50 results about "Decision boundary" patented technology

The application relates to a power image model knowledge migration method and system based on predicted divergence confrontation, which comprises the following steps: performing feature description on an original power image through a visual model, performing hidden space sampling on the feature description of the original power image through a feature extractor, and generating an initial latent vector; inputting the initial latent vector into a generator to obtain a reconstructed power image, and generating an adversarial power image through gradient symbol method-based adversarial disturbance; inputting the adversarial power image into a target model and a substitute model respectively, and calculating a predicted divergence loss; training the substitute model to learn the knowledge of the target model through the predicted divergence loss; and when the performance of the substitute model on a verification set is stable and close to that of the target model, completing knowledge migration. Compared with the prior art, the application significantly reduces data labeling cost, improves the adaptability of knowledge migration in a resource-limited scene, and improves the learning efficiency of a substitute model for key features of a target model and the decision boundary exploration ability.

The application belongs to the technical field of power grid transient stability evaluation, and specifically discloses a boundary sample enhancement method and system for power grid transient stability evaluation, which comprises the following steps: training a transient stability evaluation model by using an initial training sample set, predicting the initial training sample by using the trained transient stability evaluation model, and screening boundary samples based on the obtained prediction probability; calculating the local density of all non-boundary stable samples, and performing undersampling on the non-boundary samples; training a mask autoencoder generative adversarial network by using the screened boundary samples, generating new boundary samples without labels, obtaining sample labels by using a time domain simulation technology, and adding the new samples that pass the test to the training sample set after undersampling to realize boundary sample enhancement. The application can effectively identify the samples near the classification decision boundary of the transient stability evaluation model, and provides reliable training data basis for boundary sample generation and enhancement.

This invention discloses an open-set radar interference pattern recognition method, apparatus, and electronic device. The method includes: matched filtering of the radar received signal to construct a dual-modal input of a one-dimensional range sequence and a two-dimensional time-frequency matrix; feature extraction via a dual-branch network and adaptive fusion based on cosine similarity; introducing cross-modal consistency regularization during the training phase, constructing pseudo-unknown samples by shuffling intra-batch modes, and combining energy constraint loss to compress the energy of known classes and increase the energy of unknown classes to form a clear decision boundary; and adaptively setting a threshold based on the energy distribution of the validation set during the inference phase to achieve accurate identification of known interference and effective rejection of unknown interference. This invention overcomes the performance degradation defects of traditional closed-set methods in the face of unknown interference, and improves the generalization ability, robustness, and engineering practicality of radar interference recognition in complex electromagnetic environments.

The disclosure provides a data set distillation method based on a boundary perception diffusion model, comprising: pre-training an initial expert model using an original training set, fitting a decision boundary of the initial expert model to a true boundary by adjusting a weight parameter; generating a multi-condition fine-tuning loss reflecting different categories using a class probability predicted by the trained expert model as a confidence weight, and guiding fine-tuning of the diffusion model through minimization of a weighted sum of the multi-condition fine-tuning loss; performing multi-condition sampling using the fine-tuned diffusion model, and guiding the image generation of the diffusion model to approach the decision boundary region between the two categories by introducing labels to weight and average the noise prediction results of the target class and adjacent competing classes according to a mixing coefficient; and dynamically mixing boundary discrimination samples generated through multi-condition sampling and intra-class representative samples generated through single-condition sampling according to a preset ratio to construct a lightweight synthetic data set as a final result of data set distillation.

This invention relates to the field of data processing technology, and in particular to a method and system for processing imbalanced data based on nearest neighbor constraints and consistency screening. The method includes acquiring industrial fault detection data; performing data preprocessing based on the acquired industrial fault detection data; constructing a minority class MNN skeleton based on DenMune clustering and performing two-layer noise removal; local adaptive oversampling based on MNN sparsity double-layer weights and MNN neighborhood constraint interpolation; dual-threshold screening based on source sub-cluster consistency and original sample dominance; and using a trained model to predict and classify fault samples. This effectively suppresses the introduction of new noise during oversampling, making the final training dataset more stable near the local structure and decision boundary, which is beneficial for improving the stability and robustness of the industrial fault detection model in imbalanced scenarios.

Computer systems and computer-implemented methods modify a machine learning network, such as a deep neural network, to introduce judgment to the network. A “combining” node is added to the network, to thereby generate a modified network, where activation of the combining node is based, at least in part, on output from a subject node of the network. The computer system then trains the modified network by, for each training data item in a set of training data, performing forward and back propagation computations through the modified network, where the backward propagation computation through the modified network comprises computing estimated partial derivatives of an error function of an objective for the network, except that the combining node selectively blocks back-propagation of estimated partial derivatives to the subject node, even though activation of the combining node is based on the activation of the subject node.

The application discloses a boundary sample data enhancement method and device for knowledge distillation and a computer storage medium. The method comprises the following steps: before knowledge distillation is performed, the output of a teacher model is used to modify samples in each original data set along the decision boundary of the teacher model step by step, and a plurality of boundary samples suitable for knowledge distillation are expanded. In each iteration, the original sample or each sample modified in the last iteration is used as a basic sample, the approximate tangent plane of the decision boundary near the sample is calculated by using the output of the teacher model, and the sample is modified along multiple directions on the tangent plane; then, the modified sample is modified to be located near the boundary; finally, a plurality of samples farthest from other basic samples are selected as the result of the modification in the round and the basic samples for the next iteration. The application can meet the demand for data enhancement in current image classifier knowledge distillation.

This invention discloses a method and system for constructing a data model for a business intelligence agent based on deep learning, relating to the field of deep learning technology. Through continuous processing from steps S1 to S4, the business intelligence agent no longer trains its model solely based on the final adopted decision results. Instead, it incorporates candidate information explicitly rejected during historical decision-making processes into the modeling scope. It retains excluded candidate decisions by constructing a set of rejected information (Una), and transforms the original rejected information into computable multi-dimensional structured data by generating a set of rejected features (Ufc). Then, based on the density of rejected information (Ind), it characterizes the concentration of rejection behavior at different decision stages, forming a rejected boundary representation (Nbd) which is applied to the business intelligence agent. When generating new candidate decisions, it can proactively identify and avoid historically high rejection density regions. Compared to existing methods that rely solely on input and output results for training, this approach overcomes the problem of insufficient learning of decision boundaries.

The application relates to a model stealing attack query detection method for a machine learning-based network intrusion detection system and relates to the technical field of network attack detection.The application aims to solve the technical problem of combining query samples and model responses for comprehensive analysis, adopting a combination of decision boundary leakage evaluation and reconstruction error evaluation to quickly identify model stealing attacks with a small amount of query samples.The technical points are as follows: the prediction results of the query samples are used to construct a cumulative distribution function, the model uncertainty is measured, and potential boundary detection behaviors are detected; the reconstruction error loss of the query samples is calculated by using an automatic encoder, potential abnormal queries deviating from the normal distribution but not showing obvious malicious characteristics are identified; based on the results of the decision boundary leakage evaluation and the reconstruction error evaluation, a model safety coefficient is calculated to comprehensively evaluate the possibility of whether the model is subjected to a model stealing attack in a set period of time.The application is suitable for scenes where the ML-NIDS needs to be quickly detected and real-timely defended.

A machine tool component fault early warning and diagnosis method based on open set adversarial learning, after preprocessing and field construction of collected machine tool component monitoring data, an open set adversarial fitting network capable of self-adaptive fitting of classification decision boundary and hypothesis space boundary is constructed, dynamic adversarial learning strategy is used to ensure network training stability, Nash equilibrium of deep feature mapping module and boundary integrated fitting module is realized, open set fusion decision technology is used to fuse the output labels of multiple samples on the open set adversarial fitting network, accurate machine tool component fault early warning and diagnosis is realized. The present application includes both labeled data and unlabeled data in model training, which can diagnose both healthy operation type and fault type within the hypothesis space, and identify abnormal operation type outside the hypothesis space, breaking through the limitation of the prior art that can only identify a few fault types, which is of great significance to ensure the operation reliability of machine tool components.

The application discloses a phase modifier automatic start-stop control method based on energy consumption self-adaptive optimization, electronic equipment and a storage medium. The method comprises the following steps: acquiring multi-source operation data of the phase modifier and performing pretreatment and feature extraction, and constructing a standardized feature matrix; based on the matrix, a device dynamic energy consumption benchmark model is constructed by combining a long short-term memory network with a self-attention mechanism to predict a theoretical energy consumption baseline, and the theoretical energy consumption baseline is fused with power grid dispatching and load prediction data, and a future reactive power demand prediction sequence and a marginal energy consumption coefficient sequence are synchronously output through a time series convolution network; based on the above sequence, a multi-class energy consumption model containing start-stop transient, no-load and load operation loss is constructed, a dynamic start-stop decision boundary is generated through multi-objective optimization, and finally a start-stop control instruction resistant to disturbance is output by combining a sliding window comparison and fuzzy logic processing. The application realizes adaptive optimization of the phase modifier start-stop control, and improves the economy and reliability while ensuring the stability of the power grid.

This invention belongs to the field of artificial intelligence security and adversarial machine learning technology, and discloses a dataset compression and concealed backdoor attack method and related apparatus based on distribution matching. The dataset compression and concealed backdoor attack method includes: training an agent model based on a selected initial compressed dataset; analyzing the decision boundary of the agent model and selecting source and target class pairs according to the degree of confusion; adding triggers to clean samples of the source class to form poisoned samples; mixing the poisoned samples with clean samples of the target class to construct a poisoned training set; and using a distribution matching algorithm to compress the poisoned training set to generate a compressed dataset with an embedded concealed backdoor for training downstream models. This invention jointly optimizes three key indicators—attack success rate, model clean test accuracy, and attack concealment—in the dataset compression stage, solving the core problem that existing technologies struggle to achieve a balance among these three aspects.

The present application relates to the technical field of computer neural network, and discloses a kind of intelligent prediction model training system of pelvic abdominal cooperative function state, comprising: time delay parameter calculation module, for determining discrete time delay index based on envelope cross-correlation;Asymmetric convolution coding network, index offset convolution layer is embedded to call index offset memory reading pointer;Collaborative feature fusion module, element-by-element multiplication is performed;Time sequence misplacement suppression module, annular shift negative sample is constructed and inhibitory penalty is applied, the present application realizes the physical cause alignment of heterogeneous signal under the premise of not increasing network depth by implanting deterministic time sequence bias in convolution operator bottom layer, and constructs cause decision boundary using counterfactual training strategy, eliminates false co-occurrence misjudgment and reduces computing overhead.

The application discloses a kind of fusion knowledge enhancement and the method and system of original meaning prediction of confrontation training, it is related to natural language processing technical field, obtain target word and its corresponding original dictionary explanation;Original dictionary explanation is carried out semantic integrity evaluation using semantic perception selective enhancement intelligent agent;Target word, original dictionary explanation and supplementary context are structured as structured input sequence;Structured input sequence is sent into original meaning encoder, semantic features are extracted, and original meaning prediction vector is obtained;Using the method based on fast gradient, dynamic confrontation disturbance is applied to word embedding layer during model training process, clean sample loss and confrontation sample loss are optimized jointly to update model parameters;The original meaning label set corresponding to target word is output.The application introduces two big mechanisms of selective knowledge enhancement and confrontation training, and systematically solves the problem that decision boundary is weak due to static dictionary information bottleneck and original meaning long tail distribution in traditional original meaning prediction.

The application discloses a kind of intelligent weighted fusion and dynamic threshold early warning method for project implementation.For the problems of insufficient project feature representation capability and static early warning threshold easily causing early warning shock in prior art, the application proposes an algorithm architecture that integrates multi-scale feature analysis and mathematical statistics dynamic threshold control.In particular, the application first constructs a feature sequence, and extracts multi-scale features in parallel input deep separable convolution, selective state space and causal self-attention branch, generates an abnormal score through gated network weighting and combining evolutionary difference items;Then, the dynamic starting threshold is solved by fitting the residual error with Pareto distribution using extreme value theory, and the state comparison is completed by combining the double delay decision boundary to output the early warning signal.The application significantly improves the capture ability of implicit anomalies, effectively overcomes early warning shock, and provides a high-precision closed-loop early warning and supervision scheme for project implementation.

The application discloses a stream data self-adaptive classification method and system based on artificial intelligence, relates to the technical field of artificial intelligence and machine learning, and comprises the following steps: acquiring stream input data and a classification model of a current time step; using the model to obtain a relative position feature representing a decision boundary and to calculate an instantaneous gradient vector; updating the feature to a feature sequence, determining a target update resistance threshold based on the distribution discrete degree of the feature in the time sequence dimension; attenuating a historical accumulated gradient vector and superimposing the instantaneous gradient vector to obtain a target accumulated gradient vector; then judging whether the length of the vector is greater than the resistance threshold; if not, keeping the model parameters unchanged and retaining the accumulated gradient to the next time step; if yes, updating the classification model parameters based on the difference and resetting the target accumulated gradient vector; and the application effectively removes transient random interference and real concept drift, and breaks the noise resistance and sensitivity bottleneck.

This application relates to a method and system for risk early warning of geotechnical infrastructure based on a multi-scale damage model. The method includes: acquiring environmental monitoring data of a target geotechnical area based on a preset monitoring scale; transforming the environmental monitoring data into a multi-scale damage model through multi-scale coupling analysis modeling; extracting spatiotemporal features of damage from the multi-scale damage model using a pre-trained convolutional neural network, and determining risk patterns by matching with a preset risk pattern library; inputting the spatiotemporal features of damage and risk patterns into a pre-trained risk diffusion probability prediction model to obtain risk propagation prediction results; constructing a multi-dimensional risk decision boundary using a nonlinear classifier based on the risk patterns and risk propagation prediction results; and generating graded early warning instructions through preset dynamic risk decision rules. This method can achieve accurate and dynamic early warning of risks to geotechnical infrastructure, preventing safety accidents caused by multi-scale damage evolution.

This invention discloses a method and system for generating virtual out-of-distribution samples to suppress overconfidence in neural networks. The method includes: sampling training images and inputting the training image samples into an encoder module, which maps the training image samples from a high-dimensional feature space to a low-dimensional feature space; obtaining in-distribution training image samples located at the edges and identifying K pairs of in-distribution samples that are farthest apart; generating virtual out-of-distribution sample candidates based on the K pairs of in-distribution samples that are farthest apart; filtering the virtual out-of-distribution sample candidates to obtain virtual out-of-distribution samples; and using the generated virtual out-of-distribution samples during the training of the neural network, assigning these virtual out-of-distribution samples a confidence level below a preset threshold. The virtual out-of-distribution samples generated by this method are closer to the in-distribution samples, better constrain the decision boundary of the in-distribution samples, reduce the number of reference out-of-distribution samples, and alleviate the training burden of deep neural networks.

The application discloses an open set state monitoring method for an unmanned underwater vehicle based on a composite loss function constraint, comprising the following steps: collecting original vibration signals of an execution mechanism and performing time-frequency analysis to generate a two-dimensional time-frequency spectrum; constructing an open set state monitoring model based on a deep neural network and defining a composite loss function; training the model by using the time-frequency spectrum, compressing the distance within the known state class in the feature space and pushing away potential unknown samples by minimizing the composite loss function; based on extreme value theory, performing Weibull fitting on the tail of the distance distribution of the training samples to construct a dynamic open set decision boundary; in the online monitoring stage, processing vibration signals in real time and extracting features, and combining the dynamic boundary to determine whether the current state is a known state or an unknown abnormality. The application guarantees the recognition accuracy of the known state and realizes effective rejection of the unknown abnormal state, thereby significantly improving the state monitoring reliability of the unmanned underwater vehicle in a complex marine environment.

This invention belongs to the field of machine tool cutting state monitoring and intelligent control technology, specifically relating to a method for controlling the cutting state of a tool during the boring process of a space frame member. The method includes: real-time acquisition and preprocessing of multi-dimensional cutting data during the boring process to obtain a basic feature sequence; calculation of the ratio of the average effective cutting power of the sliding window to the instantaneous material removal rate, and obtaining a steady-state cutting energy index after deducting the initial cutting energy benchmark; calculation of the time-dependent decay weight of historical samples using an exponential decay function, and then obtaining the drift compensation boundary probability through root mean square fusion to dynamically correct the isolation forest model's decision boundary; further fusion of the average path length and normalization constant, and obtaining an anti-baseline drift correction anomaly score through the product of the exponential term and the logarithmic suppression term; setting a safety threshold, and performing anti-chipping closed-loop control based on the comparison result between the anomaly score and the threshold. This invention improves the accuracy of tool control during the boring process of space frame members.

The application belongs to the field of industrial visual inspection, and particularly relates to an industrial anomaly detection method and system based on pseudo-anomaly feature space optimization. First, an anomaly-driven denoising diffusion probability model AD-DDPM is constructed, and Berlin noise disturbance is introduced in the reverse denoising process to synthesize pseudo-anomaly samples with real physical textures. Second, a "normal-anomaly-mask" triplet is constructed using the synthesized information, and a multi-scale feature selection module MSFS is used to select a feature subset with high discriminability. Then, a center-guided contrast learning mechanism is introduced, a global normal class center is set in the feature space, normal samples are gathered to the center, and pseudo-anomaly samples are pushed away from the center to optimize the decision boundary. Finally, the reconstruction residual containing the most anomaly information is selected to generate an anomaly score, and the precise positioning of defects is realized. The application effectively eliminates the dependence of the model on real anomaly data, and significantly improves the detection accuracy of small defects and the robustness of the model in a complex background.

The application discloses a kind of port oil pipeline leakage monitoring method and system, belong to pipeline defect monitoring technical field.The port oil pipeline leakage monitoring method, comprising the following steps: constructing machine learning model for identifying pipeline leakage and its mode;Real-time monitoring the sound signal of port oil pipeline, and the sound signal is input into the machine learning model;Whether the oil pipeline leaks and the leakage mode are judged by the machine learning model whether the oil pipeline leaks and the leakage mode are judged by model;Machine learning model includes the following steps: initial model is constructed using random forest algorithm, decision boundary is determined, to obtain trained machine learning model, and the trained machine learning model is verified using the test set Performance.The port oil pipeline leakage monitoring method of the application can realize non-invasive, online monitoring of pipeline, thereby improving the effectiveness and flexibility of pipeline leakage monitoring.

This invention discloses a method for generating natural fairness test cases for machine learning models. It simulates the decision-making process of a machine learning model in the latent space by constructing a dataset, deriving an approximate proxy decision boundary. Furthermore, leveraging the model's inrocity at the decision boundary, a latent vector candidate detection strategy is designed to generate fairness test cases that conform to natural laws near the proxy decision boundary. This helps to better test the fairness characteristics of machine learning models in real-world scenarios and improves the fairness and credibility of machine learning models. This invention can be applied to generating natural fairness test cases for structured data such as tables and unstructured data such as images, demonstrating good application value and practical effectiveness.

This invention belongs to the field of image classification technology, specifically relating to a federated learning image classification method based on semantic bridging. This paper introduces a server-side semantic adapter and proposes a semantic bridging federated learning framework based on a pre-trained visual language model. The client utilizes a pre-trained visual language model (such as CLIP) as a stable semantic anchor, enabling local training to align visual features with the semantic knowledge of the global model. The server aggregates global semantic knowledge through the semantic adapter and optimizes the global classification head. At the local client level, this invention employs dual semantic distillation, forcing local features to align with the global semantic anchor at both the backbone and projection levels, while simultaneously utilizing memory replay technology to repair the decision boundaries of missing categories. Experiments show that this method significantly improves performance compared to state-of-the-art federated learning methods in non-independent and identically distributed environments.

The application discloses a hard label black box model stealing method based on semantic saliency, and through a three-stage cooperative mechanism of "static semantic division + saliency guided detection + information gain weighted refinement", blind pixel-level disturbance is improved to high-explanation semantic block-level detection within a limited query budget, and decision boundary information is efficiently triggered. Meanwhile, a consistency gating mechanism is introduced to realize high-value sample reinforcement learning and unreliable sample robust error correction, and the cloning accuracy and behavior consistency of a substitute model to a victim model are significantly improved.

The application relates to the technical field of industrial data processing, and discloses a vehicle-mounted mobile pump station abnormal data detection method and system. The method comprises the following steps: acquiring real-time vibration parameters and pressure fluctuation data, and obtaining a preliminary feature vector through preprocessing; clustering the preliminary feature vector to obtain grouped data clusters; training a classifier to obtain a decision boundary, and obtaining an abnormal probability score through probability mapping; if the threshold is exceeded, analyzing the feature contribution degree and matching a mapping dictionary to obtain a preliminary abnormal label; performing cross-validation and confidence evaluation by fusing historical environmental factor data to obtain a refined abnormal source classification; combining real-time feedback to adjust grouping parameters to obtain an optimized data grouping model; and using the optimized model to perform abnormal quantitative analysis on real-time data flow, and if the attribution probability exceeds an early warning limit, a fault early warning signal is generated and a blocking logic is triggered. The method can realize high-precision abnormal detection and real-time early warning of a vehicle-mounted mobile pump station under a complex environment.

The application discloses a kind of drop lightning arrester multi-source signal discrimination and alarm method, it is related to the on-line monitoring and fault early warning technical field of power system equipment, the application can effectively separate high-frequency sub-band signal by wavelet packet decomposition technology, adapt to non-stationary vibration environment, ensure the robustness of feature extraction;The introduction of approximate entropy and peak factor and other characteristics, quantifies the randomness and impact of signal, can sensitively capture the subtle changes in the fatigue accumulation process of lead wire;Single-class classification model trained only by healthy sample is used, such as support vector data description, decision boundary is constructed, so that the model has high sensitivity to abnormal state, and can early warning before lead wire fracture occurs;This kind of evaluation mode based on healthy baseline avoids the false alarm problem of traditional threshold method in complex environment.

This invention belongs to the interdisciplinary field of sensor design and artificial intelligence, specifically a sensor design method that proactively adapts to the application environment. This invention employs an active learning approach, actively selecting the most informative samples from unlabeled data for labeling, thereby maximizing model performance with less data acquisition. Compared to traditional methods of multiple sensor fabrications, active learning dynamically assesses the uncertainty or representativeness of data samples during model training, prioritizing data points that have the greatest impact on the model's decision boundaries and the strongest performance improvement potential. This method is not only applicable to a single type of sensor chip, but its database design and algorithm model are compatible with various material systems (including organic / inorganic, flexible / rigid, etc.), and can achieve rapid adaptive modeling and parameter recommendation for different application scenarios through transfer learning, demonstrating good scalability and versatility.