Linear-Nonlinear-Poisson Neuron Networks Perform Bayesian Inference On Boltzmann Machines
This provides a biologically plausible model for deep networks, potentially linking brain function to machine learning algorithms.
The paper tackles the problem of bridging neuroscience and machine learning by showing that Linear-Nonlinear-Poisson neuron networks can represent any Boltzmann machine and perform semi-stochastic Bayesian inference, lying between Gibbs sampling and variational inference.
One conjecture in both deep learning and classical connectionist viewpoint is that the biological brain implements certain kinds of deep networks as its back-end. However, to our knowledge, a detailed correspondence has not yet been set up, which is important if we want to bridge between neuroscience and machine learning. Recent researches emphasized the biological plausibility of Linear-Nonlinear-Poisson (LNP) neuron model. We show that with neurally plausible settings, the whole network is capable of representing any Boltzmann machine and performing a semi-stochastic Bayesian inference algorithm lying between Gibbs sampling and variational inference.