News
home
News
home
๐Ÿง 

Explainable AI

Overview

์„ค๋ช… ๊ฐ€๋Šฅํ•œ AI(XAI, Explainable Artificial Intelligence)๋Š” ๋จธ์‹  ๋Ÿฌ๋‹ ์•Œ๊ณ ๋ฆฌ์ฆ˜์œผ๋กœ ์ž‘์„ฑ๋œ ๊ฒฐ๊ณผ์™€ ์ถœ๋ ฅ์„ ์ธ๊ฐ„์ธ ์‚ฌ์šฉ์ž๊ฐ€ ์ดํ•ดํ•˜๊ณ  ์ด๋ฅผ ์‹ ๋ขฐํ•  ์ˆ˜ ์žˆ๋„๋ก ํ•ด์ฃผ๋Š” ์ผ๋ จ์˜ ํ”„๋กœ์„ธ์Šค์™€ ๋ฐฉ๋ฒ•๋ก ์ž…๋‹ˆ๋‹ค. ์„ค๋ช… ๊ฐ€๋Šฅํ•œ AI๋ฅผ ์‚ฌ์šฉํ•˜๋ฉด AI ๋ชจ๋ธ, ์ด์˜ ์˜ˆ์ƒ๋œ ์˜ํ–ฅ ๋ฐ ์ž ์žฌ์  ํŽธํ–ฅ์„ ๊ธฐ์ˆ ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ์ด๋Š” AI ๊ธฐ๋ฐ˜ ์˜์‚ฌ๊ฒฐ์ •์—์„œ ๋ชจ๋ธ ์ •ํ™•์„ฑ, ๊ณต์ •์„ฑ, ํˆฌ๋ช…์„ฑ ๋ฐ ์ตœ์ข… ๊ฒฐ๊ณผ๋ฅผ ํŠน์„ฑํ™”ํ•˜๋Š” ๋ฐ ์œ ์šฉํ•ฉ๋‹ˆ๋‹ค. ์„ค๋ช… ๊ฐ€๋Šฅํ•œ AI๋Š” ๊ธฐ์—…์ด AI ๋ชจ๋ธ์„ ์ƒ์‚ฐ์— ํˆฌ์ž…ํ•  ๋•Œ ์‹ ๋ขฐ๊ฐ๊ณผ ์ž์‹ ๊ฐ์„ ์–ป๋Š” ๋ฐ ์žˆ์–ด์„œ ๋งค์šฐ ์ค‘์š”ํ•ฉ๋‹ˆ๋‹ค. AI ์„ค๋ช…๊ฐ€๋Šฅ์„ฑ์€ ๊ธฐ์—…์ด AI ๊ฐœ๋ฐœ์— ์ฑ…์ž„ ์žˆ๋Š” ์ ‘๊ทผ ๋ฐฉ์‹์„ ์ ์šฉํ•˜๋Š” ๋ฐ๋„ ๋„์›€์ด ๋ฉ๋‹ˆ๋‹ค.
AI๊ฐ€ ๊ณ ๋„๋กœ ๋ฐœ์ „ํ•จ์— ๋”ฐ๋ผ, ์ธ๊ฐ„์€ AI ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ๊ฒฐ๊ณผ ๋„์ถœ ๊ณผ์ •์„ ํŒŒ์•…ํ•˜๊ณ  ์ด๋ฅผ ์—ญ์ถ”์ ํ•ด์•ผํ•˜๋Š” ๋‚œ์ œ์— ๋ด‰์ฐฉํ•˜๊ฒŒ ๋˜์—ˆ์Šต๋‹ˆ๋‹ค. ์ „์ฒด ๊ณ„์‚ฐ ํ”„๋กœ์„ธ์Šค๋Š” ํ•ด์„์ด ๋ถˆ๊ฐ€๋Šฅํ•œ ์†Œ์œ„ ๋งํ•˜๋Š” "๋ธ”๋ž™๋ฐ•์Šค"๋กœ ์ „ํ™˜๋ฉ๋‹ˆ๋‹ค. ์ด๋Ÿฌํ•œ ๋ธ”๋ž™๋ฐ•์Šค ๋ชจ๋ธ์€ ๋ฐ์ดํ„ฐ๋กœ๋ถ€ํ„ฐ ์ง์ ‘ ๊ตฌ์ถ•๋ฉ๋‹ˆ๋‹ค. ๋˜ํ•œ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ๋งŒ๋“œ๋Š” ์—”์ง€๋‹ˆ์–ด๋‚˜ ๋ฐ์ดํ„ฐ ์‚ฌ์ด์–ธํ‹ฐ์ŠคํŠธ์กฐ์ฐจ๋„ ๊ทธ ๋‚ด๋ถ€์—์„œ ๋„๋Œ€์ฒด ๋ฌด์Šจ ์ผ์ด ๋ฐœ์ƒํ•˜๋Š”์ง€ ํ˜น์€ AI ์•Œ๊ณ ๋ฆฌ์ฆ˜์ด ํŠน์ • ๊ฒฐ๊ณผ๋ฅผ ์–ด๋–ป๊ฒŒ ๋„์ถœํ•˜๋Š”์ง€๋ฅผ ํŒŒ์•…ํ•˜๊ฑฐ๋‚˜ ์„ค๋ช…ํ•  ์ˆ˜๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค.
AI ๊ธฐ๋ฐ˜ ์‹œ์Šคํ…œ์ด ํŠน์ • ๊ฒฐ๊ณผ๋ฅผ ๋„์ถœํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ํŒŒ์•…ํ•  ์ˆ˜ ์žˆ์œผ๋ฉด ๋งค์šฐ ์œ ์šฉํ•ฉ๋‹ˆ๋‹ค. ์„ค๋ช…๊ฐ€๋Šฅ์„ฑ์€ ๊ฐœ๋ฐœ์ž๊ฐ€ ์‹œ์Šคํ…œ์ด ์˜ˆ์ƒ๋Œ€๋กœ ์ž‘๋™ ์ค‘์ธ์ง€๋ฅผ ๋ณด์žฅํ•˜๋Š” ๋ฐ ์œ ์šฉํ•˜๋ฉฐ, ๊ทœ์ œ ๋ฐฉ์‹์˜ ํ‘œ์ค€์„ ๋”ฐ๋ผ์•ผ ํ•  ์ˆ˜๋„ ์žˆ์Šต๋‹ˆ๋‹ค. ๋˜ํ•œ ์ด๋Š” ์˜์‚ฌ๊ฒฐ์ •์œผ๋กœ ์นจํ•ด๋ฅผ ๋‹นํ•œ ์‚ฌ๋žŒ๋“ค์ด ํ•ด๋‹น ๊ฒฐ๊ณผ์— ์ด์˜๋ฅผ ์ œ๊ธฐํ•˜๊ฑฐ๋‚˜ ์ด๋ฅผ ๋ฐ”๊ฟ€ ์ˆ˜ ์žˆ๋„๋ก ํ—ˆ์šฉํ•˜๋Š” ๋ฐ ์žˆ์–ด์„œ ์ค‘์š”ํ•  ์ˆ˜๋„ ์žˆ์Šต๋‹ˆ๋‹ค.
Explainable artificial intelligence (XAI) is a set of processes and methods that allows human users to comprehend and trust the results and output created by machine learning algorithms. Explainable AI is used to describe an AI model, its expected impact and potential biases. It helps characterize model accuracy, fairness, transparency and outcomes in AI-powered decision making. Explainable AI is crucial for an organization in building trust and confidence when putting AI models into production. AI explainability also helps an organization adopt a responsible approach to AI development.
As AI becomes more advanced, humans are challenged to comprehend and retrace how the algorithm came to a result. The whole calculation process is turned into what is commonly referred to as a โ€œblack box" that is impossible to interpret. These black box models are created directly from the data. And, not even the engineers or data scientists who create the algorithm can understand or explain what exactly is happening inside them or how the AI algorithm arrived at a specific result.
There are many advantages to understanding how an AI-enabled system has led to a specific output. Explainability can help developers ensure that the system is working as expected, it might be necessary to meet regulatory standards, or it might be important in allowing those affected by a decision to challenge or change that outcome. (Reference: https://www.ibm.com/kr-ko/watson/explainable-ai)

People

Search

Projects

์„ค๋ช…๊ฐ€๋Šฅํ•œ AI ๊ธฐ๋ฐ˜ ๋””์ง€ํ„ธํŠธ์œˆ (XAI-DTw) ์ž์œจ์šด์˜ ์„œ๋น„์Šค ๊ธฐ์ˆ  ๊ฐœ๋ฐœ

์ง€์› ๊ธฐ๊ด€ : ์‚ฐ์—…ํ†ต์ƒ์ž์›๋ถ€
์‚ฌ์—…๋ช… : ์ง€์‹์„œ๋น„์Šค์‚ฐ์—…๊ธฐ์ˆ ๊ฐœ๋ฐœ์‚ฌ์—…
1.
๋””์ง€ํ„ธํŠธ์œˆ ์ง€์‹์‚ฌ๋ฌผ์ธํ„ฐ๋„ท (IoT)์„ ํ†ตํ•ด ์‹ค์ œ ๋ฐ์ดํ„ฐ๋ฅผ ์ž…๋ ฅ๋ฐ›๊ณ , ์ธ๊ณต์ง€๋Šฅ (AI) ๊ธฐ์ˆ ์„ ํ†ตํ•ด ์ž…๋ ฅ๋œ ๋ฐ์ดํ„ฐ์— ๋Œ€ํ•œ ์˜ํ–ฅ์„ ์˜ˆ์ธก ๋˜๋Š” ์‹œ๋ฎฌ๋ ˆ์ด์…˜ํ•˜์—ฌ, ๊ทธ ์ถœ๋ ฅ๊ฐ’์„ ๊ธฐ๋ฐ˜์œผ๋กœ ๋ฌผ๋ฆฌ์  ๊ฐ์ฒด๋ฅผ ์ตœ์ ํ™”ํ•˜์—ฌ ์šด์˜ํ•  ์ˆ˜ ์žˆ๋„๋ก ์ง€์›
2.
์„ค๋ช…๊ฐ€๋Šฅํ•œ AI ์ธ๊ณต์ง€๋Šฅ์ด ์ž…๋ ฅ๋œ ๋ฐ์ดํ„ฐ๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ๋„์ถœ๋œ ํŠน์ • ํŒ๋‹จ๊ฒฐ๊ณผ์˜ ๊ณผ์ •๊ณผ ์ด์œ ๋ฅผ ์‚ฌ๋žŒ์ด ์ดํ•ดํ•  ์ˆ˜ ์žˆ๋„๋ก ์ œ๊ณตํ•˜๋Š” ๊ธฐ์ˆ ๋กœ์„œ, ์ด๋ฅผ ํ†ตํ•ด ๋””์ง€ํ„ธํŠธ์œˆ์ด ํŠน์ • ์ƒํ™ฉ์— ๋Œ€ํ•ด ์ž…๋ ฅ๋œ ๋ฐ์ดํ„ฐ๋ฅผ ์ฒ˜๋ฆฌ ๋ฐ ๋ถ„์„ํ•˜์—ฌ ์˜์‚ฌ๊ฒฐ์ • ๊ฒฐ๊ณผ (insights)๋ฅผ ๋„์ถœํ•˜๋Š” ๊ณผ์ •๊ณผ ๊ทผ๊ฑฐ๋ฅผ ํšจ๊ณผ์ ์œผ๋กœ ์ œ์‹œํ•  ์ˆ˜ ์žˆ๋Š” ์„œ๋น„์Šค ์ œ๊ณต
3.
์„ค๋ช…๊ฐ€๋Šฅํ•œ AI ๊ธฐ๋ฐ˜ ๋””์ง€ํ„ธํŠธ์œˆ ์‹ค์‹œ๊ฐ„ ์ˆ˜์ง‘ ๋ฐ์ดํ„ฐ๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ์‹ค๊ฐํ˜• 3D ๋ชจ๋ธ๋กœ ์ž๋™์ƒ์„ฑ ๋‚ด์žฌ๋œ ํ•™์Šต๋ชจ๋ธ์„ ํ†ตํ•ด ๋ฌธ์ œ์ƒํ™ฉ์— ๋Œ€ํ•œ ์›์ธ๊ณผ ๊ฒฐ๊ณผ ๋ฐ ํ•ด๊ฒฐ๋ฐฉ์•ˆ์„ ๋„์ถœ ๋„์ถœ๋œ ๊ฒฐ๊ณผ๋ฅผ ๊ธฐ๋ฐ˜์œผ๋กœ ๋Œ€์‘ํ•˜๋Š” ์ „ ๊ณผ์ •์„ ๋Šฅ๋™์ ์œผ๋กœ ์ˆ˜ํ–‰ํ•˜๋Š” ์ž์œจ์šด์˜ ๋””์ง€ํ„ธํŠธ์œˆ์˜ ์ƒ์„ฑ์—์„œ๋ถ€ํ„ฐ ํ™œ์šฉ๊นŒ์ง€์˜ ์ „ ๊ณผ์ •์„ ์ž์œจํ™”ํ•˜๊ณ , ์ด๋ฅผ ์œ„ํ•œ ์ˆ˜๋ฆฌ์ ์ด๊ณ  ๋…ผ๋ฆฌ์ ์ธ ๋ชจ๋ธ๋“ค์˜ ์‚ฌ๊ณ ์ฒด๊ณ„๋ฅผ ์‚ฌ์šฉ์ž์—๊ฒŒ ์ดํ•ดํ•  ์ˆ˜ ์žˆ๋Š” ๋ฐฉ๋ฒ•์œผ๋กœ ์ œ์‹œ

Publications

Search
Year
Title
Venue
Tag
Authors
URL
Date
Index
Award
2024
ํ•œ๊ตญCDEํ•™ํšŒ ๋™๊ณ„ํ•™์ˆ ๋Œ€ํšŒ
Artificial Intelligence
Explainable AI
Hyeokmin Lee, Byounghyun Yoo
2024/01/31 โ†’ 2024/02/01
2024
ํ•œ๊ตญCDEํ•™ํšŒ ๋™๊ณ„ํ•™์ˆ ๋Œ€ํšŒ
Artificial Intelligence
Virtual Reality
Computer Vision
Ji Hyun Seo, Byounghyun Yoo
2024/01/31 โ†’ 2024/02/01
2024
ํ•œ๊ตญCDEํ•™ํšŒ ๋™๊ณ„ํ•™์ˆ ๋Œ€ํšŒ
Artificial Intelligence
Explainable AI
Digital Twin
Machine Learning
Data Reconstruction
Youngkyu Kim, Qingkai Kong, Youngsoo Choi, Byounghyun Yoo
2024/01/31 โ†’ 2024/02/01
2023
ํ•œ๊ตญCDEํ•™ํšŒ ํ•˜๊ณ„ํ•™์ˆ ๋Œ€ํšŒ
Artificial Intelligence
Computer Vision
Image-based Modeling
XR
Virtual Reality
Ji Hyun Seo, Byounghyun Yoo
2023/08/23 โ†’ 2023/08/26
Best Poster Award
2023
ํ•œ๊ตญCDEํ•™ํšŒ ํ•˜๊ณ„ํ•™์ˆ ๋Œ€ํšŒ
Digital Twin
Artificial Intelligence
Machine Learning
Explainable AI
Data Reconstruction
Youngkyu Kim, Youngsoo Choi, Byounghyun Yoo
2023/08/23 โ†’ 2023/08/26
2023
IEEE International Conference on Acoustics, Speech and Signal Processing
Artificial Intelligence
Digital Twin
Jisoo Kim, Hyebin Ahn, Byounghyun Yoo
https://doi.org/10.1109/ICASSP49357.2023.10095412
2023/06/04 โ†’ 2023/06/10
SCOPUS
2022
ํ•œ๊ตญCDEํ•™ํšŒ ํ•˜๊ณ„ํ•™์ˆ ๋Œ€ํšŒ
Digital Twin
IoT
Explainable AI
Jisoo Kim, Muhammad Atif, Jee Young Moon, Chanhyuk Lee, Jaephil Choi, Byounghyun Yoo
2022/08/24 โ†’ 2022/08/26
2022
Journal of Computational Design and Engineering
IoT
Digital Twin
Machine Learning
Muhammad Atif, Shapna Muralidharan, Heedong Ko, and Byounghyun Yoo
https://doi.org/10.1093/jcde/qwac037
2022/05/23
SCIE
SCOPUS
2022
ํ•œ๊ตญ์ •๋ฐ€๊ณตํ•™ํšŒ ์ถ˜๊ณ„ํ•™์ˆ ๋Œ€ํšŒ
IoT
Artificial Intelligence
Explainable AI
Jisoo Kim, Jee Young Moon, Chanhyuk Lee, Byounghyun Yoo
2022/05/11 โ†’ 2022/05/13
2022
ํ•œ๊ตญCDEํ•™ํšŒ ๋™๊ณ„ํ•™์ˆ ๋Œ€ํšŒ
IoT
Artificial Intelligence
Explainable AI
Jisoo Kim, Jee Young Moon, Byounghyun Yoo
2022/02/09 โ†’ 2022/02/12
2021
Journal of Computational Design and Engineering
3D Scanning
Computer Vision
Machine Learning
Robotics
Digital Twin
Ji Hyun Seo, Inhwan Dennis Lee, Byounghyun Yoo
https://doi.org/10.1093/jcde/qwab049
2021/09/15
SCIE
SCOPUS
2020
Journal of Computational Design and Engineering
IoT
Digital Twin
Machine Learning
Muhammad Atif, Shapna Muralidharan, Heedong Ko, Byounghyun Yoo
https://doi.org/10.1093/jcde/qwaa048
2020/10/13
SCIE
SCOPUS
2020
IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)
Digital Twin
IoT
Artificial Intelligence
Computer Vision
Chanwoong Lee, Hyorim Choi, Shapna Muralidharan, Heedong Ko, Byounghyun Yoo, Gerard J. Kim
https://doi.org/10.1109/MFI49285.2020.9235269
2020/09/14 โ†’ 2020/09/16
2020
IEEE Transactions on Robotics
Computer Vision
3D Scanning
Machine Learning
Robotics
Digital Twin
Inhwan Dennis Lee, Ji Hyun Seo, Young Min Kim, Jonghyun Choi, Soonhung Han, Byounghyun Yoo
https://doi.org/10.1109/TRO.2020.2980161
2020/08/01
SCIE
SCI
SCOPUS
2019
ํ•œ๊ตญCDEํ•™ํšŒ ํ•˜๊ณ„ํ•™์ˆ ๋Œ€ํšŒ
Computer Vision
3D Scanning
Machine Learning
Robotics
Digital Twin
Ji Hyun Seo, Inhwan Lee, Byounghyun Yoo
2019/08/19 โ†’ 2019/08/22
Best Poster Award
About | News | People | Research | Publications | Facilities | Career | Contact