大数据时代的人工智能范式

发布时间：2018-03-27 11:42

本文选题：脑科学　切入点：感知　出处：《江苏大学学报(自然科学版)》2017年04期

【摘要】：针对传统人工智能范式存在的知识获取成本高、质量差、只能模拟低级智能等弊端,受大数据全样本统计、近似搜索和相关推理的启发,提出了4种AI范式,分别以感知、数据、脑科学、认知为中心.以感知为中心的AI范式主要研究一组分布的、松散耦合的主体如何协同运用它们的知识、技能、信息,为尽可能好地实现各自的或全局的目标或规划;以数据为中心的AI范式的形成是数据从量变到质变的过程,其核心是让数据说话;以脑科学为中心的AI范式旨在打造基于信息通信技术的综合性研究平台;以认知为中心的AI范式研究的目的是用大数据来解决"心智"问题.范式决定了AI的实现途径,不同的范式各有特点、相互补充,无法用一种范式统一全部,最后把AI的各种范式归结为科学研究4大范式.
[Abstract]:Aiming at the disadvantages of the traditional artificial intelligence paradigm, such as high cost of knowledge acquisition, poor quality, and only simulating low-level intelligence, and inspired by big data's total sample statistics, approximate search and related reasoning, four AI paradigms are proposed to perceive, respectively. Data, brain Science, Cognitive Center. The AI paradigm centered on perception focuses on how a group of distributed, loosely coupled agents can work together to use their knowledge, skills, and information. The formation of data-centric AI paradigm is the process of data changing from quantity to quality, the core of which is to let data speak. The AI paradigm centered on brain science aims to create a comprehensive research platform based on information and communication technology, while the cognitive paradigm aims to solve "mental" problems with big data. Different paradigms have their own characteristics and complement each other, so they cannot be unified with one paradigm. Finally, the AI paradigms are reduced to four paradigms of scientific research.
【作者单位】：南通大学计算机科学与技术学院;南通先进通信技术研究院;中拓联兴信息技术有限公司;
【基金】：国家自然科学基金资助项目(61340037,61171132) 南通大学自然科学基金资助项目(14Z008) 南通大学-南通智能信息技术联合研究中心科研项目(KFKT2016B06)
【分类号】：TP18

【相似文献】