Latent space inference speeds up high-dimensional simulation parameter estimation
High-Dimensional Simulation-Based Inference in Latent Spaces
Machine Learning
Summary
It can be hard to figure out many unknown values in complex simulations, especially when there are lots of parameters and complicated data like images. The authors studied a new approach that first shrinks the big set of unknown parameters into a smaller, simpler form, does the inference there, and then converts results back. They found this method works just as well as traditional ways but can be much faster. This could help make simulation-based studies more efficient, especially when dealing with very complex models.
What this means in practice
- •For computational modelers: Run complex simulations with many parameters faster by doing inference in a simpler compressed space without losing accuracy.
- •For autonomous system developers: Improve real-time decision-making by quickly estimating uncertain parameters from high-dimensional sensor data using latent-space inference.
Authors
Lars Kühmichel, Stefan T. Radev, Bhanu Prasanna Koppolu, Masoumeh Davoudi, Jerry M. Huang, Paul-Christian Bürkner
Abstract
Neural simulation-based inference (SBI) has been widely successful in inferring a relatively small number of interpretable parameters from potentially high-dimensional observations, such as images or time series. Accordingly, representation learning in SBI has focused almost exclusively on compressing the observations used to condition the posterior. More recently, however, SBI has begun to target increasingly high-dimensional parameter spaces, raising the complementary question of whether the inference target itself should be compressed. Our answer is a practical merger of SBI and latent generative modeling, which learns a low-dimensional representation of the simulator parameters, performs posterior inference directly in this latent space, and maps posterior samples back to the original parameter space. We characterize the conditions under which latent-space inference recovers the desired target posterior and systematically study its empirical trade-offs. Across four case studies and three generative families, we compare latent and standard estimators while controlling for network capacity, regularization, optimization, and training compute. At matched training compute, latent-space inference achieves accuracy and marginal calibration comparable to direct target-space inference while sampling up to more than an order of magnitude faster.