Censoring is the central problem in survival analysis where either the time-to-event (for instance, death) or the time-to-censoring (such as loss of follow-up) is observed for each sample. The majority of existing survival analysis methods assume that survival is conditionally independent of censoring given a set of covariates, an assumption that cannot be verified since only marginal distributions are available from the data. The existence of dependent censoring and the inherent bias in current estimators has been demonstrated in various applications, accentuating the need for a more nuanced approach. However, existing methods that adjust for dependent censoring require practitioners to specify the ground truth copula. This requirement poses a significant challenge for practical applications, as model misspecification can lead to substantial bias. This talk will discuss a flexible survival analysis method that simultaneously accommodates dependent censoring and eliminates the requirement for specifying the copula. We theoretically prove the identifiability of our model under a broad family of copulas and survival distributions and empirically demonstrate that our method achieves significantly lower estimation bias when compared to existing approaches.
Speaker Bio
Dr. Weijia Zhang holds the position of lecturer in the Data Science and Statistics discipline at the University of Newcastle. Prior to this role, he worked in various academic positions at the University of South Australia and Southeast University, China. His research primarily focuses on causal inference, weakly-supervised machine learning, and survival analysis. Dr. Zhang contributes to the program committee of many international conferences, including AAAI, AISTATS, ICML, IJCAI, NeurIPS, UAI, and ICLR. Additionally, he holds editorial positions with the journals Computers in Industry and Data Science & Engineering.
More Details
- When: Wed 24 April 2024, at 1:00 pm (GMT+10)
- Speaker: Dr Weijia Zhang (University of Newcastle)
- Host: Dr Miao Xu
- Venue: 49-313A
- Zoom: https://uqz.zoom.us/j/82441902377
No.24-03 Towards Open-World Object Segmentation and Detection