Method and system for performing geometric problem reasoning by using aligned multi-modal features
The method addresses the limitation of existing geometric problem solvers by aligning and fusing multi-modal features within the geometric problem reasoning model, resulting in improved accuracy and reasoning capabilities.
CN119721234AActive Publication Date: 2025-03-28GUANGDONG UNIV OF TECH
3 Cites 1 Cited by
Patent Information
- Application Number
- CN202411494117.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-24
- Publication Date
- 2025-03-28
- Estimated Expiration
- 2044-10-24
AI Technical Summary
Technical Problem
Existing methods for solving geometric problems using deep models lack the ability to fully utilize aligned multi-modal features, limiting their performance in understanding and reasoning about geometric issues.
Method used
A method that initializes a geometric problem reasoning model with pre-trained Chinese CLIP, processes multi-modal data through feature extraction, alignment, and fusion, and uses a dynamic word table for decoding to generate accurate reasoning results.
Benefits of technology
This approach enhances the accuracy of geometric problem solving by effectively aligning and fusing multi-modal features, improving the model's multi-step reasoning capabilities and reducing semantic distance between modal features.
✦ Generated by Eureka AI based on patent content.
Abstract
The invention relates to the field of natural language processing and machine learning, and discloses a geometric problem reasoning method and system using aligned multi-modal features, and the method comprises the following specific steps: initializing a training data set of geometric problems and a geometric problem reasoning model based on a pre-trained Chinese CLIP model; inputting the multi-modal data into a geometric problem inference model for training, and introducing a dynamic word list into a training process to obtain a trained geometric problem inference model; and obtaining a to-be-reasoned geometric problem and inputting the to-be-reasoned geometric problem into the trained geometric problem reasoning model to obtain a geometric problem reasoning result. The problem that in the prior art, cross-modal semantic representation cannot be fully utilized to solve geometric problems is solved, and the method has the advantage of being capable of generating more accurate reasoning results.
Need to check novelty before this filing date? Find Prior Art