Bayesian updating with structural reliability methods (BUS) [1] is a novel approach for calibrating models within uncertainty analysis, transforming Bayesian updating into a structural reliability problem. However, reliability analysis in the case of high-dimensional problems remains a significant challenge, especially due to the significant computational cost, which depends on the required level of accuracy. To overcome this issue, the paper introduces an innovative Bayesian updating framework based on convolutional autoencoders [2]. This approach successfully overcomes the challenges associated with high dimensions. It first involves the training of a failure-informed convolutional autoencoder aiming to create a failure surface in a low-dimensional latent space. The dimensionality of the input space is reduced by the encoder, while the reconstruction of the output is obtained by the decoder. The highdimensional reliability problem can be handled by replacing the limit state function in the latent space using a novel, highly efficient active learning Kriging model, known as qAK and previously proposed by the authors [3]. Therefore, an active learning technique is adopted for the training of the model in order to produce points in the vicinity of the limit state surface. This approach improves the accuracy it also improves the model updating procedure. A highdimensional structural dynamics example is used to demonstrate the effectiveness of the proposed method. REFERENCES [1] D. Straub, I. Papaioannou, Bayesian Updating with Structural Reliability Methods, Journal of Engineering Mechanics, 141, 2015, https://doi.org/10.1061/(ASCE)EM.1943-7889.0000839. [2] S. Nikolopoulos, I. Kalogeris, V. Papadopoulos, Non-intrusive surrogate modeling for parametrized time-dependent partial differential equations using convolutional autoencoders, Engineering Applications of Artificial Intelligence, 109: 104652, 2022, https://doi.org/10.1016/j.engappai.2021.104652. [3] I. Prentzas, M. Fragiadakis, Quantified single and multiple active learning Kriging models for structural reliability analysis, Reliability Engineering and System Safety (2023) (under review)