Riemann-Theta Boltzmann Machine
arXiv:1712.07581 · doi:10.1016/j.neucom.2020.01.011
Abstract
A general Boltzmann machine with continuous visible and discrete integer valued hidden states is introduced. Under mild assumptions about the connection matrices, the probability density function of the visible units can be solved for analytically, yielding a novel parametric density function involving a ratio of Riemann-Theta functions. The conditional expectation of a hidden state for given visible states can also be calculated analytically, yielding a derivative of the logarithmic Riemann-Theta function. The conditional expectation can be used as activation function in a feedforward neural network, thereby increasing the modelling capacity of the network. Both the Boltzmann machine and the derived feedforward neural network can be successfully trained via standard gradient- and non-gradient-based optimization techniques.
arXiv comment29 pages, 11 figures, final version published in Neurocomputing
References in corpus (7)
- Jet Substructure Classification in High-Energy Physics with Deep Neural Networks
- Phase Diagram of Restricted Boltzmann Machines and Generalised Hopfield Networks with Arbitrary Priors
- Phase transitions in Restricted Boltzmann Machines with generic priors
- Gaussian-binary Restricted Boltzmann Machines on Modeling Natural Image Statistics
- On the determination of probability density functions by using Neural Networks
- Riemann-Theta Boltzmann Machine
- Efficient computation of multidimensional theta functions