trained to reconstruct the high-dimensional features, which preserves the learned information and regularizes the classifier.

The invention discloses a robust water meter reading identification method under a challenging scene based on YOLOv11, provides a comprehensive solution, and improves the accuracy and robustness of water meter reading identification under the challenging condition by solving the problems of environmental factors and mechanical digital jump. Firstly, a diversified water meter image data set is planned, robust learning of challenging samples is ensured, the adaptability of the model is enhanced, and a solid foundation is laid for model training and generalization. And secondly, the CSFE and the PSFM are integrated in the water meter reading identification model, a YOLOv11n framework is optimized, and the identification robustness in a challenging scene is remarkably improved. Finally, two new indexes TTMRrate and WADRrate are provided for comprehensive and practical performance evaluation, and more accurate identification performance evaluation of the water meter reading identification model is provided. According to the method, 98.3% of mAP at 50 and 96.19% of TTMRrate are realized, and the performance of an existing model based on YOLO on single character and sequence level identification tasks is exceeded.

The invention relates to a system, a method and a computer readable medium for selecting data to be stored on an edge device. More specifically, the present invention relates to an optimized data selection process that can improve pertinence to data observation samples that can be stored on edge devices to maximize edge learning quality. The to-be-stored data selected at the edge side can cover various categories of specific questions, improve the performance of the selected learning model, and incrementally maintain a support set for subsequent learning. According to the invention, edge-side robust learning can be realized, delay is improved, and privacy is enhanced.

The invention discloses an equipment health state intelligent identification-oriented low-quality data robust learning method. The method comprises the following steps of: obtaining multi-source sensor time sequence data of an electromechanical composite transmission system and preprocessing the multi-source sensor time sequence data; constructing a multi-sensor fusion identification network composed of a feature encoder, a projection head and a classification head; generating neighbor consensus tags and consistency measurement for samples according to the neighbor consistency of the embedded space, dynamically screening high-confidence samples and similar sample pairs, performing characterization pre-training by using a joint target of supervised contrast learning, classified cross entropy and sample pair similarity, and introducing a hybrid enhanced smooth discrimination boundary to obtain a robust encoder; freezing a robust encoder, training a new classification head only by using a high-confidence set, and obtaining a state recognition model; a state recognition model is used for directly reasoning a complete channel sample, a feature channel adaptability calibration module is introduced into a sensor channel data missing scene to dynamically compensate feature distribution offset, and probability distribution of each state category is output.

This invention discloses a cross-modal medical image segmentation method based on pathological anchoring. Using pathological priors as semantic anchors, this method integrates cross-scale collaboration and uncertainty perception in a closed-loop prompting refinement process. This achieves interpretable suppression and robust learning of cross-modal domain shifts, forming a closed-loop path from structure to semantics to noise control. Under strong cross-modal zero-sample settings such as multi-source ultrasound cross-domain segmentation and color dermoscopy, stable and consistent improvements are achieved, manifested in more precise boundaries, better calibration, and more robust generalization. This provides technical support for disease image diagnosis.

The invention discloses a noise label robust learning method, system and equipment for remote sensing change detection, and proposes an edge-guided cross-level difference enhancement module, improves the edge of a low-level difference feature through low-level-high-level feature interaction, and guides a high-level double-time-phase feature to perform difference enhancement by using the difference feature after the edge improvement, so that the noise label robust learning efficiency is improved. Edge information and change information of a dual-time-phase image are fully mined, a noise label joint detection strategy based on uncertainty analysis is combined with an uncertainty value and a prediction probability of a sample, a potential noise label is identified and corrected, and the risk of identifying a difficult sample as a noise sample is reduced; a difficult sample adaptive enhanced noise robust loss function is designed, a loss weight is adaptively distributed to a sample based on an uncertain value of the sample, the larger the uncertain value is, the more difficult the sample learning is, and the larger the loss weight is, and in this way, the network learning is forced to be more complex and more difficult in feature representation, and the network learning efficiency is improved. And the robustness of the model to a noise label is enhanced.

The invention discloses an intelligent legal file retrieval method and system based on a multi-modal heterogeneous large model, and the method comprises the steps: obtaining original legal file data containing a text, an image and a table, and carrying out the multi-modal analysis and structured preprocessing; extracting depth features of each modal through a multi-modal heterogeneous large model, and fusing the depth features by adopting a multi-modal cross attention mechanism; constructing a case element sparse graph and optimizing the case element sparse graph through an optimization algorithm; carrying out weight adjustment and vector optimization by using a robust learning real-value multi-index model algorithm; candidate case subsets are screened through a random recovery sampling technology, and multi-dimensional matching calculation is carried out; and carrying out visual presentation and interactive analysis processing on the similar historical case list. Through the multi-modal fusion understanding, sparse graph optimization, robust learning and other technologies, the legal file intelligent retrieval accuracy and efficiency are effectively improved, and accurate case reference and decision support are provided for legal workers.

The invention discloses an unsupervised cross-modal retrieval method based on robust consensus learning, and the method comprises the following steps: extracting image and text features of input data through a pre-training backbone network, and obtaining multi-level feature representation through a three-layer projector; generating a prototype as a pseudo label based on top layer feature clustering, calculating a sample weight by using contour coefficients of three levels, performing optimization through layered consensus prototype comparison loss, and finally obtaining a high-reliability pseudo label; the noise tolerance triple alignment loss and the modal consistency loss are jointly optimized, and the robustness of the model to noise is enhanced; and cross-modal retrieval is realized in the optimized shared semantic space. According to the method, a two-stage training strategy is adopted, the pseudo-labels are generated through top-layer clustering, the sample weight is optimized through multi-layer semantic information, robust learning is carried out based on the reliable pseudo-labels, the problems of pseudo-label noise and semantic granularity mismatch in an unsupervised environment are solved, and the accuracy and robustness of cross-modal retrieval are remarkably improved.

The invention relates to the technical field of artificial intelligence, distributed systems and information security, and provides a Byzantine robust learning method and system based on model segmentation, a medium and equipment. The Byzantine robust learning method based on model segmentation comprises the following steps: sending a current service model and partial weight parameters equally divided by all weight parameters of the current service model to each first device, so that the first device uses the partial weight parameters to train the current service model by using local data; the gradient after the training of the current business model is finished is received from each first device, and after all the gradients are sorted, the weighted average value of all the gradients after the extreme values are removed is calculated and used for updating the current business model; the updated service model is used as the current service model in the next round of model training process. According to the invention, the overweight calculation burden of the first equipment caused by uneven weight distribution can be reduced, so that the model training efficiency of the Byzantine robust learning system is improved.

The application provides an evaluation method based on an operation reliability intelligent evaluation model and related devices. System data of a power system is obtained, and after a minimum load shedding model is established according to the system data of the power system, the inherent mode of the minimum load shedding model is determined, and a robust learning loss function is constructed based on the inherent mode of the minimum load shedding model. Finally, strong generalization learning is performed according to the minimum load shedding neural network, the robust learning loss function and load disturbance, and an operation reliability intelligent evaluation model is obtained. The system data of the power system is input into the operation reliability intelligent evaluation model, and a result capable of evaluating the operation reliability of the system can be obtained, effectively improving the efficiency and accuracy of the evaluation result.

The invention provides a learning resource recommendation method for intention-aware double contrast learning, and relates to the technical field of sequence recommendation. The method comprises the following steps: introducing a mask language model (MLM) strategy, and using a historical learning resource sequence of a user to pre-train and finely adjust a Transform; a pre-trained Transform is used as a basic encoder, and long-term learning intention embedding of a user is captured in combination with an attention mechanism; constructing a learning resource map, and extracting short-term learning intention embedding of the user from pairwise learning resource conversion; a self-supervised double contrast learning strategy is designed, and modeling of long and short term learning intentions is enhanced; and adaptively fusing long and short term learning intentions, predicting a next learning resource, and constructing and optimizing an overall loss function. Through double-layer learning intention modeling and a self-supervised contrast learning strategy, long and short term learning intentions of a user are comprehensively captured, challenges of sparse learning behaviors and noise data are effectively coped, and personalized, accurate and robust learning resource recommendation is provided.

The present invention belongs to the field of computer vision and medical imaging fusion, and specifically relates to a medical image segmentation method based on multi-domain semi-supervision, including: constructing a semi-supervised teacher-student segmentation model and training it, inputting alienated domain data to be segmented into the trained segmentation model to obtain a segmentation result; the teacher-student network of the present invention mines a large number of high-level semantic features of unlabeled multi-domain medical images, adopts a network with a self-attention disentanglement mechanism to extract domain features and segmentation part features, uses a domain feature similarity screening mechanism and a multi-domain high-level semantic comparison loss function for robust learning, introduces an exponential sliding average algorithm to heterogeneously transform the student model into a teacher model, and uses a pixel-level errata solution under the constraint of teacher-student consistency, which improves the accuracy of segmentation, expands the applicability and generalization of multiple domains, improves the effect of deep models on image segmentation, and promotes the development of related technical fields.

The present application relates to the fields of artificial intelligence, distributed systems and information security technology, and provides a Byzantine robust learning method, system, medium and equipment based on model segmentation. The Byzantine robust learning method based on model segmentation comprises: sending a current business model and partial weight parameters divided by all weight parameters of the current business model to each first device, so that the first device uses the partial weight parameters to train the current business model using local data; receiving gradients of respective current business models after training from each first device, sorting all the gradients, calculating a weighted average of all the gradients after removing extreme values, and using the weighted average to update the current business model; and using the updated business model as the current business model in the next round of model training process. The method can reduce the heavy computing burden of the first device caused by uneven weight distribution, thereby improving the model training efficiency of the Byzantine robust learning system.

The application discloses a kind of legal dossier intelligent retrieval method and system based on multimodal heterogeneous big model, comprising: obtaining original legal dossier data containing text, image and table, carrying out multimodal analysis and structured preprocessing;Through multimodal heterogeneous big model, each modal depth feature is extracted and is fused using multimodal cross attention mechanism;Case element sparse graph is constructed and is optimized by optimization algorithm;Robust learning real value multi-index model algorithm is applied to weight adjustment and vector optimization;Through random recovery sampling technology, candidate case subset is screened and multidimensional matching calculation is carried out;Similar historical case list is visualized and interactive analysis and processing are carried out.The application improves the accuracy and efficiency of legal dossier intelligent retrieval by multimodal fusion understanding, sparse graph optimization and robust learning technology, provides accurate case reference and decision support for legal workers.

The invention discloses an industrial data prediction method and device based on a robust cyclic random configuration network, and aims to solve the problem that the modeling precision and robustness are insufficient when an existing model processes industrial data which is uncertain in input order, strong in time sequence and contains noise and abnormal values. Through combination of a robust learning mechanism and a recursive structure, order identification is not needed, reserve pool nodes are randomly generated based on a supervision mechanism, a network structure is constructed, and the global approximation capability of any nonlinear mapping function is ensured. The method not only inherits the efficient modeling capability of a cyclic network for time sequence characteristics, but also significantly enhances the stability and generalization capability of the model in a complex interference environment, and is suitable for accurate prediction and intelligent decision support scenes in a complex dynamic industrial process.

The invention discloses an intersection pedestrian-left-turn vehicle conflict risk factor quantification method based on causal machine learning. The method specifically comprises the following steps: 1, extracting pedestrian-left-turn vehicle conflict data; 2, matching the pedestrian-left turning vehicle conflict data with a signal control type Ttype and a safety influence factor; 3, constructing a directed acyclic graph; 4, on the basis of a directed acyclic graph, using a dual robust learning algorithm to construct causal inference models under different signal control types, and mining factors related to pedestrian-left turn vehicle conflicts from safety influence factors as conflict risk factors; and the heterogeneity causal effect of different signal control types on the pedestrian-left turn vehicle conflict risk is estimated. According to the method, confusion factors can be effectively identified and processed, and the robustness and generalization ability of the model facing real world data are improved.

The invention relates to the technical field of backscatter communication, particularly discloses an unmanned aerial vehicle assisted EH-based BackCom network resource allocation method and system, and provides a robust resource allocation problem which aims at maximizing the total throughput of all unmanned aerial vehicles. Key parameters (BackCom time allocation, transmitting power, master user interference protection and reflection coefficient) are jointly optimized. In order to reduce adverse effects of uncertainty of CSI parameters and reflection coefficients, a robust learning algorithm based on a polyhedron is introduced when a resource allocation problem is solved, and an outage probability constraint is converted into a deterministic constraint. And finally, decomposing the non-convex optimization problem into two sub-problems by adopting an alternating optimization method, and solving by adopting a Lagrange multiplier method. Simulation results show that the method achieves relatively high network throughput rate and energy utilization rate in an EH-based BackCom network assisted by the unmanned aerial vehicle.

The present invention relates to a noise-robust multi-modal video highlight detection learning method. This method takes the visual and audio multi-modal representation information of the video as input, and then compresses the high-dimensional features into a low-dimensional feature through linear mapping to reduce the number of parameters and video memory occupancy, while improving the running speed. Then, it combines the internal global-local spatial feature information of video segments and the temporal feature information between video segments to enhance the modal representation information. Next, through a multi-modal interaction and fusion module, the visual and audio modal representation information is interacted and fused. Finally, a target loss function is constructed to optimize the network. This method can effectively improve the accuracy and robustness of video highlight detection, has a wide range of applications, and is particularly suitable for video highlight detection of noisy label data.

The invention discloses a noisy label cross-modal retrieval method based on neighborhood perception instance refining, which comprises the following steps of: performing feature extraction on input noisy cross-modal data, and mapping the data to a shared semantic space; constructing a cross-modal boundary keeping module, and enhancing the global discrimination capability of feature representation through global mechanism regularization; constructing a cross-modal neighborhood consensus module, generating a soft label based on the cross-modal neighborhood consensus module, dynamically dividing a data set, and performing a targeted optimization strategy on the data set; integrating the loss functions of all the modules to perform end-to-end joint optimization, and training to obtain a cross-modal retrieval model with high robustness to noise tags; and efficient cross-modal retrieval is realized. According to the method, the robust learning method, the instance selection method and the label calibration method are integrated together, so that maximum utilization and fine processing of all training data are realized, and the retrieval performance in a high-noise environment is remarkably improved.

The invention belongs to the technical field of distributed machine learning, and more specifically relates to a heterogeneous distributed robust learning gradient aggregation method combining SVD and K-means. The method comprises the following steps: constructing a distributed learning system; calculating a gradient by using a stochastic gradient descent algorithm and uploading the gradient to a parameter server; calculating an SVD score of the gradient; calculating a K-means score of the gradient; and calculating a comprehensive score of each node by combining the SVD score and the K-means score, and selecting n-f gradients with the minimum comprehensive score for average aggregation to obtain a final global gradient, and using the global gradient obtained by aggregation for parameter updating. In the heterogeneous distributed machine learning system, the Byzantine nodes can be effectively identified and eliminated, so that the performance of the heterogeneous distributed machine learning system is improved.

The invention discloses an incomplete multi-mode learning method based on graph routing and prompt distillation, which comprises the following steps of: firstly, constructing a mode sensing routing and element self-adaptive prompt distillation framework oriented to incomplete multi-mode learning, capturing a dependency relationship between condition modes through a mode sensing graph to obtain fusion representation, and then activating a special path by sparse routing to obtain a fusion representation; according to the method, a low-rank adapter and a meta-adaptive prompt are generated, flexible fusion aiming at various missing modes is realized, finally, uncertainty perception distillation is adopted to carry out final cross-modal reasoning and robust learning, and incomplete multi-modal learning is realized. According to the method, a mode perception element prompt framework is provided, uncertainty perception prompt distillation is designed, the problem that a multi-mode learning model faces mode deficiency uncertainty and the problem that the multi-mode learning model faces mode deficiency performance suboptimum are solved, instance self-adaptive reasoning is achieved, and the method has the advantages of being simple in structure, convenient to operate and high in practicability. Reliable cross-modal knowledge migration and coherent representation learning can be realized in various missing scenes.

The invention provides a traffic filling rate gap position covering prediction and correction method, which comprises the following steps of: collecting historical and real-time traffic requests, bidding responses and actual filling results, eliminating abnormal data and constructing samples, defining and denoising filling rate gaps, extracting multi-dimensional features such as user groups, media bits, regions and time periods, and performing prediction and correction on the filling rate gaps. Establishing a tendency estimation model to output a complement source selection probability, realizing sample distribution balance through stabilizing inverse probability weight, performing consistency verification on processing identification, a probability range and a sample size, constructing a control group based on a tendency matching and weighting method, and calculating an anti-fact filling rate gap to generate a confidence interval; the method comprises the following steps of: obtaining a mapping function to form a complement source and anti-fact reference mapping table, estimating a causal gain by using a dual-robust learner, obtaining a gap closable degree through the mapping function, and screening out a reliable complement source. According to the method, dynamic prediction and intelligent correction of the gap can be realized in a complex putting environment, and the filling rate stability and the resource utilization efficiency are improved.

The present invention discloses a control method for drone information collection, comprising: defining the process of drone information collection in a continuous state and action space, taking maximizing the amount of information collected by the drone from multiple nodes within a given time as the optimization goal; converting the optimization goal solution for drone information collection in a continuous state and action space with state estimation errors into the solution of an optimal control strategy in robust reinforcement learning; and utilizing a robust Actor-Critic learning algorithm based on policy gradients to obtain the optimal control strategy in a continuous state and action space with state estimation errors. The present invention, for the first time, proposes how to handle the problem of autonomous control of information collection in a comprehensive situation in a continuous state and action space, where there are estimation errors in the drone state and the information of the collected information nodes is unknown, and converts the problem into the solution of an optimal control strategy in robust reinforcement learning, and finally solves it using a robust Actor-Critic learning algorithm, thereby expanding the application scenarios.

The application discloses a chemical experiment video online action segmentation method based on a perspective robust learning, relates to the technical field of video understanding and intelligent education, and comprises the following steps: frame-level sampling is performed on a real-time chemical experiment video stream, and a pre-trained visual feature extraction network is used to obtain a space-time feature sequence of continuous frames; feature separation is performed on the space-time feature sequence to obtain perspective-related context representation features and perspective-independent action representation features; the perspective-independent action representation is input into an online time sequence segmentation network, classification probability is dynamically corrected in combination with short-term historical segmentation results, and a final time sequence segmentation result of a current frame is obtained. Through the introduction of the perspective-independent action representation learning mechanism, the application reduces the dependence on and limitation of a shooting perspective of an experiment video, and is more suitable for the actual application requirements of intelligent chemical experiment operation analysis in a real teaching scene.

The invention provides a noise label cross-modal retrieval method based on gradient guide dynamic division, and belongs to the technical field of computers. According to the cross-modal retrieval method with the noise label, the problem that the noise label in multi-modal data affects the cross-modal retrieval precision can be solved, and the noise label sample is identified and processed by constructing a dynamic division strategy based on a sample loss function gradient. The method comprises the steps of common semantic representation learning, gradient-guided screening and correction, dynamic division and robust learning, cross-modal consistency learning and overall objective function construction, and finally cross-modal cross retrieval is carried out through the optimized overall objective function. According to the method, the multi-modal data with noise can be effectively utilized, and the accuracy and robustness of cross-modal retrieval are remarkably improved.