AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Amortized Bethe Free Energy Minimization for Learning MRFs

Sam Wiseman, Yoon Kim

NeurIPS 2019 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Annealing and divergence-based variational objectives · 6 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use amortized BFE minimization instead of Loopy BPIsing model learning · Ising models, n=15held out NLL: 59.79 -> 54.91Improved
Use amortized BFE minimization instead of exact inferenceIsing model learning · Ising models, n=15held out NLL: 52.24 -> 54.91Hurt
Use the inference network for marginal estimation instead of Loopy BPIsing model inference · Ising models, n=15marginal correlation: 0.942 -> 0.981Improved
Use the inference network for marginal estimation instead of Loopy BPIsing model inference · Ising models, n=15Mean L1 distance: 0.065 -> 0.040Improved
Use amortized BFE minimization instead of Loopy BPRBM learning · UCI digitsheld out average NLL: 25.47 -> 23.43Improved
Use amortized BFE minimization instead of PCDRBM learning · UCI digitsheld out average NLL: 21.24 -> 23.43Hurt
Use amortized BFE minimization instead of Loopy BPUndirected 3rd Order HMM learning · Penn Treebankaverage NLL: 108.74 -> 115.86Hurt
Use an undirected pairwise-factor HMM and amortized BFE rather than the directed model with a mean-field VAE3rd Order HMM learning · Penn Treebankaverage NLL: 119.27 -> 115.86Improved
Use an undirected pairwise-factor HMM and amortized BFE rather than the directed model with a 1st Order HMM VAE3rd Order HMM learning · Penn Treebankaverage NLL: 118.35 -> 115.86Improved
Use amortized BFE minimization rather than Loopy BPUndirected 3rd Order HMM learning · Penn Treebankseconds/epoch: 247 -> 70Improved
Replace a Transformer inference network with a BLSTM (ablation)RBM and HMM inference networks · RBM and Penn Treebank HMM experimentsheld out performanceImproved
Use KL divergence rather than L2 distance for the consistency penalty (ablation)RBM and HMM inference networks · RBM and Penn Treebank HMM experimentsheld out performanceImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy