1419 results about "Positive sample" patented technology

The invention provides an intelligent interaction method and device and an agent-based interaction system, relates to the technical field of artificial intelligence such as natural language processing, large language models and deep learning, and can be applied to multi-agent collaboration based on a large language model (LLM). Comprising the steps of determining a knowledge base, a positive sample reference statement and a negative sample reference statement which are currently associated with a dialogue system; performing intention recognition on the current first input statement based on the knowledge base, the positive sample reference statement and the negative sample reference statement, and determining a first intention of the first input statement and the definition of the first intention; outputting an intention inquiry statement under the condition that the definition of the first intention is smaller than a first threshold value; obtaining a second input statement, and returning to execute the step of performing intention recognition on the first input statement based on the second input statement until a target intention with the definition greater than a first threshold value is obtained; and outputting a reply statement corresponding to the first input statement based on the target intention.

The invention belongs to the technical field of data processing, and discloses an enterprise multi-type data labeling method and system based on feature engineering, and the method comprises the steps: collecting initial data of an enterprise internal information system, and dividing the initial data to obtain a text data set, a table data set and an image data set; feature vectors of different types of data are fused after features are extracted and screened, and a multi-dimensional feature space is constructed; constructing an annotation model according to the feature information of the feature space, constructing similar sample pairs as positive sample pairs and irrelevant sample pairs as negative sample pairs for unannotated new data, and comparing, learning and optimizing the similarity between the similar sample pairs; and carrying out label inference based on the feature vectors obtained by comparative learning, and generating a labeling result. According to the method, efficient and automatic labeling of various types of data can be achieved, the method is suitable for various heterogeneous data such as texts, tables and images, and high labeling precision and expansibility are achieved.

A computer-implemented method for artificial intelligence (AI) based risk / value assessment of a geographic area includes performing feature engineering to contextually enrich collected data. Three datasets are generated from the contextually enriched data, where a first dataset is generated by combining positive samples of the contextually enriched collected data with hard negative samples of the contextually enriched data, a second dataset is generated by combining the positive samples with soft negative samples of the contextually enriched data, and a third dataset is generated by combining the positive samples, hard negative samples, and soft negative samples. A machine learning model is trained to generate three different types of predictions for the risk / value assessment of the geographic area based on the three generated datasets.

The invention discloses a standard video-in-channel monitoring early warning model method and terminal, and the method specifically comprises the steps: creating a universal video access platform, and configuring an early warning model, specifically, the configuration of the early warning model comprises the steps: analyzing a video stream frame by frame, judging whether an image is abnormal or not, and immediately giving out an early warning by a system once the abnormality is detected, real-time playback or storage of related frames is triggered; the video corresponds to a static wide angle, video analysis is carried out through static background comparison and sound analysis, whether a change occurs is judged, and a change result is recorded; a video middle station realizes deep understanding and alarm management of video contents through an intelligent analysis technology and in combination with multi-dimensional information, an algorithm model is trained and optimized by constructing a positive sample set and an error sample set, configuration is adjusted through a visual interface, and the alarm accuracy and efficiency are improved; dynamic video electronic fence early warning is established, and through real-time monitoring and intelligent analysis, abnormity identification is realized, so that an alarm mechanism is triggered.

A system for increasing the accuracy in summarization techniques is disclosed. The system generates a set of summaries for text. The system determines a label for each summary based on a set of composite metrics. The label for the summary indicates the truthfulness and faithfulness of the summary with respect to the text. The system determines that more than a threshold number of the set of composite metrics indicate that a first summary is assigned with a first label. In response, the system adds the first summary paired with the text as a positive sample to a dataset. The system determines that more than a threshold number of composite metrics indicate that a second summary is assigned with a second label. In response, the system adds the second summary paired with the text as a negative sample to the dataset. The system trains a summarization algorithm with the dataset.

The invention discloses a method and system for evaluating the susceptibility of avalanches based on GIS and machine learning, and belongs to the technical field of avalanche susceptibility evaluation. The method comprises the following processes: determining the influencing factors causing avalanches in a target area according to the terrain data, meteorological data and avalanche data of the target area; performing correlation detection on the influencing factors causing avalanches preliminarily selected to determine the evaluation factors; constructing a positive sample set and a negative sample set according to the evaluation factors and avalanche point data information of the target area, training a machine learning model according to the positive sample set and the negative sample set, and obtaining an avalanche susceptibility evaluation model after training; producing an avalanche susceptibility zoning map according to the prediction results of the avalanche susceptibility evaluation model, and determining the avalanche susceptibility of each area; the method can improve the rationality and scientificity of the avalanche susceptibility evaluation results, and provide theoretical support and method guidance for the avalanche disaster prevention and control system.

The invention discloses a privacy protection-oriented personalized federal recommendation system and method, and relates to the field of federal recommendation systems. According to the method, an article preference embedding vector is constructed for each client, and a global article embedding orthogonal mapping matrix and a double-layer perceptron used for generating an article embedding mixed factor matrix are trained, so that the global article embedding matrix and the article preference embedding vector of a user are fused; a mixed article embedding matrix which is more personalized and finer in granularity is constructed for the client, and the mixed article embedding matrix is input into a personalized recommendation prediction perceptron constructed for the client to obtain a recommendation result, so that the final recommendation effect is improved; when the client side uploads data to the central server, local differential privacy encryption is carried out; according to the technical scheme, negative sampling is carried out on each client side, the positive samples and the negative samples are mixed and then jointly used for training, the training effect is improved, meanwhile, the server cannot distinguish the historical interaction situation between the client sides and the articles according to the update data uploaded by the client sides, and user privacy is protected.

The invention provides an information recommendation method and device based on collaborative sampling of a knowledge graph and an adjacent graph, and relates to the technical field of information recommendation. According to the method, the triple is constructed by obtaining the historical interaction data of the user, and the triple is fused with the project attribute graph to construct the collaborative knowledge graph; performing expression alignment through projection transformation to obtain initial embedding expression; performing multi-hop neighbor sampling through biased random walk to obtain a multi-hop neighbor set; determining a neighbor which is most similar to the embedding representation of the positive sample item as an anchor neighbor; then linear interpolation is carried out, and the embedding representation of the difficult negative sample is synthesized; on the basis of the multi-head graph attention network, training by adopting a positive sample and a synthesized difficult negative sample, and aggregating to obtain an updated embedded representation; and calculating an interaction probability based on the updated embedded representation, and generating a recommended item for the user. According to the method, the problems of user-project interaction sparsity and false negative example interference are effectively solved, and stronger semantic comprehension and generalization ability are provided for a recommendation system.

The invention relates to the technical field of recommendation algorithms, provides a graph collaborative filtering course recommendation method based on interactive attention and comparative learning, and aims to solve the problem that a traditional recommendation system is insufficient in modeling ability in a sparse interaction scene. The method comprises the following steps: firstly, constructing a user-course bipartite graph, and utilizing a dynamic attention mechanism guided by an interactive opposite-end node: carrying out vector dot product through original embedding of the opposite-end node (for example, course embedding is used during user aggregation) and current embedding of a neighbor node, and generating an attention coefficient in combination with temperature parameter normalization; and multi-level structure information aggregation is realized. Afterwards, local context features are fused through a multilayer graph convolutional network, random noise disturbance is introduced to generate a multi-view comparison sample, the consistency of positive samples is maximized in combination with an InfoNCE loss function, and the robustness of the model to noise and sparse data is enhanced; and finally, optimizing user-course embedding in combination with Bayesian personalized sorting loss and comparison loss, and generating a personalized recommendation list. According to the method, the key interaction relationship is screened through guided attention, the representation discrimination is improved in combination with comparative learning, and the recommendation precision in cold start and data sparse scenes can be improved.

The invention relates to the field of mechanical arm motion planning, relates to a multi-mechanical arm motion planning method based on an improved artificial potential field method and deep reinforcement learning, and aims to solve the problems that in a multi-mechanical arm system, early-stage rewards are sparse, the training speed is low, and effective strategies are difficult to learn in deep reinforcement learning. The invention provides an improved dynamic artificial potential field expert demonstration scheme, and in the training process, an artificial potential field reward function and a dynamic artificial potential field expert demonstration mechanism which dynamically change along with the training process are added, so that the mechanical arm obtains higher learning efficiency and more positive samples in the early-stage training process; in later training, the mechanical arm is assisted to reach a high-precision target, and then the learning speed of the mechanical arm is increased. The training process mainly comprises the following steps of constructing a simulation environment, describing and modeling a multi-mechanical-arm path planning task, generating a training sample, executing the training task and performing simulation, dynamically adjusting expert data according to a simulation result, collecting experience and updating a strategy. Compared with a traditional method, the method has the advantages that the convergence speed is higher, the training time is shorter, the precision is higher, and particularly, the early-stage convergence speed is higher, and the fitting precision is higher.

The invention relates to the technical field of artificial intelligence, in particular to a method and device for selecting a knowledge base and a storage medium. Comprising the steps of obtaining a current query intention and a current user portrait of a user; obtaining a positive sample data set in the sample data set; selecting a positive sample data set of the large model for the knowledge base to generate a task scene cue word; constructing a knowledge base selection large model based on the task scene cues; inputting the current query intention and the current user portrait into a knowledge base selection large model to obtain a knowledge base selection score of each currently available knowledge base; according to the preset selection rule and the knowledge base selection score of each currently available knowledge base, the target knowledge base aiming at the current query intention of the user is selected from each currently available knowledge base, the personalized demand of the user is considered, the target knowledge base required by the user can be accurately determined, and the user experience is improved. And the efficiency and accuracy of subsequently answering user questions and answers are improved.

The invention discloses a data construction and fine tuning method for a knowledge retrieval model in the field of energy and power, and the method comprises the steps: carrying out the preprocessing of document data in the field of energy and power, and segmenting the document data into document segments suitable for the input of a retrieval model; performing question generation based on a large language model, generating a question-document pair positive sample set according to document fragments, and sampling to generate a question-document pair negative sample set; and in combination with contrast learning of the positive sample set and the negative sample set and LoRA parameter fine tuning, training a retrieval model, and performing problem retrieval by using the trained retrieval model. According to the method, high-quality problem-document pairs and challenging negative samples are generated, a comparative learning technology is combined, the retrieval model is optimized, the retrieval accuracy of the vector model in the field is remarkably improved, specific retrieval requirements in the energy power field can be deeply understood, and retrieval suggestions which are highly reliable and conform to habits in the field are generated.

The present disclosure provides a training method and apparatus for an image-text matching model, a device and a storage medium. The method includes: acquiring a positive sample and a negative sample; where the positive sample includes text and an image, the text in the positive sample is used to describe content of the image in the positive sample; the negative sample includes text and an image, the text in the negative sample describes content that is inconsistent with content of the image in the negative sample; training the image-text matching model by using the acquired positive sample and the acquired negative sample based on a manner of contrastive learning; where the image-text matching model is used to predict, for an input image and input text, whether the input text is used to describe content of the input image.

The invention discloses a consistent learning-oriented sample enhancement and optimization method, which comprises the following steps of: respectively performing knowledge extraction on professional knowledge data, professional vocabulary interpretation, historical practice data and an expert experience set in an unstructured text to obtain an original question and answer pair and a professional knowledge base; processing the original question and answer pairs through a diversified strategy to generate enhanced sample question and answer pairs with consistent semantics and various forms, and expanding or simplifying the enhanced sample question and answer pairs to obtain positive samples; introducing an interference expression or a fact error into the enhanced sample question and answer pair to obtain a negative sample; adjusting the positive and negative sample proportion according to the large model accuracy; and taking the positive sample and the negative sample as training data, and performing parameter fine tuning based on the pre-trained large model and the professional knowledge base to obtain a professional large model. According to the method, the adaptability and robustness of a large model in different scenes are improved, and the occurrence probability of irrelevant information and wrong answers is reduced.

The invention relates to a difficult negative sample generation method, an intention recognition method, electronic equipment and a storage medium. The method comprises the following steps: extracting a keyword from a positive sample corresponding to a target intention; a target combination is constructed through the keyword, and the target combination is a phrase used for describing key information of the target intention; inputting a context guide description corresponding to the target intention and the target combination into a large language model (LLM) to obtain a candidate negative sample; and screening the candidate negative samples through semantic similarity to determine difficult negative samples. Through the method and the device, the technical problem of low quality of difficult negative samples dug in related technologies is solved, and the technical effect of improving the quality of the difficult negative samples is achieved.

The present application relates to a model training method and an apparatus. The model training method includes: processing a text in a corpus to obtain a plurality of samples, where the plurality of samples includes a plurality of positive samples and a plurality of negative samples; building a plurality batches of training sets; inputting each batch of training set into a model for training, and obtaining a category of each sample in each batch of training set; and obtaining an overall loss value according to label information and the category corresponding to each sample, a sample quantity and an overall loss function in each batch of training set, and adjusting the model according to the overall loss value to obtain a trained model.

The invention relates to the technical field of image classification and recognition, in particular to a semi-supervised classification method and a semi-supervised classification system for lumbar diseases based on semantic contrast fusion uncertain sensing, and adopts a semi-supervised classification network model as a semi-supervised classification framework for lumbar disease recognition. The cross entropy loss of the labeled samples and the consistency loss of the unlabeled samples are used as the loss of the model; selecting the labels with the prediction confidence greater than a threshold value and the prediction uncertainty less than the threshold value as pseudo labels; taking the labeled data with the same label and the unlabeled data with the false label as a positive sample pair, and training by adopting a supervised comparative learning method to obtain comparative learning loss of the model; semi-supervised model loss, uncertainty loss and comparative learning loss are used as total loss functions for joint training of the model. The problems that semantic information of annotated data is not utilized to constrain unannotated data, and a pseudo tag has confirmation deviation are solved, and the lumbar disease recognition accuracy is improved.

The invention relates to the field of image recognition, and provides a positive and negative sample algorithm combined power transmission line high-altitude anomaly detection method. The objective of the invention is to improve the accuracy and robustness of detection. By comprehensively utilizing the advantages of the positive sample algorithm and the negative sample algorithm, the method effectively solves the problems of background complexity and abnormal data collection. Firstly, a detection picture and a base picture are cut and masked and focused on a power transmission line area; next, a positive sample algorithm is used to identify an abnormal region, while a negative sample algorithm is focused on detecting a specific type of anomaly. Through a multi-stage filtering mechanism including target detection filtering and template matching filtering, false alarms are effectively reduced, and the detection effect is improved. According to the method, a memseg algorithm and a YOLO model are adopted in model training, and an anomaly filtering model is introduced to detect common misinformation targets such as birds and vegetation. Finally, by combining positive and negative sample results, high-precision and high-recall-rate detection performance is realized.

The invention discloses a thermal power plant risk early warning method and system based on deep learning, and relates to the technical field of artificial intelligence. The method comprises the following steps: firstly, based on target risk state parameters and historical time sequence data of all monitoring parameters, aiming at all target risk associated monitoring parameters and all deep learning models, applying all positive sample data of corresponding parameters, and optimizing hyper-parameters of corresponding models based on an optimization algorithm; obtaining an optimal search result which corresponds to the hyper-parameter and is used for minimizing an objective function which adopts a product of an output error condition index value and a time length index value required for calculation, and then obtaining a target risk occurrence prediction model of each monitoring parameter based on the optimization result; according to the method, the prediction probability of the target risk event occurring at the future moment is obtained through model estimation, then the final prediction probability is obtained through superposition calculation, and finally when the final prediction probability reaches the preset probability threshold value, the risk early warning action is triggered and executed, so that the risk early warning accuracy and timeliness can be improved.

The invention relates to a knowledge base construction method and device and computer equipment. The method comprises the following steps: in response to a construction request for a domain knowledge base of a target domain, screening a new corpus set of the target domain through a text classification model; the text classification model is obtained by training in advance based on a positive sample formed by a historical corpus set of a target field and a negative sample formed by corpus sets of other fields matched with the category of the target field; performing relation extraction on the basis of the screened new corpus set to obtain a relation between a plurality of entities, and performing error information correction on the relation between the plurality of entities; and constructing a domain knowledge base of the target domain according to the corrected multiple entities and the relationship among the entities. By adopting the method, the quality of the domain knowledge base can be improved.

The invention discloses a watermark imperceptible embedding and recovering method based on a deep learning network structure. Firstly, a mask guide watermark embedding scheme is designed, a mask generation module is constructed by using a residual dense feature extraction module and an attention mask generation module, and a watermark is adaptively guided to be embedded into an image texture rich area so as to improve the invisibility of the watermark. And secondly, constructing a watermark decoding network based on comparative learning, and by comparing a loss function, taking the decoding features of the same watermark image under different noise conditions as positive samples and taking the decoding features of different watermark images as negative samples so as to enhance the consistency of the decoding features, thereby improving the robustness of the watermark. The deep learning network structure can improve the robustness of the watermark in a real screen shooting scene, and has a huge application value in copyright protection and traceability tracking.

The invention belongs to the technical field of industrial operation and maintenance term processing, and provides a big language model training method based on a knowledge graph, and the method comprises the steps: carrying out the cooperation with operation and maintenance personnel in advance to set a fault dictionary, collecting fault instances, carrying out the classification and preprocessing, extracting causal trigger words of the fault instances, and setting a text dependency relationship rule. Splitting and extracting a triple of the fault instance, and importing to form a knowledge graph; mapping the instance attribute into a node feature vector, outputting an embedded representation of a node, and fusing the embedded representation with the multi-modal feature vector to form a fused feature embedded representation; replacing semantic and degree words with fault instances in the knowledge graph to generate positive samples, randomly replacing subjects, objects and causal relationships to generate negative samples, inputting the negative samples into a semantic encoder, mapping the negative samples to the same metric space, and calculating similarity to obtain a comparison loss value to adjust graph convolutional network parameters; and after the fusion feature embedding representation retrieval of the fault instance is carried out and a subgraph is generated, maintenance suggestions are given step by step along a causal chain.

The invention discloses a small target detection method based on scale perception label distribution and context enhancement. The method comprises the following steps of: 1, organizing an image file and a corresponding annotation file to obtain image data, and dividing the image data into a training image and a test image to form a required data set; 2, preprocessing and training the image data through feature extraction and small target detection capability of an optimization model, and storing the weight; and step 3, testing the test image according to the weight stored by training, verifying the small target detection capability of the model in a real scene through evaluation on a test set by using the weight stored by training, and adjusting or optimizing model parameters. According to the invention, sufficient positive samples can be reasonably distributed for the targets of different scales for training according to the target scale change, so that the influence on the detection precision of the targets of other scales is reduced on the basis of improving the small target detection precision; the problem that a traditional multi-scale detection network fails to challenge small targets is effectively solved.

A method for controlling an artificial intelligence (AI) device can include obtaining, via a processor in the AI device, an AI model trained on a dataset that includes a majority class and at least one minority class, generating, via the processor, at least one evaluation metric for the AI model based multiplying a first score for positive samples of a target class within the dataset by a number of negative samples of the target class within the dataset and multiplying a second score for the negative samples within the dataset by a number of the positive samples, and outputting, via an output unit in the AI device, the at least one evaluation metric. Also, the method can further include adding the trained AI model to a pool of trained AI models and selecting a best AI model from the pool for deployment based on the at least one evaluation metric.

The invention discloses a protein generation model optimization method based on deep learning, and relates to the technical field of protein generation. The optimization method comprises the following steps: generating a candidate protein electron density distribution diagram based on the structural characteristics of a target spot by adopting a pre-trained diffusion model; converting the candidate protein electron density distribution diagram into a corresponding amino acid sequence; determining a functional index value corresponding to the amino acid sequence, and constructing a feedback data set; setting a reward threshold value, and marking the feedback data set as a positive sample data set and a negative sample data set according to the reward threshold value; constructing a utility function taking the reward threshold as a reference point, and respectively calculating utility values of the positive sample data set and the negative sample data set; and performing iterative optimization on the diffusion model according to the utility value. By adopting the technology provided by the invention, the dependence on preference paired data of large-scale and high-quality protein sequences can be avoided, and the performance of the protein generation model can be effectively and continuously improved.

The invention provides a potential risk event intelligent analysis method based on comparative learning, and belongs to the technical field of risk event intelligent analysis. A memory unit is regarded as a prototype of a normal video mode, and on the basis of memorizing video data features, contrast learning is introduced into the memory unit for training. A data enhancement strategy is adopted: different video sections in the same video belong to positive samples; data enhancement is performed on a video, and spatio-temporal information of the video, such as disorder, random rotation of video frames, continuous repetition of single video frames and the like, is destroyed as a negative sample sequence. Reconstruction errors are used as a standard for judging the drowning event, the similarity between an input video stream and a memory unit is used as an auxiliary standard for judging the drowning event, and the specific mixed decision-making method is completed in a weighting mode. According to the method, the general problem of network reconstruction and the problem of a memory model in anomaly detection are solved.

The present disclosure provides a fault prediction method, a fault prediction model training method, a computing device, a storage medium, and a computer program product. The fault prediction method comprises: obtaining abnormal log data and unit attribute information of a service processing unit; determining an abnormal event sequence on the basis of the abnormal log data; and inputting the unit attribute information and the abnormal event sequence into a fault prediction model to obtain a fault prediction result of the service processing unit, wherein the fault prediction model is obtained by performing training on the basis of a positive sample, a first sample label corresponding to the positive sample, a negative sample, and a second sample label corresponding to the negative sample, the positive sample comprises a positive sample abnormal event sequence and sample unit attribute information, and the negative sample comprises a negative sample abnormal event sequence and sample unit attribute information. The training data of the fault prediction model is richer, so that the accuracy of prediction results of the fault prediction model during applications is improved.

The invention provides a cross-modal image-text retrieval method and system based on adaptive comparative learning. The method comprises the steps that two boundary parameters are introduced into probability distribution of each pair of positive samples, the probability of an image I corresponding to T is calculated, semantic consistency of each pair of positive samples is adjusted, and distribution characteristics of negative samples are distinguished and cloned; defining and calculating a saliency score of each batch of samples, and selecting negative sample sequences with the maximum saliency score and the minimum saliency score as an observation clone negative sample and a saliency negative sample; according to a Gaussian discriminant analysis method, predicting all potential clone negative samples of each batch of samples and generating anchor point samples; and iteratively updating the two boundary parameters according to the two boundary conditions of the similarity of the anchor point samples. According to the method, potential clone negative samples in each training batch are predicted by utilizing Gaussian discriminant analysis without additional training for clone negative samples easily appearing in an image-text data set. And a foundation is laid for subsequent further learning of potential high-dimensional key semantics of the cloned negative sample.

The invention relates to an underwater target detection method and device based on comparative learning and a medium, and the method comprises the following steps: obtaining an underwater image, inputting a trained underwater target detection model based on comparative learning, and obtaining an underwater target detection result; wherein the underwater target detection model based on comparative learning comprises a backbone network, a feature pyramid network and a classification regression network which are connected in sequence, and a target detection precision evaluation module and a comparative feature enhancement module which are connected with the backbone network; wherein the target detection precision evaluation module is used for calculating detection scores of different image enhancement methods, and generating a positive sample and a negative sample corresponding to an original image according to the detection scores; and the comparative feature enhancement module is used for guiding the backbone network to extract features close to the features of the positive samples through a comparative learning method according to the positive samples and the negative samples. Compared with the prior art, the underwater target detection precision can be further improved.

The invention relates to a mineralization prediction method for synthesizing rare data based on a mineralization space structure. The method comprises the following steps: collecting and preprocessing geological data based on a prediction demand; based on the geological background of the to-be-studied area, constructing a metallogenic prediction model, determining a prediction variable and creating a prediction layer; performing mineralization space structure analysis, synthesizing a preset number of prediction variables of rare data by combining the prediction variables of the to-be-studied area, and further generating a plurality of training data sets; adopting a plurality of training data sets to respectively train the constructed metallogenic prediction model so as to predict the metallogenic probability of the to-be-researched area; and performing performance evaluation on the metallogenic prediction model, and delineating a prospecting target region based on a prediction result of the metallogenic prediction model with relatively good performance. According to the metallogenic prediction method disclosed by the invention, the nuclear density analysis is carried out on the space distribution of the known ore deposit, so that a certain number of positive samples can be synthesized under the condition of considering the geological connotation, the training of the metallogenic prediction model is facilitated, and the prospecting prediction precision is improved.