Variational Inference for Gaussian Process Models for Survival Analysis

Minyoung Kim, Vladimir Pavlovic
Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence, PMLR R16:434-444, 2018.

Abstract

Gaussian process survival analysis model (GP- SAM) was recently proposed to address key deficiencies of the Cox proportional hazard model, namely the need to account for uncertainty in the hazard function modeling while, at the same time, relaxing the time-covariates factorized assumption of the Cox model. However, the existing MCMC inference algorithms for GPSAM have proven to be slow in practice. In this paper we propose novel and scalable variational inference algorithms for GPSAM that reduce the time complexity of the sampling approaches and improve scalability to large datasets. We accomplish this by employing two effective strategies in scalable GP: i) using pseudo inputs and ii) approximation via random feature expansions. In both setups, we derive the full and partial likelihood formulations, typically considered in survival analysis settings. The proposed approaches are evaluated on two clinical and a divorce-marriage benchmark datasets, where we demonstrate improvements in prediction accuracy over the existing survival analysis methods, while reducing the complexity of inference compared to the recent state-of-the-art MCMC-based algorithms.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR16-kim18a, title = {Variational Inference for {G}aussian Process Models for Survival Analysis}, author = {Kim, Minyoung and Pavlovic, Vladimir}, booktitle = {Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence}, pages = {434--444}, year = {2018}, editor = {Globerson, Amir and Silva, Ricardo}, volume = {R16}, series = {Proceedings of Machine Learning Research}, month = {06--10 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r16/main/assets/kim18a/kim18a.pdf}, url = {https://proceedings.mlr.press/r16/kim18a.html}, abstract = {Gaussian process survival analysis model (GP- SAM) was recently proposed to address key deficiencies of the Cox proportional hazard model, namely the need to account for uncertainty in the hazard function modeling while, at the same time, relaxing the time-covariates factorized assumption of the Cox model. However, the existing MCMC inference algorithms for GPSAM have proven to be slow in practice. In this paper we propose novel and scalable variational inference algorithms for GPSAM that reduce the time complexity of the sampling approaches and improve scalability to large datasets. We accomplish this by employing two effective strategies in scalable GP: i) using pseudo inputs and ii) approximation via random feature expansions. In both setups, we derive the full and partial likelihood formulations, typically considered in survival analysis settings. The proposed approaches are evaluated on two clinical and a divorce-marriage benchmark datasets, where we demonstrate improvements in prediction accuracy over the existing survival analysis methods, while reducing the complexity of inference compared to the recent state-of-the-art MCMC-based algorithms.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Variational Inference for Gaussian Process Models for Survival Analysis %A Minyoung Kim %A Vladimir Pavlovic %B Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2018 %E Amir Globerson %E Ricardo Silva %F pmlr-vR16-kim18a %I PMLR %P 434--444 %U https://proceedings.mlr.press/r16/kim18a.html %V R16 %X Gaussian process survival analysis model (GP- SAM) was recently proposed to address key deficiencies of the Cox proportional hazard model, namely the need to account for uncertainty in the hazard function modeling while, at the same time, relaxing the time-covariates factorized assumption of the Cox model. However, the existing MCMC inference algorithms for GPSAM have proven to be slow in practice. In this paper we propose novel and scalable variational inference algorithms for GPSAM that reduce the time complexity of the sampling approaches and improve scalability to large datasets. We accomplish this by employing two effective strategies in scalable GP: i) using pseudo inputs and ii) approximation via random feature expansions. In both setups, we derive the full and partial likelihood formulations, typically considered in survival analysis settings. The proposed approaches are evaluated on two clinical and a divorce-marriage benchmark datasets, where we demonstrate improvements in prediction accuracy over the existing survival analysis methods, while reducing the complexity of inference compared to the recent state-of-the-art MCMC-based algorithms. %Z Reissued by PMLR on 04 October 2026.
APA
Kim, M. & Pavlovic, V.. (2018). Variational Inference for Gaussian Process Models for Survival Analysis. Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R16:434-444 Available from https://proceedings.mlr.press/r16/kim18a.html. Reissued by PMLR on 04 October 2026.

Related Material