YucongDuan的个人博客分享 http://blog.sciencenet.cn/u/YucongDuan

博文

可解释、可信、负责任的多模态人工智能是什么?--DIKWP模型(超越ChatGPT)

已有 1958 次阅读 2023-2-11 15:23 |系统分类:论文交流

首先回答什么是人工智能(Artifical Intelligence, AI)?

整个数字世界和认知世界中的主体和对象都可以被一致映射到DIKWP模型的五个成分及其转换:数据图谱(Data Graph),信息图谱(Information Graph),知识图谱(Knowledge Graph),智慧图谱(Wisdom Graph),意图图谱(Purpose Graph)。

每个DIKWP成分对应认知的语义层面、人类语言的概念和概念实例层面:{语义层面、{概念,实例}}

模型<DIKWP Graphs>

::=(DIKWP Graphs)*(Semantics, {Concept, Instance})

::={  DIKWP Graphs*Semantics, DIKWP Graphs*Concept, DIKWP Graphs*Instance }

::={ DIKWP Semanics GraphsDIKWP Concept GraphsDIKWP Instance Graphs }


交互场景<DIKWP Graphs>

::={DIKWP内容模型(DIKWP Content Graph)包括:数据内容图谱(Data Content Graph),信息内容图谱(Information Content Graph),知识内容图谱(Knowledge Content Graph),智慧内容图谱(Wisdom Content Graph),意图内容图谱(Purpose Content Graph);

DIKWP认知模型(DIKWP Cognition Graph)包括:数据认知图谱(Data  Cognition Graph),信息认知图谱(Information  Cognition Graph),知识认知图谱(Knowledge  Cognition Graph),智慧认知图谱(Wisdom  Cognition Graph),意图认知图谱(Purpose  Cognition Graph)。

}



人工智能就是DIKWP交互的能力部分。

AI::=(DIKWP Graphs)*(DIKWP Graphs)*

狭义定义:人工智能是DIKWP交互中面向发展的消除重复的存储-计算-传输一体化迭代能力和跨DIKWP的面向(Open World Assumption)OWA范围转换能力。

我们正在进行ChatGPT的人工智能能力测试,期待与大家分享。


我们的相关工作举例:

http://www.yucongduan.org/patents.html


面向类型化资源的价值导向的存储与计算一体化优化系统 (已授权)
  Value Driven Storage and Computing Collaborative Optimization System for Typed Resources
   申请公布号:CN107734000A

全部相关发明专利列表见:

2019-2022三年间DIKWP团队授权的段玉聪第一发明人中国国家发明专利列表(共69件/241件)

请访问:https://blog.sciencenet.cn/blog-3429562-1354842.html



可解释、可信、负责任的多模态人工智能前言--DIKWP模型(超越ChatGPT)

from 《Explainable, trustworthy and responsive intelligent processing of biological resources integrating data, information, knowledge, and wisdom—Volume II》

KEYWORDS DIKW, DIKW graph, explainability and interpretability, trustworthy AI, responsive ability, knowledge, information retreival, data comprehension

机器翻译:

人工智能(AI)在生物和生物医学资源中的日益实践面临着多模态、交织、交互式生物和生物医学数据的可解释、可验证、响应式人工智能处理的挑战,这需要数据、信息、知识、智慧的集成 和目的 (DIKWP) 跨越客观内容和主观认知/目的。 数据、信息、知识和智慧之间的转换,为应对数据样本不完整、信息不足、无效知识的脆弱性和智慧策略失衡等不确定性,实现数据更精确、更稳健、可重现、更少重复操作提供了可能 研究主题和信息综合,通过多源推理和抽象更全面的知识再现性。 此外,随着 COVID 紧急情况,越来越多的注意力集中在平衡社会福利、文化道德和涉及隐私保护数据研究主题和法律信息使用的生物学实践,在国际政治和技术谈判的快速迭代下,朝着负责任的方向发展。 人工智能支持的人工智能治理实现公正、透明和公平。 本研究课题旨在收集最新的研究成果,致力于在一个统一的语义理解空间中构建多模态数据、信息、知识和智慧的集成和转换能力,以验证数据、检索信息、抽象 对信息进行知识假设,并进行平衡优化。

英文:

"The increasing practice of Artificial Intelligence (AI) in biological and biomedical resources faces challenges of the explainable, rustworthy, responsive AI processing of multi-modal, intertwined, interactive biological and biomedical data, which requires the integration of data, information, knowledge, wisdom and purpose (DIKWP) across objective content and subjective cognition/purpose. Transformations among data, information, knowledge and wisdom open possibilities to comply with uncertainties originating in the incompleteness of data samples, insufficiency of information, vulnerability of invalid knowledge and imbalanced wisdom strategies, towards achieving more precise, robust, reproducibility and less repeated operations of data Research Topic and information synthesis, and more comprehensive knowledge reproducibility through multiple sources reasoning and abstraction. Moreover, alongside the COVID emergency, more and more attention is focused on balancing social welfare, cultural moralities, and the biological practices involving privacypreserving data Research Topic and legal information usage, under rapid iterations of international political and technical negotiations, towards a responsible AI-enabled AI governance implementing justice, transparency and fairness. This Research Topic aimed to collect the latest research efforts devoted to building capabilities of integration and transformation of multi-modal data, information, knowledge and wisdom in an integrated semantic understanding space unifying subjective purposes and objective formalism, to validate data, retrieve information, abstraction on information to attain knowledge hypotheses, and balanced optimization."

原文见:


https://www.researchgate.net/publication/366863567_Editorial_Explainable_trustworthy_and_responsive_intelligent_processing_of_biological_resources_integrating_data_information_knowledge_and_wisdom-Volume_II




https://m.sciencenet.cn/blog-3429562-1375794.html

上一篇:答复:ChatGPT/AIGC怎么会打破资本主义-DIKWP数字经济视角
下一篇:基于DIKWP的信息计算与推理-超越ChatGPT系列

2 杨正瓴 许培扬

该博文允许注册用户评论 请点击登录 评论 (0 个评论)

数据加载中...

Archiver|手机版|科学网 ( 京ICP备07017567号-12 )

GMT+8, 2024-4-28 16:01

Powered by ScienceNet.cn

Copyright © 2007- 中国科学报社

返回顶部