JZUS - Journal of Zhejiang University SCIENCE

Frontiers of Information Technology & Electronic Engineering

Accepted manuscript available online (unedited version)

Miniaturized five fundamental issues about visual knowledge

Author(s): Yun-he Pan
Affiliation(s): Institute of Artificial Intelligence, College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China
Corresponding email(s): panyh@zju.edu.cn
Key Words:

Share this article to： More \|Next Paper >>>

Yun-he Pan. Miniaturized five fundamental issues about visual knowledge[J]. Frontiers of Information Technology & Electronic Engineering,in press.https://doi.org/10.1631/FITEE.2040000

@article{title="Miniaturized five fundamental issues about visual knowledge",
author="Yun-he Pan",
journal="Frontiers of Information Technology & Electronic Engineering",
year="in press",
publisher="Zhejiang University Press & Springer",
doi="https://doi.org/10.1631/FITEE.2040000"
}

%0 Journal Article
%T Miniaturized five fundamental issues about visual knowledge
%A Yun-he Pan
%J Frontiers of Information Technology & Electronic Engineering
%P 615-618
%@ 2095-9184
%D in press
%I Zhejiang University Press & Springer
doi="https://doi.org/10.1631/FITEE.2040000"

TY - JOUR
T1 - Miniaturized five fundamental issues about visual knowledge
A1 - Yun-he Pan
J0 - Frontiers of Information Technology & Electronic Engineering
SP - 615
EP - 618
%@ 2095-9184
Y1 - in press
PB - Zhejiang University Press & Springer
ER -
doi="https://doi.org/10.1631/FITEE.2040000"

Abstract
Chinese Summary
Academic Network
Reviewer Comment

Abstract: The five fundamental issues on visual knowledge are expression of visual knowledge, visual recognition, simulation of visual imagery thinking, learning of visual knowledge, and multiple expressions of knowledge. Our analysis shows that the distinct advantages of visual knowledge are its capacity to generate comprehensive imagery, its spatio-temporal evolution capacity and imagery display capacity. These are the features currently lacking in character knowledge and DNN. Integration of AI and CAD/CG/CV technologies will provide a vital foundation for the development of AI in terms of creation, prediction, and man-machine integration. The study of visual knowledge and multiple expressions of knowledge is the key to the development of visual intelligence and the main theory and technology to enable AI 2.0 to make major breakthroughs. It is a desolate, clammy, and fertile “Great Northern Wilderness,” but also a “depopulated land” full of hope and worthy of multi-disciplinary cooperation.

视觉知识的五个基本问题

潘云鹤
浙江大学计算机科学与技术学院人工智能研究所，中国杭州市，310027

概要：认知心理学早已指出，人类知识记忆中的重要部分是视觉知识，被用来进行形象思维。因此，基于视觉的人工智能（AI）是AI绕不开的课题，且具有重要意义。本文继《论视觉知识》一文，讨论与之相关的5个基本问题：（1）视觉知识表达；（2）视觉识别；（3）视觉形象思维模拟；（4）视觉知识的学习；（5）多重知识表达。视觉知识的独特优点是具有形象的综合生成能力，时空演化能力和形象显示能力。这些正是字符知识和深度神经网络所缺乏的。AI与计算机辅助设计/图形学/视觉的技术联合将在创造、预测和人机融合等方面对AI新发展提供重要的基础动力。视觉知识和多重知识表达的研究是发展新的视觉智能的关键，也是促进AI 2.0取得重要突破的关键理论与技术。这是一块荒芜、寒湿而肥沃的"北大荒"，也是一块充满希望值得多学科合作勇探的"无人区"。

关键词组：视觉知识表达；视觉识别；视觉形象思维模拟；视觉知识学习；多重知识表达

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference

[1]Brown TB, Mann B, Ryder N, et al., 2020. Language models are few-shot learners. https://arxiv.org/abs/2005.14165

[2]Pan YH, 2016. Heading toward artificial intelligence 2.0. Engineering, 2(4):409-413.

[3]Pan YH, 2019. On visual knowledge. Front Inform Technol Electron Eng, 20(8):1021-1025.

[4]Pan YH, 2020. Multiple knowledge representation of artificial intelligence. Engineering, 6(3):216-217.

[5]Tang SL, Zhang Q, Zheng TP, et al., 2018. Two step joint model for drug drug interaction extraction. https://arxiv.org/abs/2008.12704

[6]Xu DF, Zhu YK, Choy CB, et al., 2017. Scene graph generation by iterative message passing. Proc IEEE Conf on Computer Vision and Pattern Recognition, p.5410-5419.

[7]Zellers R, Yatskar M, Thomson S, et al., 2018. Neural motifs: scene graph parsing with global context. Proc IEEE Conf on Computer Vision and Pattern Recognition, p.5831-5840.

[8]Zhang NY, Deng SM, Zhang W, et al., 2020. Relation adversarial network for low resource knowledge graph completion. Proc Web Conf, p.1-12.

[9]Zhang SY, Tan ZQ, Zhou Z, et al., 2020. Comprehensive in-formation integration modeling framework for video titling. Proc SIGKDD Int Conf on Knowledge Discovery & Data Mining, p.2744-2754.

[10]Zhuang YT, Jain R, Gao W, et al., 2017. Panel: cross-media intelligence. Proc 25^th ACM Int Conf on Multimedia, p.1173.

Open peer comments: Debate/Discuss/Question/Opinion

<1>

- Go to

视觉知识的五个基本问题

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference