Below are some references on EBM's that Will Grathwohl provides for his talk:
EBM history:
OG POE w/ CD from da boss Geoff: https://www.cs.toronto.edu/~hinton/absps/tr00-004.pdf
PCD for RBMs: https://www.cs.toronto.edu/~tijmen/pcd/pcd.pdf
Classic paper on Deep belief nets: https://www.cs.toronto.edu/~hinton/absps/fastnc.pdf
Evaluation of RBMs with AIS/RAISE: https://arxiv.org/abs/1412.8566
Modern EBMs:
SGLD: https://www.stats.ox.ac.uk/~teh/research/compstats/WelTeh2011a.pdf
Seminal work on langevin trained deep EBMs: https://arxiv.org/abs/1903.12370
Scaling up to larger images: https://arxiv.org/abs/1903.08689
JEM: https://arxiv.org/abs/1912.03263
CoopNets: http://www.stat.ucla.edu/~ywu/CoopNets/main.html
Short run MCMC: https://arxiv.org/abs/1904.09770
Alternative training procedure:
Noise Contrastive Estimation: https://proceedings.mlr.press/v9/gutmann10a/gutmann10a.pdf
Score matching: https://jmlr.org/papers/volume6/hyvarinen05a/old.pdf
Kernelized Stein Discrepancy: https://arxiv.org/abs/1602.03253
Learning the Stein Discrepancy: https://arxiv.org/abs/2002.05616
Variational EBM Learning:
VERA: https://arxiv.org/abs/2010.04230
Flow contrastive Estimation: https://arxiv.org/abs/1912.00589
Flow-based sampling: https://arxiv.org/pdf/2006.06897.pdf
Discrete Sampling:
Locally-Balanced Proposals: https://arxiv.org/abs/1711.07424
Gibbs-With-Gradients: https://arxiv.org/pdf/2102.04509.pdf
Path-auxillury proposals (GWG extension): https://openreview.net/forum?id=JSR-YDImK95