An Extendable Python Implementation of Robust Optimisation Monte Carlo
This work offers a practical tool for researchers in computational statistics and machine learning dealing with intractable likelihood models, though it is incremental as it focuses on implementation and extensibility rather than new algorithmic breakthroughs.
The authors implemented the Robust Optimisation Monte Carlo (ROMC) method in the Python package ELFI to address accuracy and efficiency limitations in likelihood-free inference, providing an easy-to-use, parallelizable framework that yields accurate weighted posterior samples.
Performing inference in statistical models with an intractable likelihood is challenging, therefore, most likelihood-free inference (LFI) methods encounter accuracy and efficiency limitations. In this paper, we present the implementation of the LFI method Robust Optimisation Monte Carlo (ROMC) in the Python package ELFI. ROMC is a novel and efficient (highly-parallelizable) LFI framework that provides accurate weighted samples from the posterior. Our implementation can be used in two ways. First, a scientist may use it as an out-of-the-box LFI algorithm; we provide an easy-to-use API harmonized with the principles of ELFI, enabling effortless comparisons with the rest of the methods included in the package. Additionally, we have carefully split ROMC into isolated components for supporting extensibility. A researcher may experiment with novel method(s) for solving part(s) of ROMC without reimplementing everything from scratch. In both scenarios, the ROMC parts can run in a fully-parallelized manner, exploiting all CPU cores. We also provide helpful functionalities for (i) inspecting the inference process and (ii) evaluating the obtained samples. Finally, we test the robustness of our implementation on some typical LFI examples.