University of Cambridge > Talks.cam > Astro Data Science Discussion Group > Hybrid Summary Statistics: Telling Neural Networks where to look

Hybrid Summary Statistics: Telling Neural Networks where to look

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact km723.

Neural networks can capture an impressive range of patterns in supervised and unsupervised settings. This makes them useful for capturing data features from simulations or training datasets that can be used for highly informative and in some cases optimal simulation-based inference. However, asymptotic optimality often requires both enormous networks and large vats of training data.

I will present a way to capture high-information posteriors from training sets that are sparsely sampled over the parameter space with smaller networks for robust simulation-based inference. In physical inference problems, we can often apply domain knowledge to define traditional summary statistics to capture some of the information in a dataset. I will show that augmenting these statistics with neural network outputs to maximise the mutual information improves information extraction compared to neural summaries alone or their concatenation to existing summaries and makes inference robust in settings with low training data. I will also informally discuss extensions to this information-theoretic framework that go beyond cosmological parameter inference.

This talk is part of the Astro Data Science Discussion Group series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2024 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity