# Differentiable Likelihoods for Fast Inversion of 'Likelihood-Free' Dynamical Systems

@article{Kersting2020DifferentiableLF, title={Differentiable Likelihoods for Fast Inversion of 'Likelihood-Free' Dynamical Systems}, author={Hans Kersting and Nicholas Kr{\"a}mer and Martin Schiegg and Christian Daniel and Michael Tiemann and Philipp Hennig}, journal={ArXiv}, year={2020}, volume={abs/2002.09301} }

Likelihood-free (a.k.a. simulation-based) inference problems are inverse problems with expensive, or intractable, forward models. ODE inverse problems are commonly treated as likelihood-free, as their forward map has to be numerically approximated by an ODE solver. This, however, is not a fundamental constraint but just a lack of functionality in classic ODE solvers, which do not return a likelihood but a point estimate. To address this shortcoming, we employ Gaussian ODE filtering (a… CONTINUE READING

#### Citations

##### Publications citing this paper.

SHOWING 1-3 OF 3 CITATIONS

## A Fourier State Space Model for Bayesian ODE Filters

VIEW 4 EXCERPTS

CITES METHODS

## Bayesian ODE Solvers: The Maximum A Posteriori Estimate

VIEW 1 EXCERPT

CITES BACKGROUND

## Variational Autoencoding of PDE Inverse Problems

VIEW 1 EXCERPT

CITES METHODS

#### References

##### Publications referenced by this paper.

SHOWING 1-10 OF 54 REFERENCES

## Scalable Variational Inference for Dynamical Systems

VIEW 5 EXCERPTS

HIGHLY INFLUENTIAL

## Accelerating Bayesian Inference over Nonlinear Differential Equations with Gaussian Processes

VIEW 10 EXCERPTS

HIGHLY INFLUENTIAL

## A Conceptual Introduction to Hamiltonian Monte Carlo

VIEW 7 EXCERPTS

HIGHLY INFLUENTIAL

## Bayesian Filtering and Smoothing

VIEW 3 EXCERPTS

HIGHLY INFLUENTIAL

## Exponential convergence of Langevin distributions and their discrete approximations

VIEW 8 EXCERPTS

HIGHLY INFLUENTIAL

## Information Theory, Inference, and Learning Algorithms

VIEW 1 EXCERPT

HIGHLY INFLUENTIAL

## Optimization Methods for Large-Scale Machine Learning

VIEW 2 EXCERPTS

HIGHLY INFLUENTIAL

## The frontier of simulation-based inference

VIEW 3 EXCERPTS

HIGHLY INFLUENTIAL