424 results about "Task learning" patented technology

Provided in the present application is an intelligent question answering method based on a collaboration between a large language model and a knowledge graph, relating to the technical fields of artificial intelligence and natural language processing, the method comprising: decomposing a complex question into a plurality of simple questions, and analyzing the degree of association between the simple questions and a basic function so as to form a multi-hop reasoning path; automatically extracting structured information from the simple questions on the basis of a multi-task learning framework of a large model, so as to construct a knowledge graph; and constructing a cumulative reasoning learning framework on the basis of a logic reasoning large model, and performing iterative verification on a process result formed by the knowledge graph on the basis of the multi-hop reasoning path, so as to correct the reasoning path until a correct answer is inferred.

The invention relates to the technical field of psychological health, and discloses a psychological health multi-modal evaluation method, system and device and computer equipment, and the method comprises the steps: obtaining multi-modal data, and carrying out the feature extraction of the multi-modal data, and obtaining multi-modal features; performing emotion-oriented cross-modal attention mechanism analysis on the multi-modal features to respectively obtain a preset-dimension emotion scale vector, an emotion semantic feature, a multi-dimensional emotion feature and an uncertainty quantitative feature; fusing the preset dimension emotion scale vector, the emotion semantic feature, the multi-dimensional emotion feature and the uncertainty quantitative feature to obtain a multi-modal fusion feature; and obtaining a mental health assessment result based on the multi-modal fusion features and a preset multi-task learning framework. Through a cross-modal attention mechanism, deep semantic fusion of five modals of vision, audio, physiology, text and behavior is realized, and intelligent mapping from original multi-modal data to accurate psychological state judgment is realized through a deep learning technology.

The invention discloses a spine image key point detection algorithm based on multi-task learning. The algorithm comprises the following steps: constructing a multi-task deep learning network model comprising a segmentation branch and a key point positioning branch; a feature aggregation module based on multi-scale cavity convolution is embedded in the segmented branches, and the multi-scale cone feature extraction capability of the model is enhanced through splicing fusion of multiple cavity rate convolution branches and global pooling branches; a cross-task attention fusion module is introduced between the two branches, and bidirectional dynamic interaction and complementation between segmentation features and key point features are realized by generating and fusing first-order and second-order context attention maps; semantic alignment loss is designed in a training stage, collaborative optimization of two tasks is promoted by constraining the consistency of segmentation masks and key point heat maps in a high-level feature space, global information of segmentation and local information of key point detection are fully utilized, and the accuracy and stability of spine centrum key point positioning and the accuracy of a segmentation result are improved.

The invention particularly relates to a multi-task motion recognition method based on a physical constraint guide condition diffusion model, which comprises the following steps: acquiring original observation data of multi-task motion based on an inertial measurement unit and preprocessing to obtain preprocessed original observation data; constructing a conditional diffusion model, and training the conditional diffusion model by using the preprocessed original observation data based on a preset loss function to obtain a trained conditional diffusion model; performing data enhancement on a key motion state category determined based on the motion state category label corresponding to the original observation data according to the trained conditional diffusion model to obtain a key motion state category sample, and constructing a sample balance data set based on the key motion state category sample; and based on the sample balance data set and a preset multi-task learning network loss function, constructing a multi-task learning pedestrian motion recognition model and carrying out multi-task recognition. Therefore, the problems of few key state samples, unreasonable data and the like in motion recognition are solved.

The invention relates to the technical field of target behavior prediction systems based on causal inference and multi-task learning, and particularly discloses a target behavior prediction system based on causal inference and multi-task learning. The system comprises a central coordination server and a plurality of participant clients, constructs a global causal graph through a federated causal discovery algorithm in a collaborative manner, determines a causal feature subset of each prediction task, and carries out multi-task model training through a federated average algorithm under the constraint. In the process, differential privacy and homomorphic encryption technologies are comprehensively applied to protect data privacy. According to the method, more accurate causal discovery and more reliable prediction model training can be realized on the premise of protecting data privacy of all parties.

The invention discloses a multi-modal sensing fusion dam body structure state intelligent diagnosis and early warning system and method and application. The system comprises a multi-modal sensing network composed of a space deformation monitoring subsystem, an internal response monitoring subsystem and an environment quantity monitoring subsystem; a data fusion and feature extraction module based on a space-time diagram attention network, which is used for deeply mining a complex space-time coupling relationship between multiple measurement points and multiple physical quantities; the health state diagnosis module based on multi-task learning can synchronously output dam body structure health indexes and abnormal types of key areas; and a future state evolution prediction and multi-stage early warning module based on a Transform model. According to the method, digital twinning of the dam body is constructed, a physical entity and an information model are closely combined, a full-chain closed loop from multi-dimensional perception to intelligent diagnosis to prospective early warning is achieved, and the accuracy, comprehensiveness and timeliness of dam body structure state evaluation are improved.

Disclosed in the present invention is a road-scene depth completion method based on the guidance of semantic information and a color image. The method comprises: designing a dual-branch network consisting of a color-image-guided branch and a semantic-guided branch, and introducing semantic information into the network, so as to perform road-scene depth completion. The idea of multi-task learning is used, and a backbone is shared with a depth map prediction layer in the color-image-guided branch, such that the semantic information can be obtained simply by adding a semantic segmentation layer, without the need to add an entire network. Moreover, semantic labels generated by the semantic segmentation layer are fed into the semantic-guided branch as input, such that parameters of the layer can also be adaptively adjusted during network training. By integrating high-precision semantic information and an RGB image, the present invention completes a sparse depth map provided by LiDAR, thereby improving the accuracy and real-time performance of scene understanding in applications such as digital twin, virtual reality, digital infrastructure and intelligent transportation.

The invention relates to an aircraft flow field and aerodynamic performance joint prediction method and system based on multi-task learning. The method comprises the following steps: collecting flow field data sets of different symmetric airfoils; after data preprocessing is carried out on the flow field data set, an input and output data set is constructed; constructing a neural network model, wherein the neural network model comprises a backbone network, a branch network and a fusion network; taking the space time sequence coordinates as input of the backbone network, and outputting backbone features; taking the static input characteristics and the historical flow field time sequence data as inputs of the branch network, and correspondingly outputting flow field prediction branch characteristics and aerodynamic coefficient prediction branch characteristics respectively; outputting a predicted value through the fusion network; constructing a weighted loss function; training the neural network model; and carrying out neural network model lightweight processing, and carrying out flow field prediction. The flow field characteristic prediction of the aircraft under the flow condition is realized quickly and accurately.

Application of multi-task learning technique(s) to machine logic (for example, software) used to detect financial transactions that are fraudulent or at least considered likely to be fraudulent. Some embodiments include adjustments and / or additions to conventional multi-task learning techniques in order to make the multi-task learning techniques more suitable for use in fraud detection software. One example of this is compensation for class imbalances that are to be expected as between the likely-fraud and not-likely-fraud classes of data sets (for example, training data sets, runtime data sets).

The invention discloses an intelligent detection method and system for medicine components based on dynamic multi-task learning, and relates to the technical field of intelligent component detection. The method comprises the following steps: acquiring original spectral data of a medicine to be identified, and preprocessing the original spectral data; performing dynamic target detection on the preprocessed data according to a type rule, wherein a target comprises a regression task and a classification task; based on a multi-method fusion feature selection strategy, performing feature selection on the preprocessed data by using a plurality of feature selection methods to obtain component features; constructing a fusion network, dynamically activating a corresponding output head of the fusion network according to a target detection result, and training the fusion network by adopting a weighted mixed loss function; and processing the component features by using the trained fusion network to obtain a component detection result. According to the invention, highly-automatic, flexible and traceable dynamic prediction of the medicine components can be realized based on a multi-task learning mechanism.

The invention discloses a large language model control fine tuning method and system based on multi-task cooperative regulation and control, and belongs to the technical field of large language models. Generating a gating coefficient and an initial dynamic evaluation signal through the intelligent regulation and control network; a task difficulty index is obtained by combining multi-index fusion and historical moving average, and the sampling probability and the exclusive learning rate are dynamically adjusted; weighting the fusion gradient and carrying out back propagation to update parameters; and closed-loop feedback monitoring is carried out and parameters of the regulation and control network and the scheduling policy device are optimized. The system comprises a data coding module, a collaborative intelligent regulation and control module, a dynamic balance control module, a joint optimization module and a closed-loop feedback module. According to the method, gradient conflicts among tasks are relieved, the problems of convergence instability and performance imbalance are solved, the multi-task training efficiency, convergence stability and generalization ability of a large language model are improved, and the method can be widely applied to multi-class multi-task learning scenes.

A text sequence recommendation method and system based on a large language model is disclosed, belonging to the technical field of recommendation algorithms. The method includes: a data preprocessing stage, a large language model pre-training stage, a sequence model fine-tuning stage and a matching stage. According to this disclosure, a large language model is introduced into a text sequence recommendation task, so that text can be better modeled by utilizing rich pre-training corpus of the large language model; meanwhile, sequence modeling is performed on the text, the capability of sequence recommendations modeling in a large model is activated, an ID-based recommendation paradigm in a traditional recommendation algorithm is eliminated, and recommendation task learning processing is better performed in a cold start scenario and a knowledge transfer scenario; and finally, a recommendation result is finally optimized by a sequence model.

The invention discloses a federated learning and hybrid clustering network intrusion detection method and system, and relates to the technical field of network intrusion detection. The method comprises the following steps: a server issues a current global model parameter to a client participating in the current round of training; after each client receives the global model parameter, taking the global model parameter as an initialization parameter of a local model, performing training by using local data, introducing a clustering-based soft label generation mechanism and a classification and clustering parallel dual-task learning framework in the training process for training, and updating the local model parameter; the client encrypts and uploads the trained model parameters to the server; the server aggregates all client model parameters, generates a new generation of global model, and issues the parameters back to the client for next round of training; and after all rounds of training are finished, the server issues a final model for the client to carry out network intrusion discrimination. According to the method, the negative influence of data non-independent identical distribution is effectively overcome, and sparse attacks are accurately detected.

The invention provides an intelligent recommendation method based on a big language model thinking chain and a heterogeneous graph, and relates to the technical field of recommendation systems.The method comprises the steps that a dynamic heterogeneous data set is constructed; constructing a dynamic heterogeneous graph which comprises three types of nodes and four types of weighted edges, and dynamically updating an edge weight and a topological structure through a time window; aggregating the spatio-temporal features by adopting a dynamic heterogeneous graph neural network, and generating a feature representation fusing cross-domain association and time sequence evolution; extracting cross-domain association features by using a large language model, dynamically strengthening the weight of a cross-domain edge in the graph structure, and generating semantic enhancement features; the method comprises the following steps: starting from a user high-frequency interaction project, constructing a deepened and cross-domain extended reasoning path in a fusion domain, and generating recommendation features with both time sequence evolution and cross-domain logic by dynamically weighting recent nodes; dynamic isomerism and semantic features are fused, multi-task learning and time sequence loss optimization recommendation are combined, and interest changes are adaptively learned through a dynamic window.

The invention relates to the technical field of intelligent decision making, can be applied to business scenes such as financial science and technology and medical health, and discloses a data analysis method, device and equipment based on progressive task learning and a medium. A basic language model is initialized as an intra-domain data autonomous analysis agent, tasks with different difficulties are received to obtain analysis operation tracks and analysis result output, reward signals are generated based on quality feedback, agent parameters are updated in combination with demonstration data, and a trained agent is obtained through multiple rounds of training. And executing the target analysis task in the interactive data environment to obtain a target analysis result. According to the method, a training mechanism combining a progressive task system and a demonstration track is introduced, so that the intelligent agent can steadily improve the complex task processing capability, and autonomous analysis output facing real tasks is realized.

The invention discloses an end-to-end multi-task learning method and system based on Swin Transform, and belongs to the technical field of artificial intelligence and computer vision crossing, and the method comprises the steps: collecting low-quality paper document images which have physical degradation characteristics and cause substantial obstacles to information recognition, carrying out the corresponding text labeling of each image, and constructing a target data set; based on the target data set, adopting a backbone network capable of extracting multi-scale hierarchical features as a shared image encoder, and taking an image enhancement task and a text recognition task as two parallel downstream branches to construct an end-to-end multi-task learning network architecture; based on the output of the image enhancement task and the text recognition task, image enhancement loss and text recognition loss are calculated respectively, a joint loss function is constructed, and multi-task collaborative optimization is realized through end-to-end training; the robust recognition capability of fuzzy, continuous and low-quality handwritten characters is improved, and a recognition-friendly high-quality image is generated.

The invention relates to the field of artificial intelligence and self intelligence, and discloses a cross-space, cross-task and cross-ontology learning large model and reasoning method, the method comprises the following steps: a pre-training multi-modal large language model as a backbone network is used for processing visual and text input and generating a response; the intention bridging interface is used for compressing a hidden state output by the multi-mode large language model into semantic intention representation with a fixed length; the action strategy head is used for generating a continuous action sequence based on the semantic intention; the state encoder is used for encoding the body sensing state of the robot; and the motion encoder / decoder is used for embedding and reconstructing the motion. According to the method, cross-space migration, cross-task learning and cross-ontology generalization are realized in a single model, and the performance and generalization ability of digital space reasoning and physical space control are improved.

The invention discloses an electric power construction behavior safety detection method and system, and relates to electric power safety. The method comprises the following steps: collecting historical electric power work ticket data, and constructing a work ticket risk feature library; according to the work ticket risk feature library, training a risk identification model based on multi-task learning; collecting text data of the current electric power work ticket, extracting work content and security measure fields, and generating a text feature vector; searching optimal parameter configuration in a parameter space of the risk identification model by using a discretization doliolaria optimization algorithm; the method comprises the following steps: collecting real-time image data of a working site, carrying out target detection and feature extraction on the image data, identifying position coordinates of working personnel and types and wearing state marks of safety protection appliances, encoding a detection result into structured data, and generating an image feature vector; aiming at the lack of a dynamic association verification mechanism between an electric power work ticket text specification and an actual behavior of a work site in the prior art, the static requirement of a work ticket is converted into a dynamic behavior supervision standard.

The invention discloses a digital eyestrain detection method based on multi-task learning and an intelligent terminal. The method mainly solves the problems that the prior art depends on a single mode, only pays attention to a single task (for example, only the opening and closing state is detected), and comprehensive evaluation on the fatigue state of a user, especially the eye dryness condition is lacked. The invention provides a digital eye fatigue detection method based on multi-task learning and an intelligent terminal, and aims to realize high-precision fatigue and distance detection on low-cost hardware through fusion of multi-task learning and a monocular vision geometric model, solve the problems of limited precision of a single mode and redundancy of separate deployment calculation in the prior art, and improve the detection precision of the eye fatigue. The method is suitable for mobile terminals such as smart phones and tablet computers. After the scheme is deployed on the intelligent terminal, a user can be helped to keep a healthy screen use distance and a healthy screen display parameter, a healthy blinking habit is maintained, and dry eyes, astringent eyes and eyestrain are reduced.

The invention discloses a phosphorus chemical industry production equipment on-line state monitoring and intelligent diagnosis system based on multi-parameter fusion, and relates to the technical field of chemical industry equipment intelligent monitoring and diagnosis. Functional modules such as data input and preprocessing, equipment wear evaluation, real-time data analysis, fault prediction, personnel risk analysis and comprehensive diagnosis are deeply integrated, and the system accurately adapts to non-stationary characteristics of the phosphorus chemical process by dynamically adjusting a time window and a self-attention weight. The comprehensive diagnosis module innovatively constructs a causal diagram based on self-attention features, generates a structured diagnosis report by using a Transform decoder, and optimizes the diagnosis probability in combination with Bayesian reasoning, so that the accuracy, comprehensiveness and intelligence level of monitoring diagnosis are improved, accurate early warning can be realized, deep insight and optimization operation and maintenance decisions can be provided, and the system has a wide application prospect. And the production risk and economic loss are effectively reduced.

Methods, systems, and computer program products for knowledge graph based embedding, explainability, and / or multi-task learning may connect task-specific inductive models with knowledge graph completion and enrichment processes.

The invention discloses an original message binary feature extraction method based on a time sequence convolutional network. The method comprises the following steps: generating an embedded vector sequence of a message; constructing a time sequence convolutional network architecture; pre-training and classifier training are carried out through self-supervised pre-training, supervised fine tuning and multi-task learning; model interpretation is carried out, and quantized behavior features are extracted based on interpretation results; converting the features into a Snort / Suricata rule format, and establishing a mapping relation from the features to original message segments; and integration to an IDS / IPS engine is realized. The invention also provides an original message binary feature extraction system based on the time sequence convolutional network. A closed loop from feature discovery to automatic rule deployment is constructed, and the detection response efficiency and accuracy of complex network threats are remarkably improved.

The invention discloses a multi-working-condition industrial process soft measurement method based on multi-task learning and probability modeling, and aims to solve the problem of insufficient measurement precision caused by heterogeneous mixing of multi-working-condition process samples. The method comprises three core modules, namely a feature decoupling coding module, a hierarchical feature fusion module and a probability information aggregation module. Firstly, a spatial-temporal feature extractor is designed to explicitly decouple multi-working-condition data into working condition shared features and specific features, and hybrid feature expression and working condition recognition are achieved. Then, a hierarchical feature fusion module is constructed, deep fusion of information between working conditions is realized through a hierarchical expert gating network, and a complex interaction relationship between the working conditions is modeled; and finally, proposing a probability information aggregation strategy, inputting the fusion features into corresponding predictors, and weighting prediction results by using the working condition identification probability to generate final prediction output. According to the method, a classification task and a regression task are incorporated into a unified multi-task learning framework, and the good performance of multi-working-condition process performance index soft measurement is ensured.

This invention provides a method for fish passage behavior recognition based on a multimodal large model. First, imaging sonar image sequences and underwater optical video sequences of fish passing through fish passages or facilities are simultaneously acquired, and the multimodal data are aligned through time stamp synchronization and spatial calibration. Then, fish target detection, segmentation, and feature extraction are performed on the sonar image sequences and optical video sequences respectively to obtain the fish's spatial location, body length, water depth, motion state, morphological structure, and posture features. The obtained acoustic and optical features are input into a multimodal coding and fusion model based on the Transformer architecture, and a unified fish behavior representation vector is constructed through a cross-modal attention mechanism. During the training phase, model parameters are adjusted through a multi-task learning approach using cross-entropy loss, mean squared error loss, and contrastive learning loss, ultimately achieving fish passage behavior category recognition and prediction of individual and group passage difficulty scores.

The invention discloses a lightweight bridge and dynamic task vector federated multi-task learning method, which comprises a federated learning system, the system comprises a plurality of clients and a central server, and the method comprises the following steps: 1) initialization and distribution; 2) performing local training and processing on the client; 3) performing dynamic task vector decomposition and compression transmission; 4) aggregation and reconstruction of the central server side; and 5) updating and issuing the model. According to the method, the communication efficiency is improved in order of magnitude, the communication bottleneck is relieved fundamentally, the personalized performance of the model is remarkably enhanced and exceeds that of an existing baseline method, inter-task interference is effectively restrained, training stability and convergence are improved, excellent balance between communication overhead and calculation / personalized performance is achieved, and good adaptability and robustness are achieved.

The invention provides a method for automatically labeling work order types on the basis of agents, which comprises the following steps of: performing punctuation standardization processing on an original work order, and converting spoken and non-standardized work order texts into segmented word segments conforming to field specifications in combination with word segmentation in a power field dictionary; unifying and normalizing the segmented word segments through a preset synonym mapping table to obtain a standardized text sequence; the standardized text sequence is input into a bidirectional encoder expression model to output a semantic vector sequence, and deep semantic understanding of the work order text is achieved; related external information is called to be coded into a feature vector, and then the feature vector is fused with the semantic vector sequence through an attention mechanism to generate an enhanced semantic vector; multi-dimensional label prediction tasks are executed in parallel based on a multi-task learning architecture, and multi-class labels and probability distribution are output; the confidence coefficient is obtained by calculating the maximum value of the probability distribution, and the preset process is executed, so that automation and quality management and control of label generation are realized, and the problem of low efficiency of manual power work order processing in the prior art is solved.

The invention provides an operator data processing method and device based on multi-dimensional analysis. The method comprises the steps that after multi-source heterogeneous data are collected from multiple systems of an operator, the multi-source heterogeneous data are fused through a dynamic feature cascade mechanism, and fusion features are generated; modeling the fusion features by using a recurrent neural network to capture longitudinal correlation of the time sequence data; modeling the fusion features through multi-task learning to capture transverse correlation between data sources corresponding to the multi-source heterogeneous data; and generating business decision information based on modeling results of the longitudinal correlation and the transverse correlation. According to the method, deep mining of the potential value of the operator data is realized, and the mining accuracy and the commercial realization efficiency are improved.

The invention discloses an lncRNA and disease association prediction method based on multitask graph attention and information completion, and belongs to the field of data mining in bioinformatics. The method comprises the following steps: firstly, according to an lncRNA interaction spectrum set, a miRNA interaction spectrum set and MeSH description of a disease, respectively calculating to obtain an interaction spectrum similar matrix SL of the lncRNA, an interaction spectrum similar matrix SM of the miRNA and a semantic similar matrix SD of the disease; secondly, the similar information is combined with LMA, LDA and MDA to construct an lncRNA-miRNA heterogeneous network, an lncRNA-disease heterogeneous network and a miRNA-disease heterogeneous network; then, internal representation of similar information is obtained through a GCN to serve as initial embedding, high-order topological information of interaction nodes is extracted through a graph attention Unet model, and an incidence matrix is reconstructed through a bilinear decoder; and finally, constructing a multi-task learning framework, and realizing association prediction through an information completion method. Through experiments, the powerful performance of predicting potential correlation between the lncRNA and the disease by the model is effectively verified.

This invention relates to the field of speech translation technology, and particularly to a cross-transport obfuscation method and apparatus based on adaptive optimal transmission. The method includes: constructing an optimal transmission model within a multi-task general framework; performing attention-enhanced optimal transmission alignment on speech and text sequences; optimizing the attention-enhanced optimal transmission alignment based on a dynamic window strategy to obtain the alignment relationship between the speech and text sequences; fusing speech and text features of the speech and text sequences through a similarity-based adaptive fusion strategy; and combining a contrastive learning loss function with a multi-task learning framework, constructing positive and negative sample pairs, and combining the multi-task loss functions to obtain a unified optimization objective. This invention effectively reduces the representational differences between speech and text through dynamic window strategies, optimal transmission, and contrastive learning, achieving significant improvements in translation tasks for low-resource languages.

The invention discloses a multi-task learning-based driver driving-suitable credibility identification method and system, and relates to the technical field of multi-task learning, and the method comprises the steps: determining the multi-modal data of a driver based on the feature positions of a plurality of key face features, the corresponding feature forms and the real-time face image of the driver; the multi-modal data simultaneously contains the accurate head posture of the driver, the dynamic geometric information of the eyes and the mouth and the visual texture and expression details of fatigue, distraction or specific emotion on the face of the driver, and a preset credibility value is determined based on the previous driving event of the driver, a multi-task learning system and the current state of the driver; the corresponding driving early warning item is triggered based on the real-time comparison of the preset credibility value and the driving-suitable credibility value, so that the real-time performance and the accuracy of the driving early warning item are ensured.