基于多模态数据的AI模型决策方法、装置及电子设备
By combining multimodal data and real user network graphs, the training samples of AI models are purified and expanded, solving the problem of poor performance caused by a lot of dirty data in the training samples of AI models, and achieving higher quality model training and decision-making.
CN120781082BActive Publication Date: 2026-07-17HANGZHOU ZHITUOXING TECHNOLOGY CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HANGZHOU ZHITUOXING TECHNOLOGY CO LTD
- Filing Date
- 2025-07-17
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
The presence of dirty data in the training samples of AI models leads to poor training results.
Method used
By acquiring multimodal data that integrates text, time series, and images, as well as real user review data, user profiles and real user network graphs are generated. The learning samples of the initial AI model are purified, and the credibility assessment and data expansion are carried out by interacting with users through intermediate AI models, thereby optimizing the target AI model.
Benefits of technology
It improves the authenticity and representativeness of learning samples, reduces the impact of dirty data, enhances the robustness and generalization ability of the model, and improves the accuracy and reliability of the model.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN120781082B_ABST
Abstract
本申请提供了一种基于多模态数据的AI模型决策方法、装置及电子设备，涉及人工智能技术领域，解决了AI模型训练样本中的脏数据较多导致AI模型训练效果较差技术问题。该方法包括：获取融合有文本、时序及图像的多模态数据以及与多模态数据相关的真实用户评论数据；根据真实用户评论数据及对应评论的多模态数据生成真实用户评论数据对应的目标用户的用户画像；基于多个目标用户对应的用户画像利用多模态数据和真实用户评论数据之间的关系生成多个目标用户的用户真实网络图谱；将用户真实网络图谱添加至初始AI模型的学习样本，得到第一学习样本，并利用第一学习样本对初始AI模型进行模型训练。
Need to check novelty before this filing date? Find Prior Art