New Era in Data Protection for Machine Learning

2019-03-12
  • 3,264

*bldg. 2  NO.102/ 16:00~

 

Biography

– KAIST 전산학 학사,석사,박사 (2004 학위 취득)

– 2005-2006:       삼성전자 소프트웨어연구소

– 2006-2012:       주식회사 올라웍스 공동창업 (2009), 연구소장/CTO. 2012년에 인텔에 인수

– 2012-2018:       인텔 Principal Engineer, 인텔코리아 상무

– 2018: 주식회사 딥핑소스 창업대표이사

강의요약내용(abstract)

AI is everywhere. High-performance machine learning algorithms to achieve the AI commonly require a large collection of labeled data. The expensive data engineering and curation became one of the most valuable assets in industry. Consequentially demand for data sharing has been increasing a lot for the cost-effective data collection. However, privacy and ownership issues on the data are big challenges in the data sharing. For instances, faces/license plates in dash-cam-captured videos are privacy information, and mages crawled from social media are likely to cause copyright infringement. New approaches to promote sharing data for machine learning are introduced: manipulation of data so that humans are not able to recognize their detailed contents while machines can still utilize them, or vice versa. Based on the data protection techniques, a platform for secure data sharing is presented.

LIST