Bharti, Ayush; Huang, Daolang; Kaski, Samuel; Briol, François-Xavier; (2025) Cost-aware simulation-based inference. In: Li, Yingzhen and Mandt, Stephan and Agrawal, Shipra and Khan, Emtiyaz, (eds.) Proceedings of the 28th International Conference on Artificial Intelligence and Statistics (AISTATS). (pp. pp. 28-36). Proceedings of Machine Learning Research (PMLR): Mai Khao, Thailand.

Preview

Text Briol_bharti25a.pdf Download (2MB) | Preview

Abstract

Simulation-based inference (SBI) is rapidly becoming the preferred framework for estimating parameters of intractable models in science and engineering. A significant challenge in this context is the large computational cost of simulating data from complex models, and the fact that this cost often depends on parameter values. We therefore propose costaware SBI methods which can significantly reduce the cost of existing sampling-based SBI methods, such as neural SBI and approximate Bayesian computation. This is achieved through a combination of rejection and selfnormalised importance sampling, which reduces the number of expensive simulations needed. Our approach is studied extensively on models from epidemiology to telecommunications engineering, where we obtain significant reductions in the overall cost of inference.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item