학술행사
- 발표자 신원용 교수,연세대학교
- 조직위원 산업수학혁신센터
- 기간 2024-03-28 ~ 2024-03-28
- 장소 판교 테크노밸리 산업수학혁신센터 세미나실
- 주최 산업수학혁신센터
-
일시: 2024.3.28.(목), 14:00~16:00
-
장소: 판교 테크노밸리 산업수학혁신센터 세미나실
- 경기 성남시 수정구 대왕판교로 815, 기업지원허브 231호 국가수리과학연구소
- 무료주차는 2시간 지원됩니다.
-
발표자: 신원용 교수(연세대학교)
-
주요내용: Graph Learning vs. Graph Filtering
In the graph signal processing perspective, a series of graph filtering is shown to exhibit state-of-the-art performance with a substantially low computational complexity. This talk aims to bridge between graph filtering and graph learning. In the first part of this talk, I explain how the basic mechanism of the well-known graph convolutional network (GCN) is interpreted as graph filters. In the second part of this talk, I introduce graph filtering methods using a low-pass filter without a costly model training process. More specifically, I present graph filtering-based collaborative filtering approaches that do not require training for recommender systems. Finally, I discuss how such methodology is applicable to a broad spectrum of real-world recommendation domains.
*유튜브 스트리밍 예정입니다.
-
일시: 2024.3.28.(목), 14:00~16:00
-
장소: 판교 테크노밸리 산업수학혁신센터 세미나실
- 경기 성남시 수정구 대왕판교로 815, 기업지원허브 231호 국가수리과학연구소
- 무료주차는 2시간 지원됩니다.
-
발표자: 신원용 교수(연세대학교)
-
주요내용: Graph Learning vs. Graph Filtering
In the graph signal processing perspective, a series of graph filtering is shown to exhibit state-of-the-art performance with a substantially low computational complexity. This talk aims to bridge between graph filtering and graph learning. In the first part of this talk, I explain how the basic mechanism of the well-known graph convolutional network (GCN) is interpreted as graph filters. In the second part of this talk, I introduce graph filtering methods using a low-pass filter without a costly model training process. More specifically, I present graph filtering-based collaborative filtering approaches that do not require training for recommender systems. Finally, I discuss how such methodology is applicable to a broad spectrum of real-world recommendation domains.
