TY - GEN
T1 - Multi-view Deep Gaussian Process with a Pre-training Acceleration Technique
AU - Zhu, Han
AU - Zhao, Jing
AU - Sun, Shiliang
N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2020.
PY - 2020
Y1 - 2020
N2 - Deep Gaussian process (DGP) is one of the popular probabilistic modeling methods, which is powerful and widely used for function approximation and uncertainty estimation. However, the traditional DGP lacks consideration for multi-view cases in which data may come from different sources or be constructed by different types of features. In this paper, we propose a generalized multi-view DGP (MvDGP) to capture the characteristics of different views and model data in different views discriminately. In order to make the proposed model more efficient in training, we introduce a pre-training network in MvDGP and incorporate stochastic variational inference for fine-tuning. Experimental results on real-world data sets demonstrate that pre-trained MvDGP outperforms the state-of-the-art DGP models and deep neural networks, achieving higher computational efficiency than other DGP models.
AB - Deep Gaussian process (DGP) is one of the popular probabilistic modeling methods, which is powerful and widely used for function approximation and uncertainty estimation. However, the traditional DGP lacks consideration for multi-view cases in which data may come from different sources or be constructed by different types of features. In this paper, we propose a generalized multi-view DGP (MvDGP) to capture the characteristics of different views and model data in different views discriminately. In order to make the proposed model more efficient in training, we introduce a pre-training network in MvDGP and incorporate stochastic variational inference for fine-tuning. Experimental results on real-world data sets demonstrate that pre-trained MvDGP outperforms the state-of-the-art DGP models and deep neural networks, achieving higher computational efficiency than other DGP models.
KW - Deep Gaussian process
KW - Multi-view learning
KW - Pre-training technique
KW - Stochastic optimization
KW - Variational inference
UR - https://www.scopus.com/pages/publications/85085732605
U2 - 10.1007/978-3-030-47436-2_23
DO - 10.1007/978-3-030-47436-2_23
M3 - 会议稿件
AN - SCOPUS:85085732605
SN - 9783030474355
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 299
EP - 311
BT - Advances in Knowledge Discovery and Data Mining - 24th Pacific-Asia Conference, PAKDD 2020, Proceedings
A2 - Lauw, Hady W.
A2 - Lim, Ee-Peng
A2 - Wong, Raymond Chi-Wing
A2 - Ntoulas, Alexandros
A2 - Ng, See-Kiong
A2 - Pan, Sinno Jialin
PB - Springer
T2 - 24th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2020
Y2 - 11 May 2020 through 14 May 2020
ER -