Distributed problem solving system based on federated intelligent agent and excitation method thereof

Abstract

The invention discloses a federated agent-based distributed problem solving system and an excitation method thereof, and belongs to the technical field of artificial intelligence, and the system comprises a plurality of participant nodes, each node is provided with one or more specially-trained large language models, and the participant nodes are used for processing a specific problem domain; the central server is equipped with a high-performance large language model and is responsible for coordinating each participant node and integrating and optimizing answers returned from each node; through an intelligent distribution algorithm, model independence guarantee, dynamic learning and optimization and incentive mechanism design, the problem solving efficiency and quality are improved, and a continuously advanced and self-optimized ecological system is promoted through an incentive mechanism, so that each participant can effectively improve the efficiency and quality of problem solving while protecting own interests. And the optimization and progress of the whole system are jointly promoted. The mode is particularly valuable for modern enterprises or organizations needing to deal with a large number of complex and specialized problems, and has extremely high practicability and popularization value.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and in particular relates to a distributed problem solving system based on a federated intelligent agent and an incentive method thereof. Background Art

[0002] In existing AI applications, large language models are often deployed in a single server or data center. Although this centralized deployment method is easy to manage and maintain, it also has disadvantages such as long response time, high cost, and privacy issues. In addition, a single model often cannot cover all areas of knowledge, and when encountering professional or marginal problems, the model may not work well. Summary of the invention

[0003] The purpose of the present invention is to propose a distributed problem-solving system based on a federated intelligent agent and its incentive method in order to solve the above-mentioned problems.

[0004] On the one hand, in order to achieve the above-mentioned purpose, the present invention adopts the following technical solution: a distributed problem solving system based on a federated agent, comprising:

[0005] Multiple participant nodes, each of which is equipped with one or more large language models specially trained to handle a specific problem domain;

[0006] The central server, which is equipped with a high-performance large language model, is responsible for coordinating the various participant nodes, as well as integrating and optimizing the answers sent back from each node.

[0007] As a further description of the above technical solution:

[0008] The central server uses advanced algorithms to dynamically determine and assign problems to the most appropriate nodes.

[0009] As a further description of the above technical solution:

[0010] The model of each participant node runs in an isolated environment.

[0011] As a further description of the above technical solution:

[0012] The large language model on the central server integrates the answers, learns from each problem-solving process, and continuously optimizes its answer integration and problem parsing capabilities.

[0013] On the other hand, in order to achieve the above-mentioned purpose, the present invention adopts the following method: a distributed problem solving method based on a federated agent, which comprises the following steps:

[0014] 11) Question upload: When participants encounter complex questions, they upload the question description and related data to the central server. The uploaded questions can be text descriptions, images, or other forms of data;

[0015] 12) Problem distribution: The central server parses the received problems and distributes them to all participant nodes. When distributing problems, the server can select the most suitable node group to solve the problem based on the nature of the problem and the expertise of each node;

[0016] 13) Local model reasoning: Each participant node uses a local large language model to independently reason about the question. Each node comes up with a preliminary answer based on its own expertise and model capabilities.

[0017] 14) Answer feedback and preliminary screening: The participant nodes transmit the inferred answers back to the central server, which first screens the answers provided by each node and removes obviously irrelevant or erroneous content;

[0018] 15) Answer integration: The central server uses its large language model to further analyze and integrate the screened answers. At this stage, the server not only merges the correct information, but also reorganizes and optimizes the information;

[0019] 16) Results distribution: The final integrated answer will be sent back to the participant who asked the question, providing a detailed and accurate solution.

[0020] On the other hand, in order to achieve the above-mentioned purpose, the present invention adopts the following method: an incentive method for a distributed problem-solving system based on a federated agent, comprising the following steps:

[0021] 21) Contribution evaluation: The central server evaluates the contribution of each node based on the innovation, effectiveness and influence of the answer provided by each node;

[0022] 22) Reward distribution: Based on the evaluation results, the system rewards nodes through virtual currency, increased data access rights, and more computing resources.

[0023] The comprehensive evaluation and incentive mechanism can evaluate the contribution of each node in the problem-solving process in real time and issue rewards accordingly. The rewards are in various forms, including but not limited to virtual currency, better computing resource allocation, data access rights, etc., which effectively stimulates the enthusiasm of participants and their motivation for continued participation.

[0024] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0025] In the present invention, through intelligent distribution algorithm, model independence guarantee, dynamic learning and optimization and incentive mechanism design, not only the efficiency and quality of problem solving are improved, but also an ecosystem of continuous progress and self-optimization is promoted through incentive mechanism, so that each participant can jointly promote the optimization and progress of the entire system while protecting their own interests. This model is particularly valuable for modern enterprises or organizations that need to deal with a large number of complex and specialized problems, and has extremely high practicality and promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A block diagram of a distributed problem-solving system based on federated agents.

[0027] Figure 2 Flowchart of a distributed problem-solving method based on federated agents.

[0028] Figure 3 Flowchart of an incentive method for a distributed problem-solving system based on a federated agent. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] See also Figure 1 The present invention provides a technical solution: a distributed problem solving system based on a federated agent, comprising:

[0031] Multiple participant nodes, each equipped with one or more large language models specially trained to handle a specific problem domain; these models may focus on different technical or subject areas, such as medical, legal, or engineering problems, allowing the system to cover a wider range of knowledge;

[0032] The central server is equipped with a high-performance large language model, which is responsible for coordinating the nodes of each participant and integrating and optimizing the answers sent back from each node. The model continuously improves its ability to integrate questions and filter information through continuous learning and updating.

[0033] The central server uses advanced algorithms to dynamically determine and assign problems to the most appropriate nodes, ensuring that the problems are handled by the most capable models. This intelligent matching mechanism can greatly improve the efficiency and accuracy of problem solving;

[0034] The model of each participant's node runs in an isolated environment, ensuring the independence of the model and the security of the data. The data and knowledge of the participants do not need to be exposed, thus protecting the intellectual property rights and business secrets of the participants.

[0035] The large language model of the central server integrates the answers, learns from each problem-solving process, and continuously optimizes its answer integration and problem-solving capabilities. This continuous learning mechanism enables the system to adapt to complex and changing problem environments and continuously improve performance.

[0036] Example:

[0037] S11: Problem upload: When participants encounter complex problems, they upload the problem description and related data to the central server. The uploaded problems can be text descriptions, images, or other forms of data;

[0038] S12: Problem distribution: The central server parses the received problems and distributes them to all participant nodes. When distributing problems, the server can select the most suitable node group to solve the problem based on the nature of the problem and the expertise of each node;

[0039] S13: Local model reasoning: Each participant node uses the local large language model to independently reason about the question. Each node comes up with a preliminary answer based on its own expertise and model capabilities.

[0040] S14: Answer feedback and preliminary screening: The participant nodes transmit the inferred answers back to the central server, which first screens the answers provided by each node and removes obviously irrelevant or erroneous content;

[0041] S15: Answer integration: The central server uses its large language model to further analyze and integrate the screened answers. At this stage, the server not only merges the correct information, but also reorganizes and optimizes the information to improve the accuracy and practicality of the answers.

[0042] S16: Results distribution: The final integrated answers will be sent back to the participants who asked the questions, providing detailed and accurate solutions;

[0043] S21: Contribution evaluation: The central server evaluates the contribution of each node based on the innovation, effectiveness, and influence of the answer provided by each node;

[0044] S22: Reward distribution: Based on the evaluation results, the system rewards nodes through virtual currency, increased data access rights, and more computing resources. This is intended to motivate nodes to continuously optimize local models and increase participation.

[0045] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A distributed problem solving system based on federated agents, characterized in that: include: Multiple participant nodes, each of which is equipped with one or more large language models specially trained to handle a specific problem domain; The central server, which is equipped with a high-performance large language model, is responsible for coordinating the various participant nodes, as well as integrating and optimizing the answers sent back from each node.

2. A distributed problem solving system based on federated agents according to claim 1, characterized in that: The central server uses advanced algorithms to dynamically determine and assign problems to the most appropriate nodes.

3. A distributed problem solving system based on federated agents according to claim 1, characterized in that: The model of each participant node runs in an isolated environment.

4. A distributed problem solving system based on federated agents according to claim 1, characterized in that: The large language model on the central server integrates the answers, learns from each problem-solving process, and continuously optimizes its answer integration and problem parsing capabilities.

5. A distributed problem solving method based on federated agents as claimed in any one of claims 1 to 4, characterized in that: The steps include: 11) Question upload: When participants encounter complex questions, they upload the question description and related data to the central server. The uploaded questions can be text descriptions, images, or other forms of data; 12) Problem distribution: The central server parses the received problems and distributes them to all participant nodes. When distributing problems, the server can select the most suitable node group to solve the problem based on the nature of the problem and the expertise of each node; 13) Local model reasoning: Each participant node uses a local large language model to independently reason about the question. Each node comes up with a preliminary answer based on its own expertise and model capabilities. 14) Answer feedback and preliminary screening: The participant nodes transmit the inferred answers back to the central server, which first screens the answers provided by each node and removes obviously irrelevant or erroneous content; 15) Answer integration: The central server uses its large language model to further analyze and integrate the screened answers. At this stage, the server not only merges the correct information, but also reorganizes and optimizes the information; 16) Results distribution: The final integrated answer will be sent back to the participant who asked the question, providing a detailed and accurate solution.

6. A method for motivating a distributed problem-solving system based on a federated agent as claimed in any one of claims 1 to 4, characterized in that: The steps include: 21) Contribution evaluation: The central server evaluates the contribution of each node based on the innovation, effectiveness and influence of the answer provided by each node; 22) Reward distribution: Based on the evaluation results, the system rewards nodes through virtual currency, increased data access rights, and more computing resources.

Images