Artificial intelligence system and method based on multi-repetition number group operation rule
By mapping data features into independent dimensions and synthetic dimensions of multiple repetitive numbers, and introducing measurement constraints and Nott conservation layers, the problem of dimensional catastrophic and poor interpretation of existing AI technologies in high-dimensional data processing and multimodal information fusion is solved, achieving more efficient data processing and stronger physical interpretability.
Patent Information
- Application Number
- CN202510338290.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-21
- Publication Date
- 2025-06-27
AI Technical Summary
Existing AI technologies have problems of dimensional catastrophe and poor interpretation when dealing with high-dimensional data, multimodal information fusion and coordinated decision-making in distributed systems.
By mapping data features into independent dimensions and synthetic dimensions of multiple repetitive numbers, the orthogonal group structure of multiple repetitive numbers is used to model complex relationships, and the measurement constraints and Nott conservation layers are introduced to improve the physical interpretability of the model.
It improves the efficiency of high-dimensional data processing, reduces the amount of model parameters, and improves the physical interpretability of the model.
Smart Images
Figure CN120218280A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of artificial intelligence, and specifically relates to an AI system and method based on the operation rules of Multi-Complex Number Groups (MCNG), which are used for optimizing machine learning models, multi-modal data processing, and complex system modeling. Background Art
[0002] The existing AI technologies have the following problems when dealing with high-dimensional data, multi-modal information fusion, and distributed system collaborative decision-making: (1) Curse of dimensionality: Traditional tensor operations are difficult to efficiently model the non-linear relationships between multi-dimensional orthogonal features. (2) Poor interpretability: Black-box models cannot explicitly express the dialectical relationships between multiple dimensions (such as time-energy conservation) through mathematical structures. The multi-complex number group theory provides a mathematical basis for the above problems through orthogonal dimension synthesis, measure constraints, and Noether's law mapping, but it has not been applied to the field of AI. Summary of the Invention
[0003] 1. Core Innovation Points
[0004] (1) Embedding multi-complex number groups into AI models: Map data features to the independent dimensions and synthetic dimensions of multi-complex numbers (such as tri-complex numbers ), and model complex relationships through their orthogonal group structure (homotopy period 3 / 8).
[0005] (2) Measure-driven adaptive learning: Use measures to constrain the model parameter space and solve the overfitting problem.
[0006] (3) Noether conservation layer: Introduce conservation constraints based on Noether's law (such as energy-time conservation) in the neural network to improve the physical interpretability of the model.
[0007] 2. Technical Effects: The high-dimensional data processing efficiency is increased by 50%, and the number of model parameters is reduced by 70%.
[0008] 3. Technical Solutions
[0009] a. System Architecture
[0010] (1) Input layer: Input multi-modal data (images, text, time series signals).
[0011] (2) Encoder: Encode multi-modal data (images, text, time series signals) into the form of multi-complex numbers, single-complex numbers , double-complex numbers , tri-complex numbers . For example: RGB channels of an image → single-complex number independent dimension time series signals of → double-complex number Synthetic Dimension 。
[0012] (3)MCNG Operation Layer: Through the commutative law and the almost symplectic structure to achieve cross-dimensional feature fusion. Use the Poincaré algebra to generate rotation-invariant features.
[0013] (4)Measure Constraint Module: Apply a threshold to the output to filter out the noise dimensions.
[0014] (5)Noether Conservation Layer: Introduce a conservation term in the loss function to ensure that the model follows physical laws (such as momentum conservation).
[0015] (6)Output Layer: Give the decision result.
[0016] b. Application Scenarios
[0017] (1)Map the permission level to the independent dimensions of triple complex numbers ( , , ), and achieve the irreversible verification of cross-chain transactions through the synthetic dimension .
[0018] (2)Game AI Decision: The character behavior is modeled by the common dimensions of double complex numbers ( , , ), and dynamically switch between combat / exploration modes. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] The drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure, but do not constitute a limitation to the present disclosure. In the drawings: Figure 1 is a system architecture diagram of an AI system and method based on the operation rules of multi-complex number groups shown in an exemplary embodiment of the present disclosure.
[0021] DETAILED DESCRIPTION OF THE EMBODIMENTS, The following will describe in detail the specific embodiments of the present disclosure with reference to the drawings. It should be understood that the specific embodiments described herein are only for explaining and understanding the present disclosure, and are not used to limit the present disclosure.
[0022] Exemplary embodiments will be described in detail herein, and examples thereof are shown in the accompanying drawings. When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. On the contrary, they are merely examples of methods consistent with some aspects of the present disclosure as detailed in the appended claims.
[0023] Embodiment: Multi-Complex Number Convolutional Neural Network (MCN-CNN)
[0024] (1) Decompose the input image into four independent dimensions (R / G / B / Alpha channels → , , ).
[0025] (2) Use synthetic dimension convolutional kernels to extract cross-channel features.
[0026] (3) Prune redundant feature maps through measurement to reduce the computational amount by 80%.
Claims
1. An artificial intelligence system based on multiple repeated number groups, characterized in that: It includes multiple complex number encoding layer, orthogonal group operation layer, measure constraint module and Noether conservation layer.
2. The system as claimed in claim 1, wherein the multiple-repetition encoding layer encodes the multimodal data (image, text, time series signal) into multiple-repetition form, a repetition , double repetition , triplicate .
3. The system of claim 1, wherein the orthogonal group operation layer is implemented by the commutative law Almost symplectic structure Implement cross-dimensional feature fusion using Poincare algebra Generate rotation-invariant features.
4. The system according to claim 1, wherein the measurement constraint module outputs Apply a threshold to filter out the noise dimension.
5. The system of claim 1, wherein the Noether conservation layer is obtained by Constrain model output.