Partition Function Tracker

An implementation of the paper[1] in which the authors proposed a method to track the partition function of an RBM during learning.

To validate the implementation, this library also implements AIS[2] applied to RBMs.

Installation

pip install git+https://github.com/Ktakuya332C/tracker

Usage

See examples directory.

References

1. G. Desjardins et al. (2011), On tracking the partition function, NIPS
2. R.M. Neal (2001), Annealed importance sampling, Stat. Comput.