ML Reading Group

This is an informal reading group open to all UCI students and faculty. Each week we will read and discuss an interesting paper selected by the group. Please add to this list papers you’d like to read and discuss with the group.

Everyone who attends is expected to have read the paper and contribute to the discussion. Please note at least one point about the paper that you would like to discuss. This may be a question you have or an interesting idea that you would like to highlight.

Mailing list: mlrg

Reading list: doc

Next meeting: Friday 4–5pm, December 6, 2019
Location: DBH 4011
Facilitator: Yadong Lu
Reading: Energy-Inspired Models: Learning with Sampler-Induced Distributions
Outline: Energy-based models often suffer from intractable sampling and density evaluation due to the partition function. The interesting idea from this paper is directly treating the sampling procedure as the model of interest and optimizing the log-likelihood of the sampling procedure. The paper evaluates three instantiations of such models based on truncated rejection sampling, self-normalized importance sampling, and Hamiltonian importance sampling.