logo

  • Academia Sinica / 
  • Sitemap / 
  • MH Login / 
  • 中文
  • 正體中文
    English
search
  • Events
    • >  News
    • >  Academic
  • About IMH
    • >  Introduction
    • >  Director’s remarks
    • >  Organization
    • >  Advisory board
    • >  Research plans
    • >  Research findings
    • >  Honors
    • >  Admin Staff
  • People
    • >  Research fellows
    • >  Corresponding Research Fellows
    • >  Adjunct research fellows
    • >  Postdoctoral fellows
    • >  Doctoral candidate fellows
    • >  Research Groups
  • Activities
  • Publications
    • >  Historical sources
    • >  Monographs
    • >  Bulletin
    • >  RWMCH
    • >  Conference Volumes
    • >  Other publications
    • >  Hu Shih Publications
    • >  eBooks
    • >  Non-IMH publications
    • >  Search
    • >  Order
  • Academic exchanges
    • >  List of Partner Institutions
    • >  Visiting scholars
    • >  Life and work
    • >  Visiting scholars program
  • Resources
    • >  Research Resources Links
    • >  Special displays
    • >  Video
    • >  Picture of the Day
  • Contact
    • >  Subscribe our RSS
    • >  FAQ
    • >  Contact us

 

Home >

當AI遇到歷史學

arrow iconDate(s): 2024/11/04

arrow iconTime: 10:00~12:00

*Venue: Archives 3rd Conference Hall

*Host: 連玲玲教授(中研院近史所研究員兼副所長)

*Speaker:Prof. 郭至汶(Assistant Professor, Department of Applied History, NCU)

*Organizer: IMH數位人文共同研究專題

演講摘要:
2022年3月,一篇刊登在權威科學期刊《自然》(Nature)的文章—“Restoring and attributing ancient texts using deep neural networks”,將人工智慧的方法應用於古典歷史的研究中。這篇文章以帕卡德人文學院(The Packard Humanities Institute)所提供的近八萬份古希臘銘刻文為資料集,利用深度神經網絡訓練出一模型“Ithaca”,用以恢復受損的古希臘銘刻文。根據研究團隊的統計,歷史學家經由Ithaca的輔助,其判讀受損銘刻文的正確率由25%提升至72%,大幅提高歷史學者對文本的掌握度。這項研究的問世,為人工智慧如何影響歷史學提供了絕佳的範例。本演講將以此為基礎,探討AI可以為歷史學做什麼、如何與歷史學擦出火花,並介紹AI的工具樣態、範例資源(例如:Programming Historian)等相關議題。



arrow iconPhotos:
event photo
event photo
event photo
event photo

Return
FB網站 RSS 2010優勝網站

Copyright 2016, Institute of Modern History, Academia Sinica. All Rights Reserved.

128 Academia Rd, Sec. 2, Nankang, Taipei 115201, Taiwan Tel:886-2-2782-4166 Fax:886-2-2789-8204

Privacy policy

Profile Protection