17 results about "Unsupervised algorithm" patented technology

The invention belongs to the technical field of petroleum and natural gas engineering, and particularly relates to a priori knowledge embedded unsupervised learning geology-engineering dessert comprehensive evaluation method, which comprises the following steps: S1, obtaining an evaluation data set of a research area, and preprocessing and standardizing the evaluation data set; s2, dividing the evaluation data set into a positive index and a negative index according to the prior knowledge, constructing positive and negative ideal points through the prior knowledge, and fusing to generate an unsupervised evaluation data set; s3, clustering is carried out on the unsupervised evaluation data set by adopting principal component analysis and an incremental online k-means algorithm; and S4, according to category labels obtained by clustering, determining geological-engineering dessert grades by adopting a standardized variable distance grading evaluation method, and carrying out segmented clustering optimization and perforation position optimization on the geological-engineering dessert grades. According to the method, a supervised evaluation effect is achieved through an unsupervised algorithm, comprehensive evaluation of the dessert of the fractured horizontal well section is achieved, and the problem of unlabeled data evaluation is effectively solved.

The invention provides an FFU filter screen leakage early warning triggering system and method based on a particle counter, and relates to the technical field of filter screen leakage early warning, and the system comprises a data collection module which calculates the actual particle concentration change rate based on the collected data of an FFU; the reference curve construction module is used for establishing an associated reference curve of the rotating speed-particle concentration of the FFU motor under multiple particle sizes based on historical normal working condition data; the anomaly recognition module is used for calculating a deviation value between the actual particle concentration change rate and the associated reference curve, and marking leakage to be detected if a preset anomaly judgment standard appears; and the early warning module detects the abnormal score of the leakage to be detected through an isolated forest unsupervised algorithm, outputs the leakage probability in combination with a gradient boosting tree supervised algorithm, and triggers early warning based on the leakage probability. According to the method and the device, the technical problem of relatively low early warning accuracy of the leakage of the FFU filter screen in the prior art can be solved, and the technical effect of improving the early warning accuracy of the leakage of the FFU filter screen is achieved.

This disclosure describes techniques for analyzing network traffic to generate an actionable insight pertaining to a security threat to a network. In one example, this disclosure describes a method that includes obtaining, by a computing system, historical network activity data that includes information about authentication traffic within a network; determining, by the computing system and based on the historical network activity, a baseline of network activity; collecting, by the computing system, a set of network activity data; applying, by the computing system, an unsupervised algorithm to identify the set of network activity data as anomalous relative to the baseline of network activity; classifying, by the computing system, the network activity data into an identified threat category from among a plurality of threat categories; and taking action, by the computing system and based on the identified threat category, to mitigate a security threat posed by the network activity data.

The application discloses a time series anomaly detection method for an online monitoring system of a dressing and smelting residue field, and belongs to the technical field of the online monitoring of the dressing and smelting residue field. The steps of the application are as follows: data is extracted from a database, missing values are filled, some original indexes with relatively high correlation coefficients are deleted, statistical features, time features and time sequence features are constructed for the remaining indexes through feature engineering; all the features are normalized, and an unsupervised algorithm is used to determine an abnormal label through majority voting; feature selection is performed on the processed label data, and the importance indexes are used to improve the detection performance of the unsupervised algorithm; then, the pseudo label is corrected in combination with an industry standard; finally, the detection results of the unsupervised algorithm are weighted and fused, the statistical features, the time features and the time sequence features are constructed based on a small amount of monitoring projects, the important features are selected with the aid of a supervised algorithm, and the performance of the unsupervised anomaly detection algorithm is improved.

The invention discloses a security event response system and method based on intelligent analysis, and relates to the technical field of network security, an asset business load integrated digital model is constructed, a cross-domain attack surface is identified based on cross-environment asset interaction data in the integrated digital model, and basic data support for subsequent simulation and reasoning is formed; based on the obtained asset, business and cross-domain attack surface data, an attack framework and a dynamic attacker portrait are fused to construct a causal knowledge graph, the causal relationship of unknown attacks is complemented through transfer learning and an unsupervised algorithm, and the causal knowledge graph is updated according to the dynamic change of the environment; according to the method, the conversion of the security event from passive tracing to active prediction is realized, high-risk threats are identified in advance through cross-domain attack path simulation and risk quantification, and the defense initiative is improved.

Aspects related to predicting zero-day vulnerabilities using anomaly detection and neural network algorithms are provided. A prediction platform may train an unsupervised algorithm for identifying suspicious packets and a prediction model for generating suspicion scores and behavior patterns based on network traffic information. The platform may segment information of packets of network traffic information into a plurality of segments. The platform may compare the segments with zero-day vulnerability information to identify known zero-day vulnerabilities. The platform may use the unsupervised algorithm to identify suspicious packets that do not correspond to known zero-day vulnerabilities. The platform may generate suspicion scores and behavior patterns for suspicious packets. The platform may further train the prediction model based on behavior patterns associated with certain suspicion scores to generate vulnerability scores. The platform may generate vulnerability scores for suspicious packets using the model. The platform may output zero-day vulnerability predictions based on the vulnerability scores.

This application discloses a method, apparatus, electronic device, and medium for anomaly detection in photovoltaic module data based on an integrated unsupervised algorithm. The method may include: establishing and training corresponding base learners based on multiple unsupervised algorithms using a SCADA dataset; determining the anomaly score and weight for each base learner; and performing anomaly detection based on the anomaly score and weight. This invention can improve the automation level of detection, reduce manual workload, promptly detect and handle anomalies, effectively prevent the escalation of faults, extend the service life of photovoltaic modules, thereby improving the power generation efficiency of the entire system and reducing operation and maintenance costs.

The application discloses a kind of based on sparse auto-encoder's malicious URL detection method, this method with sparse auto-encoder as the core model to complete the detection of malicious URL, including: collecting normal URL to construct data set, as training data input;URL is preprocessed and parameter mapping;URL is converted respectively by TF-IDF and Doc2Vec method, two vectors obtained are connected as the complete feature vector of URL;A sparse auto-encoder is constructed, and the feature vector is used as input to train;The URL to be tested is input into the trained sparse auto-encoder, the spatial distance of its input and the output obtained is calculated, and whether it is a malicious URL is judged in combination with a given threshold value.The method effectively reduces the data cost required in the process of constructing the detection model using unsupervised algorithm, and also has the advantages of accurate feature extraction and efficient model training, and achieves an average accuracy of 95% in detecting new URL samples, achieving simultaneous improvement in efficiency and accuracy.

Techniques are disclosed herein that encompass image pre-processing and a semi-supervised clustering for optimization and analysis of immune-enriched single-cell proteomics data generated via multiplexed imaging technologies. This is achieved through an image pre-processing pipeline, which converts image data contained in one type of file (e.g., .mcd) into another type of file (e.g., .tiff) and removes artifact signals from the image data using various algorithms to generate improved image data. Thereafter, a semi-supervised clustering pipeline analyzes the improved image data using various techniques, including implementing a supervised algorithm to identify metaclusters such as general immune phenotypes (e.g., CD4−T-cells, Macrophages, Neutrophils, etc.) as well as non-immune phenotypes while implementing an unsupervised algorithm that enables the identification of specific subclusters and a more in-depth cellular status characterization.

The application provides a BERT-based entity attribute similarity calculation method and system, which filters out the relationship triplets, redundant or invalid triplets in the knowledge graphs through preprocessing two knowledge graphs, and obtains pre-aligned entity pairs; aligns the relationships of the first knowledge graph and the second knowledge graph; uses the BERT model to perform unsupervised learning on the tail entities, i.e. attribute texts, in the attribute triplets, and obtains a trained BERT model; according to the relationship alignment result, the attribute text similarity with the same relationship is calculated through the trained BERT model, and the results of the entity attribute similarity are obtained through calculation and integration; the application can calculate the entity attribute similarity between the knowledge graphs with high accuracy through the unsupervised algorithm in the case that the knowledge graph data is huge and lacks annotation, provides support for the fusion of the knowledge graphs, and can effectively solve the problems of high labor cost and low accuracy of the annotated samples in the knowledge graph fusion process.

The invention provides a general object counting method and system based on an unsupervised algorithm, and relates to the technical field of automatic object counting. The method comprises the following steps: acquiring an image containing to-be-counted objects; extracting an image dense feature map by adopting a self-supervised training DINO model; by interactively marking a bounding box or uploading an example graph, extracting and optimizing example features through coordinate transformation, ROI-Align operation and ellipse weighted mask; using the example features as convolution kernels to generate a similarity graph and aggregating the similarity graph; normalizing the aggregation similarity graph to obtain a density graph, and screening effective areas according to a threshold value; and dividing the small object scene into sub-regions and independently processing the sub-regions to obtain an aggregation result. The system comprises a corresponding image acquisition module, a feature extraction module and the like. The method does not need data labeling and training, simplifies batch operation, improves the counting precision of small objects, is suitable for industrial and agricultural multiple scenes, and meets the requirements of efficient and accurate counting.

The invention discloses a network physical power grid attack pre-stage detection method and system based on an isolation forest algorithm, a medium and a processor, and relates to the technical field of attack pre-stage detection. Aiming at the problem of lack of targeted detection before attack in the prior art, the method comprises the following steps: collecting core data of different levels of CPPS to eliminate a detection blind area; performing missing value processing, feature screening and other preprocessing on the original data; constructing an isolation forest model to calculate a data point anomaly score; and recognizing continuous abnormal behaviors through sliding window time sequence analysis. The system comprises an acquisition module, a preprocessing module, a calculation module and a judgment module. According to the method, multi-source data coverage and time sequence trend analysis are combined, fine anomalies before attack are accurately captured, the problem of performance imbalance of an unsupervised algorithm is solved, the anomaly recognition rate is high, the false alarm rate is low, crossing from passive response to active defense is achieved, and the CPPS active protection requirement is met.

The invention relates to the technical field of computer vision and anomaly detection, and particularly provides a cable surface anomaly detection method and device based on an unsupervised algorithm, and the method comprises the following steps: S1, collecting a cable image, and carrying out Gaussian filtering denoising, CLAHE algorithm brightness normalization, Sobel edge extraction and Otsu segmentation preprocessing; s2, improving ResNet50, and performing model pre-training through a contrast learning framework; s3, non-defect cable features are extracted based on the pre-training model, and after abnormal patch filtering and greedy core set compression, a feature library is constructed by using FAISS; s4, preprocessing a to-be-detected image, extracting features, and calculating a cosine similarity with the feature library to obtain a patch-level abnormal score; s5, dynamically adjusting a threshold value based on the similarity of the core set, and realizing abnormal segmentation in combination with expansion corrosion; and S6, outputting a result according to the abnormal patch proportion. Compared with the prior art, the cable fine defect identification capability and detection robustness can be improved, the marking cost is reduced, and the method is suitable for industrial cable quality detection.

The application relates to a cone barrel obstacle clustering and processing method based on an unsupervised algorithm and belongs to the technical field of unmanned driving, and comprises the following steps: obtaining point cloud data containing 2048 points through a laser radar; performing noise filtering on the point cloud data; performing dimension reduction processing on the point cloud data; and outputting a point cloud clustering center, that is, a cone barrel center point, by adopting an unsupervised K-means clustering algorithm. The point cloud data is transformed by a PCA algorithm and a matrix transformation method, is mapped into a space with more obvious features, and the mapped features are used as main clustering bases for obstacle detection on an actual vehicle, so that the clustering accuracy under the condition of algorithm power priority is improved; by the unsupervised algorithm, based on a large amount of point cloud data returned by the laser radar, the ROS unmanned vehicle can automatically make accurate clustering and positioning of the cone barrel center, without additional manual intervention and without being affected by light factors.

The invention provides a material defect detection method based on an unsupervised algorithm visual model. According to the method, a class encoder is introduced into an unsupervised algorithm visual model, an image generation process is guided to approach a target domain, and the situation that a generator cannot find an effective generation strategy is prevented; a loop structure is introduced to realize effective utilization of non-pairwise data, so that dependence on pairwise data is relieved, the pressure of pairwise data acquisition and processing is relieved, and the problem of data scarcity can be solved. Therefore, by means of the improved unsupervised algorithm visual model, more comprehensive information such as types, sizes, shapes, positions and distribution of the defects can be rapidly obtained, and an important reference basis is provided for defect analysis and control. In addition, the computer vision algorithm also supports real-time defect detection and monitoring, and when the algorithm is applied to real-time detection on a production line, the defects can be found and controlled in time, the influence of the defects on the product quality is avoided, and meanwhile, the production cost is reduced.

Embodiments provide methods and systems for processing unstructured and unlabelled data. A method includes generating, by a processor, a structured and unlabelled training dataset from an unstructured and unlabelled dataset. The method includes categorizing the structured and unlabelled training dataset into a plurality of clusters by executing an unsupervised algorithm. Each cluster of a selected set of clusters from the plurality of clusters is labelled with an applicable label from a set of labels. The method includes executing a supervised algorithm to generate a trained supervised model using a labelled training dataset including the set of labels and an input dataset generated from plurality of datapoints present in each cluster of the selected set of clusters. The method includes generating a Labelled Data1 (LD1) by executing the trained supervised model configured to assign applicable label from the set of labels to each datapoint of the structured and unlabelled training dataset.

The embodiment of the invention discloses an unsupervised mixed domain separation method and device for aliasing seismic data, and the method comprises the steps: carrying out the random time delay coding of original common-shot-domain multi-seismic-source aliasing data according to a seismic channel, and obtaining the randomly delayed common-shot-domain multi-seismic-source aliasing data; constructing a common-shot-domain unsupervised learning set by using the randomly delayed common-shot-domain multi-seismic-source aliasing data and the original common-shot-domain multi-seismic-source aliasing data; constructing a common detection domain secondary random aliasing unsupervised learning set by using the randomly delayed common shot domain multi-source aliasing data and the original common detection domain multi-source aliasing data; performing self-supervised cross iteration training by using the two learning sets to obtain a trained aliasing seismic data separation model; and inputting common-detection-domain multi-seismic-source aliasing data to be separated into the model for separation processing to obtain result data after aliasing separation. According to the invention, an unsupervised algorithm is adopted, and unsupervised intelligent processing of aliasing seismic data is realized.