[행사/세미나] 인공지능대학원 전문가 초청 세미나(AITRICS 이주호 박사,11/1 15:00~)
- 인공지능학과
- 조회수24373
- 2019-10-10
● 일시 : 11월 1일 금요일 15:00 ~ 16:30
● 장소 : 반도체관 400126호
Title: Deep amortized clustering
Speaker: Juho Lee (이주호 박사, AITRICS)
Homepage: https://juho-lee.github.io/
Biography. Dr. Juho Lee finished his Ph.D. at
POSTECH, majoring in machine learning under the supervision of professor
Seungjin Choi. After his Ph.D., he spent one year in the computational
statistics & machine learning group at the University of Oxford, working
with professor François Caron. Now he is working as a research scientist at
AITRICS. His research mainly focuses on Bayesian inference, especially Bayesian
nonparametric models and its application to statistical inference problems in
machine learning. He is also interested in various fields, including
variational inference, set-input neural networks, Bayesian deep learning,
random graph models, and meta-learning.
Abstract. We propose a deep amortized
clustering (DAC), a neural architecture that learns to cluster datasets
efficiently using a few forward passes. DAC implicitly learns what makes a
cluster, how to group data points into clusters, and how to count the number of
clusters in datasets. DAC is meta-learned using labeled datasets for training,
a process distinct from traditional clustering algorithms which usually require
hand-specified prior knowledge about cluster shapes/structures. We empirically
show, on both synthetic and image data, that DAC can efficiently and accurately
cluster new datasets coming from the same distribution used to generate
training datasets.