773 results about "Cold start" patented technology

The invention discloses a personalized learning resource recommendation method and system based on multi-agent collaboration and a dynamic knowledge graph, and the method comprises the steps: constructing the dynamic knowledge graph, enabling nodes to be associated with teaching resources (videos, test questions, teaching plans, PPT and the like), and enabling edges to represent the logic relation between the resources; collaborative decision is made through four layers of agents: a target determination agent generates a learning target based on student historical learning data and a graph node state; the path planning agent plans a learning path in combination with the target and the learner model; the resource screening agent matches personalized resources from the path nodes; the user portrait intelligent agent updates the learner model in real time; and finally, generating a dynamic recommendation result and feeding back the optimized knowledge graph. Through multi-agent hierarchical collaboration and dynamic interaction with the knowledge graph, the problems of cold start, incomplete resource coverage and path stiffness of a traditional recommendation system are solved, and precise and adaptive learning resource recommendation is realized.

The invention relates to an article sequence recommendation method and system based on collaborative filtering and an LLM perspective, and the method comprises the steps: obtaining a user historical data set and an article data set, and carrying out the preprocessing, and obtaining an article title similarity matrix, a user behavior sequence, and historical collaborative filtering interaction information; historical collaborative filtering interaction information is enhanced through a large language model; constructing a sequence recommendation model, inputting the user behavior sequence into the sequence recommendation model for training, and correcting the deviation of the user behavior sequence through comparative learning according to the item title similarity matrix and the enhanced historical collaborative filtering interaction information to obtain a trained sequence recommendation model; and inputting the user behavior sequence of the user into the trained sequence recommendation model for prediction, calculating prediction scores, and generating a recommendation list according to score sorting, thereby completing article sequence recommendation. According to the method, the performance of the sequence recommendation system is greatly improved by solving the cold start problem.

The invention relates to a cross-domain knowledge migration cold start recommendation method based on dynamic intention perception. A system model of the method comprises a decoupling feature extractor based on graph convolution, an intention bridging network, a self-adaptive knowledge fusion mechanism and a recommendation generation unit. The method comprises the following steps: firstly, decomposing expressions of a user and an article into a plurality of intention subspaces through a decoupling feature extractor; then, establishing a soft mapping relation between intention subspaces of a source domain and a target domain by using an intention bridging network to realize intention alignment of fine granularity; the migration degree of source domain knowledge is dynamically adjusted through an adaptive knowledge fusion mechanism, and differentiated migration strategies are adopted for different users and articles; and finally, stable learning and smooth knowledge migration of the intention mapping relation are ensured by adopting a three-stage progressive training strategy of a recommendation generation unit. The problems of non-correspondence of intention semantics, negative migration and the like in traditional cross-domain recommendation are effectively solved, the recommendation effect is remarkably improved, and the method is particularly suitable for a cold start scene with sparse target domain data.

The invention provides an intelligent service recommendation method based on multi-dimensional scene perception and dynamic portrait modeling. The intelligent service recommendation method comprises the steps that 1, distributed edge computing nodes are deployed, space-time tetrad data operated by a user are collected in real time, user behaviors and labels are stored, and a high-dimensional recommendation database is built; 2, constructing a dynamic portrait engine, and constructing a user long-term behavior pattern library; 3, deploying a real-time streaming and offline double-engine architecture, fusing real-time scene matching and offline portrait prediction, and performing personalized function and user deep demand analysis; step 4, for new users, synchronously calling geofences to obtain regional hot services, and realizing cold start optimization based on meta reinforcement learning in parallel; and 5, establishing an intelligent recommendation closed loop associated with weather characteristics. According to the method, multi-dimensional features such as space-time tetrad data, user tag information and dynamic interest weights are fully utilized, and the real-time performance, interpretability and generalization ability of a recommendation system can be effectively improved.

The invention relates to the technical field of educational resource recommendation, and discloses an educational resource intelligent recommendation method and system based on big data driving, and the system comprises a data fusion processing module, a portrait modeling module, a resource feature engine module, an intelligent recommendation core module and a closed-loop feedback module. By fusing knowledge state, cognitive ability and interest preference three-dimensional portraits, capturing user learning ability evolution in real time, dynamically updating knowledge mastery by adopting a knowledge tracking model, and perceiving interest migration in combination with an attention mechanism, the problem of learning cold start in a new field is solved, recommendation coverage range and accuracy are improved, and user experience is improved. The cognitive load sensitive ant colony optimization algorithm is designed, the learning path continuity is guaranteed through a heuristic function, the learning path structure reasonability is optimized, the cognitive burden of a user is reduced, a personalized knowledge attenuation model is constructed by fusing an Ebbinghaus forgetting curve, the knowledge long-term retention rate is increased, and the review resource release accuracy is enhanced.

An online system uses a machine-learned language model (e.g., an LLM) to improve multilingual search capabilities. The system generates a prompt for the LLM that includes a set of search queries in a first language along with their context, as well as a request for translating these queries into a second language. This prompt is sent to a model serving system, which executes it through the LLM and returns translated queries in the second language. Additionally, the concierge system accesses a first set of features derived from the search results in the first language, and updates these features based on the newly translated search queries to create a second set of features. These translated queries and the second set of features are then used to train a search model optimized for queries in the second language.

The invention relates to the technical field of data fusion, and discloses a multi-modal multi-source heterogeneous data fusion method and system.The multi-modal multi-source heterogeneous data fusion method comprises the steps that a multi-modal prototype network is constructed, different modal features are extracted, and a semantic mapping relation between modals is established; a memory module is constructed, and multi-modal characterization of historical key samples is stored through sample value evaluation; executing diversity sampling to reserve a rare abnormal mode; modal balance processing is implemented, and contribution weights of different modal gradients are dynamically adjusted; knowledge distillation is applied, and general fusion knowledge is extracted from a data-rich scene to assist small sample decision making; the method solves the problems of low fusion efficiency of small sample and cold start scenes, disastrous forgetting in incremental learning, unbalanced modal learning rate and rare mode retention under long-tail distribution, and is suitable for predictive maintenance, personalized recommendation and medical health monitoring in the manufacturing industry and scenes requiring small sample learning and incremental updating.

The invention belongs to the technical field of precision marketing and intelligent recommendation, and discloses a B2B customer deep insight and precision reaching method based on a multi-modal large model, and the method comprises the specific steps: S1, carrying out the fusion and deep analysis of multi-modal data; s2, performing customer multi-dimensional preference modeling; s3, an intention and emotion analysis engine; s4, generating a personalized marketing strategy; s5, recommendation execution and verbal skill optimization; s6, feedback collection and reward function drive optimization; and S7, applying a cold start solution and transfer learning. According to the invention, through integration of multi-mode data of client official websites, news information, social media dynamics, mail communication, product images and conference recording, panoramic insight of clients from a business level to behavior details is realized; in combination with deep analysis of a pre-trained large model on texts, images and audios, explicit demands can be obtained, and industry features, cultural characteristics and potential concerns of customers can be understood.

The invention provides a model training and information replying method and device, a storage medium and a program product, and relates to the technical field of computers. The method comprises the steps of performing continuous pre-training on a base model based on a first training sample to obtain a basic model; performing cold start supervision fine tuning training on the basic model based on the second training sample to obtain a first supervision fine tuning model; performing multiple reasoning based on the third training sample, the target information and the to-be-trained model to obtain a reasoning result; in the Mth reasoning process, the target information comprises information obtained after a target tool determined by previous M-1 reasoning is called; optimizing the to-be-trained model based on the reasoning result to obtain a first reinforcement learning model; and based on the general recognition data and reasoning data output by the first reinforcement learning model, carrying out general recognition alignment training to obtain a target model. According to the method, the target tool can be called to obtain the required target information, so that the information output by the large model is more comprehensive.

The invention discloses a large language model semantic enhancement collaborative filtering recommendation method based on comparative learning, and belongs to the field of recommendation systems.The method comprises the steps that a user-item bigraph is constructed, user-item semantic information collaborative information is provided through GNN, user-item semantic information is extracted through cue words, a deterministic topology view enhancement strategy is adopted, and user-item semantic information collaborative filtering recommendation is achieved. According to the method, a semantic neighbor extension view is generated, a semantic neighbor reconstruction view is generated, collaboration and semantic information alignment are performed, double-view structure alignment is performed, and a loss function is integrally trained, so that the accuracy and the cold start capability of a recommendation system are remarkably improved through collaborative graph structure learning and semantic enhancement.

The invention discloses a method for deploying an edge dynamic DAG (Directed Acyclic Graph) server-free function on line to realize quick starting, and belongs to the field of edge computing. The method comprises the steps of 1, defining a dynamic DAG and establishing a modeling criterion of function preheating processing, and 2, constructing a dynamic DAG model with preheating preparation and function scheduling, and the model achieves the minimization of the overall expected execution time through preheating and function deployment. And step 3, converting a quadratic programming problem into a convex optimization problem by processing a quadratic constraint condition. And for the converted problem, a search algorithm based on random rounding is adopted to solve an optimal solution. And step 4, designing an online preheating and function scheduling algorithm. The algorithm includes function weight calculation, warm container adjustment, and optimal container selection to handle multiple online arrival requests. According to the method, through an online preheating and dynamic scheduling strategy, the cold start problem is effectively relieved, and the execution efficiency of the server-free function in the edge computing environment is improved.

The invention relates to the technical field of intelligent recommendation, and discloses an AI intelligent matching method based on a knowledge graph, and the method comprises the steps: constructing a quaternary knowledge graph containing a time dimension; identifying legal entities and semantic relationships by adopting an entity identification and relationship extraction technology; multi-level semantic features are extracted through a two-layer progressive semantic matching algorithm of a grammar layer, a semantic layer and a reasoning layer; constructing lawyer ability portraits based on a heterogeneous graph neural network and a time sequence perception graph convolution technology; progressive matching calculation is adopted, the optimal matching weight is learned through a multi-layer attention mechanism, and dynamically optimized intelligent matching is achieved. The technical problems of cold start, insufficient semantic understanding ability and poor timeliness processing ability in the existing legal consultation matching system can be solved, and the matching precision and the user satisfaction are improved.

The invention discloses a modeling method of an industrial fault diagnosis model. The modeling method comprises the following steps: S1, preprocessing original data; s2, establishing an automatic sample generation labeling mechanism based on the thinking chain; s3, establishing a state transition mapping layer; s4, establishing a fault sample filtering mechanism; s5, fault diagnosis modeling based on LLMs is carried out; and S6, evaluating the method. The general industrial intelligent fault diagnosis model established by the invention is not only suitable for cold start zero sample scenes of various environmental conditions, but also can overcome the limitation of system scale, environmental conditions and topological structures on model applicability, solves the problems that samples cannot be collected in advance during industrial field cold start, the model precision is low and the generalization is poor, and improves the fault diagnosis efficiency. The method is of great significance in avoiding economic loss and major safety accidents caused by the fact that equipment faults are not found in time.

The invention relates to the technical field of text generation, provides an NL2SQL method and system based on thinking reasoning, and aims to enable a model to interact with a database environment by introducing a reinforcement learning mechanism to obtain reward signal feedback. The system comprises a table field recall component and a structured query language generation component. The table field recall component fills table field information into a preset template to construct a structured text, then vectorization processing is carried out, the structured text is stored into an elastic search engine, a relevant table is searched according to user query, and finally specific fields are screened through a field selector model. And the structured query language generation component receives the screening result, generates a structured query language in combination with user query, and performs retrieval replacement on a filtering value. In the aspect of training, the method combines supervised fine-tuning cold start and reinforcement learning reasoning training, and the model performs reasoning and thinking according to an input context through trial and error learning. In the reinforcement learning training process, a reward function iteration optimization model based on task execution result correctness is adopted.

The invention discloses an electric power detection method and system for multi-modal enhanced fine tuning, and relates to the technical field of vision-driven intelligent electric power inspection, and the method comprises the steps: 1, obtaining an electric power inspection image, and marking the image; 2, inputting the marked image into the initial vision-language model, and carrying out cold start supervision fine tuning to train the initial vision-language model to obtain a model subjected to supervision fine tuning; 3, performing reinforced fine tuning on the model subjected to supervised fine tuning, and updating model parameters to obtain a PowerGPT-R1 model; and 4, evaluating the performance of the PowerGPT-R1 model based on the test set, and verifying the reliability of power detection. According to the method, the detection precision is remarkably improved under the condition of few samples, meanwhile, the multi-modal reasoning capability which a visual perception model does not have is shown, and the method has important application value in autonomous inspection of the infrastructure of the smart power grid.

The invention discloses a personalized teaching recommendation system for higher vocational students. The system belongs to the technical field of education informatization and artificial intelligence, and comprises a student portrait construction module, a teaching resource label management module, an intelligent recommendation engine module, a recommendation display and reason generation module, a teacher management and intervention module, and a learning process monitoring and feedback module. All the modules are mutually connected to form a recommendation-feedback-optimization closed-loop recommendation structure; according to the method, the recommendation accuracy and the post matching degree are improved; cold start, behavior-driven and semantic-driven multi-strategy fusion is supported, and recommendation accuracy and adaptability are improved; recommendation result interpretability and teacher cooperation capability are realized; dynamic evolution of portraits and self-adaption of recommendation strategies are realized; the system can be widely applied to various vocational education scenes. According to the invention, an intelligent recommendation system containing multi-module cooperation is constructed, and a complete closed loop of accurate layering of students, adaptive distribution of resources and interpretability of recommendation results is realized.

The invention discloses an academic group mining and directional recommendation system for regional natural fund, and relates to the technical field of computers. The academic group mining and directional recommendation system for the regional natural fund comprises a data management system, a knowledge graph module, a mixed recommendation algorithm module, an AI robot module and a user interaction interface module, and a multi-dimensional heterogeneous relation graph, an innovative cold start solution and a lightweight graph recommendation framework are constructed. Accurate mining, intelligent recommendation and efficient management of scientific researchers and projects are realized. According to the method, potential interests and cross-domain association requirements of users can be deeply captured, the problem of data sparsity in a cold start stage is effectively solved, and rapid deployment and real-time recommendation are realized through lightweight design. And powerful technical support and decision basis are provided for reasonable distribution and scientific research and innovation of regional natural funds.

The invention provides an advertisement cold start pushing method and system based on multi-source data fusion, and relates to the technical field of advertisement pushing. The advertisement cold start pushing method based on multi-source data fusion comprises the steps that user side information is collected, and the user side information comprises device information, behavior logs and third-party annotation data; extracting user portrait features according to the collected user equipment information, behavior logs and third-party annotation data, and extracting features of advertisement content needing to be pushed; the extracted user portrait features are matched with the content features of the advertisements needing to be pushed, and interest values of the users for all the advertisements are obtained respectively; and according to the obtained interest value of the user for each advertisement, judging the type of the advertisement needing to be pushed to the client, generating a pushing list of the advertisement, and completing pushing of the advertisement according to the pushing list. According to the method, the accuracy and the putting effect of the advertisement can be greatly improved, and better user experience and business return can be brought.

The invention provides a large language model instruction optimization method based on constraint perception self-learning reasoning, which can be applied to the technical field of natural language processing and artificial intelligence. The method comprises the steps that in the cold start training stage, a thinking chain example generated by a pre-training language model is utilized to assist a target large language model in recognizing constraints in a training instruction set, and the constraints in the training instruction set comprise hard constraints and soft constraints; performing supervision and fine tuning on the target large language model by utilizing the thinking chain example and the training annotation data so as to enable a thinking chain generation process of the target large language model to meet a hard constraint and a soft constraint; in the self-learning training stage, a constraint satisfaction evaluation mechanism is constructed by using the hard constraint and the soft constraint, and a thinking chain generation strategy of the target large language model is optimized through a reinforcement learning algorithm based on the constraint satisfaction evaluation mechanism.

The invention relates to a toxic text collection method and system based on retrieval enhancement generation, and the method comprises the steps: firstly obtaining target text data through a crawling platform, manually constructing a small-scale initial data set through a cold start mode, and injecting the initial data set into a knowledge base as basic data; and performing semantic retrieval on each batch of target texts, obtaining the first k texts with semantic similarity from the knowledge base, reasoning by adopting a plurality of large language models through thinking chain reasoning in combination with the texts, generating a toxic label, and labeling the target texts to be labeled in the current batch. And injecting the labeled target text into a knowledge base, continuously collecting and labeling data in an iteration mode, and continuously optimizing the performance of a retriever through an incremental learning mechanism in the iteration process so as to realize toxic text collection. According to the method, large-scale, fine-grained and high-consistency toxic text tagging corpora can be efficiently accumulated, and the problems of high tagging cost, inconsistent quality and poor expansibility in traditional toxic text data collection are remarkably relieved.

The invention discloses an electricity customer credit rating management method and system based on a differentiated service strategy, particularly relates to the field of energy service, and solves the problem of cold start user credit evaluation deviation and power service resource dynamic adaptation optimization. Seamless migration of credit scoring parameters and smooth transition of user portraits are realized by introducing multi-dimensional feature similarity analysis of virtual data and real data and a time-driven weight attenuation strategy, and continuity and adaptive optimization capability of an evaluation system are ensured; meanwhile, an intelligent service matching network is constructed based on the credit rating updated in real time, and a closed-loop service management and control system driven by data feedback is formed by dynamically adjusting a prepayment mode threshold value and an electric charge overdraft elastic interval strategy, so that the credit value of a user is deeply associated with the power resource configuration efficiency, and the user experience is improved. The operation risk and the service cost of a power grid enterprise are reduced, and finally collaborative optimization ecology considering energy safety management and user behavior guidance is established.

The invention discloses a federal learning-based cross-device brain grain model training system. The system comprises a terminal device layer, an edge coordination layer and a central server layer, the terminal equipment layer is responsible for collecting original electroencephalogram data and executing local lightweight training; the edge coordination layer is used for aggregating model updates of a plurality of terminal devices in the jurisdiction to generate a regionalized sub-model; according to the central server layer, a central server performs secondary optimization on each regional model to generate a global unified model; the encrypted global model and strategy configuration are pushed to all edge coordination layers; and the edge coordination layer pushes the optimized model and strategy to terminal equipment. Through a three-layer federated architecture and multi-dimensional optimization, the system efficiency is remarkably improved: the terminal equipment only uploads an encrypted model increment, and source data protection is realized in combination with differential privacy and homomorphic encryption; federal transfer learning shortens the cold start time of new equipment.

The embodiment of the invention provides a distributed function scheduling optimization method and device under a server-free architecture, and the method employs a dynamic priority scheduling strategy, dynamically adjusts the priority of a function task according to a predefined priority rule and real-time monitoring data, guarantees the priority processing of a key task, and improves the scheduling efficiency. And the system response speed and the resource utilization efficiency are improved. A distributed cache is adopted to store function tasks, execute environment snapshots and a dependency library, a preheating mechanism is supported, cold start delay is reduced, and function task execution efficiency is improved. Through a monitoring feedback mechanism, index data related to system operation are collected in real time, and according to the index data, a priority and a caching strategy are dynamically adjusted, so that adaptive optimization is realized, and the stability and robustness of the system are improved. And through dynamic priority scheduling and cache optimization, the system throughput and the resource utilization rate are improved.

The invention relates to a large language model dynamic routing method and device based on context learning model characterization and a readable storage medium. Query is embedded and mapped to a language model input space by using a projection model, semantic alignment is realized, a representative evaluation set covering multi-dimensional capability is automatically screened from a benchmark question bank, and the representative evaluation set is used for evaluating the multi-dimensional capability. Performance characteristics of the model on an evaluation set are efficiently obtained at a time, and high-quality context model capability representation is formed; then real-time query embedding and context model capability representation are combined, a lightweight routing language model is used for supervised learning, so that the fine-grained model distinguishing capability is achieved, an increment embedding updating mechanism is designed, and when a new model is accessed or an old model is upgraded, cold start can be rapidly completed only through a very small number of fixed questions, so that the efficiency is improved. The calculation and maintenance cost is greatly reduced, and the accuracy, real-time performance and flexible expansibility of model routing are effectively improved.

A collaborative transformation matrix learning system extends adaptive compression and encryption architectures through federated, privacy-preserving optimization. Each node analyzes local data distributions to generate anonymized distribution profiles using differential-privacy mechanisms, securely exchanging profiles and validated transformation matrices across a collaborative network. A trust and validation engine verifies mathematical properties and evaluates claimed performance metrics. Validated matrices are integrated into local optimization when trust and performance thresholds are satisfied. The system employs secure multi-party computation, homomorphic encryption, and conflict-resolution logic to ensure integrity of shared insights while preventing exposure of sensitive information. By combining collective learning with local adaptation, the invention accelerates convergence to optimal matrix configurations, mitigates cold-start inefficiencies, and improves compression-encryption efficiency and cryptographic strength across distributed deployments.

The invention discloses an intelligent recommendation system and algorithm based on a knowledge graph, and relates to the technical field of information recommendation. The intelligent recommendation system based on the knowledge graph and the dynamic updating method thereof are achieved mainly by taking dynamic updating of the knowledge graph, multi-modal information fusion, cold start problem solving and recommendation algorithm optimization as the core, and the purpose is to establish and maintain the association relation graph between the user and the article and combine real-time data analysis to achieve the intelligent recommendation system based on the knowledge graph and the dynamic updating method of the intelligent recommendation system based on the knowledge graph. And the recommendation algorithm is continuously optimized to improve the accuracy and timeliness of recommendation.

The invention discloses a reinforcement learning unmanned ship path control method based on double-track regulation and control random network distillation. The method comprises the operation steps that an unmanned ship builds a path tracking simulation environment and a kinetic model; the unmanned ship builds a core algorithm flexible action evaluation algorithm framework; the unmanned ship deploys a priority experience playback pool based on quality and success guidance; an uncertainty perception and risk perception mechanism is introduced into the unmanned ship; the unmanned ship builds a success rate-based reward attenuation and cold start module, and the unmanned ship calculates a total reward and designs a reward softening mechanism to smooth the total reward; the unmanned ship imports hyper-parameters of all the modules, starts training circulation in a simulation environment, and dynamically adjusts exploration intensity and the like; according to the method, uncertainty and risk indexes are introduced, the exploration intensity of the intelligent agent is controlled, the intelligent agent is prevented from making dangerous actions, and the robustness is improved; a priority experience playback pool based on quality and success guidance is introduced, high-quality samples are better played back, and strategy convergence is accelerated.

The invention provides a cold start acceleration method and device, electronic equipment and a medium, which can be widely applied to the technical field of computers, and the cold start acceleration method comprises the following steps: obtaining call information of a target function, and preprocessing the call information to obtain call characteristics; inputting the calling features into an online preheating model, predicting a function calling interval through a first preheating module in the online preheating model to obtain predicted calling time, and predicting the number of containers required for calling through a second preheating module in the online preheating model to obtain the number of predicted containers; determining a prediction decision according to the prediction calling time and the prediction container number; and deploying the containers according to the prediction decision, and creating a first number of preheated containers before the prediction call time to cope with a call request of the target function. The problem of request response delay caused by cold start in related technologies can be relieved, the request response speed is increased, and the user experience is improved.

The invention discloses a personalized cross-domain recommendation method and system based on federal learning, and the method provides a personalized cross-domain recommendation service for a user on the premise that original data of interaction between the user and an article and user parameters are kept locally. The method comprises two stages of federated training: stage 1, intra-domain users cooperatively train a single-domain score prediction model by using a neural collaborative filtering method; in the second stage, overlapping users of the two domains cooperatively train a migration module based on a multi-layer neural network to capture a mapping relation represented by potential user features between the two domains; besides, each layer of network of the cross-domain recommendation model is decomposed into a base vector and a personalized vector which respectively represent common knowledge among different users and unique knowledge of the users, and a local model obtained by final training can provide personalized recommendation services for registered users in a target domain, so that the recommendation efficiency is improved, and the user experience is improved. Meanwhile, the global model obtained through training provides effective initial recommendation for new users in the target domain, and the cold start problem in a recommendation algorithm is effectively relieved.

The invention relates to the field of customs document text feature recognition, in particular to a deep learning-based customs document text feature recognition method, which comprises the following steps of: preprocessing a real-time customs document text to obtain customs document text features; establishing a customs document text classification analysis model based on deep learning according to the customs document text features; and performing feedback adjustment processing by using the customs document text classification analysis model to obtain a customs document text feature recognition result, and establishing a dynamic threshold judgment system by considering multi-dimensional features such as data types, data capacity, text content, timestamps and historical data at the same time, so that the abnormal document recognition accuracy is improved, and the user experience is improved. Meanwhile, a cross validation mechanism of the data content classification model and the data content analysis model reduces the omission ratio of high-risk receipts in the test, and effectively solves the problem of cold start of training data.