Henrik Häggström
@henhagg.bsky.social
PhD student @deptmathgothenburg.bsky.social | https://research.chalmers.se/en/person/henhagg
💡We compare our method (SeMPLE) with pseudomarginal (exact) inference (PEPSDI in the figure). Results are very similar to exact inference but our method is way faster and requires less tuning than particle MCMC.
April 28, 2025 at 1:04 PM
💡We compare our method (SeMPLE) with pseudomarginal (exact) inference (PEPSDI in the figure). Results are very similar to exact inference but our method is way faster and requires less tuning than particle MCMC.
💡 We use easy to train but expressive Gaussian Mixture Models to approximate the likelihood and posterior.
💡Approximations are amortized and can be sequentially refined for each individual.
💡The semi-amortized approximations make our methodology scalable for an increasing number of individuals.
💡Approximations are amortized and can be sequentially refined for each individual.
💡The semi-amortized approximations make our methodology scalable for an increasing number of individuals.
April 28, 2025 at 1:04 PM
💡 We use easy to train but expressive Gaussian Mixture Models to approximate the likelihood and posterior.
💡Approximations are amortized and can be sequentially refined for each individual.
💡The semi-amortized approximations make our methodology scalable for an increasing number of individuals.
💡Approximations are amortized and can be sequentially refined for each individual.
💡The semi-amortized approximations make our methodology scalable for an increasing number of individuals.
New paper on SBI for stochastic nonlinear mixed-effects models!
We propose a scalable Bayesian framework for hierarchical mixed-effects models, using amortized likelihood and posterior approximations, obtained without neural networks.
🔗 arxiv.org/abs/2504.11279
We propose a scalable Bayesian framework for hierarchical mixed-effects models, using amortized likelihood and posterior approximations, obtained without neural networks.
🔗 arxiv.org/abs/2504.11279
April 28, 2025 at 1:04 PM
New paper on SBI for stochastic nonlinear mixed-effects models!
We propose a scalable Bayesian framework for hierarchical mixed-effects models, using amortized likelihood and posterior approximations, obtained without neural networks.
🔗 arxiv.org/abs/2504.11279
We propose a scalable Bayesian framework for hierarchical mixed-effects models, using amortized likelihood and posterior approximations, obtained without neural networks.
🔗 arxiv.org/abs/2504.11279