This paper provides a comprehensive analysis of variational inference in latent variable models for survival analysis, emphasizing the distinctive challenges associated with applying variational methods to survival data. We identify a critical weakness in the existing methodology, demonstrating how a poorly designed variational distribution may hinder the objective of survival analysis tasks--modeling time-to-event distributions. We prove that the optimal variational distribution, which perfectly bounds the log-likelihood, may depend on the censoring mechanism. To address this issue, we propose censor-dependent variational inference (CDVI), tailored for latent variable models in survival analysis. More practically, we introduce CD-CVAE, a V-structure Variational Autoencoder (VAE) designed for the scalable implementation of CDVI. Further discussion extends some existing theories and training techniques to survival analysis. Extensive experiments validate our analysis and demonstrate significant improvements in the estimation of individual survival distributions.

The market for face biometric or multi-modal physical access control is a little more crowded, with new solutions launched this week by SAFR, new ievo parent company CDVI, Telaeris and Invixium. SAFR from RealNetworks is introducing the new SAFR SCAN touchless biometric access control product for commercial and office settings at ISC West 2022 in Las Vegas this week, marking a new direction for the company's facial recognition software. RealNetworks CEO Rob Glaser notes that the access control solution is the first integrated hardware product the company has ever made, and claims it is the most secure way to control access to a building or office ever made. The new SAFR SCAN is intended to run either as a standalone or networked biometric solutions, and can authenticate up to 30 people per minute, according to the announcement. It also utilizes 3D structured light and RGB for anti-spoofing liveness detection.