Unified field theoretical approach to deep and recurrent neuronal networks

TL;DR

This paper develops a unified mean-field theory for deep and recurrent neural networks using statistical physics, revealing their similar Gaussian process limits and providing insights into finite-width differences.

Contribution

It introduces a systematic derivation of the mean-field theory for both architectures from first principles, including next-to-leading-order corrections.

Findings

Gaussian kernels are identical at a single readout time or layer

Recurrent networks converge slower to mean-field theory than deep networks

Finite-width corrections are architecture-specific

Abstract

Understanding capabilities and limitations of different network architectures is of fundamental importance to machine learning. Bayesian inference on Gaussian processes has proven to be a viable approach for studying recurrent and deep networks in the limit of infinite layer width, $n\to\infty$. Here we present a unified and systematic derivation of the mean-field theory for both architectures that starts from first principles by employing established methods from statistical physics of disordered systems. The theory elucidates that while the mean-field equations are different with regard to their temporal structure, they yet yield identical Gaussian kernels when readouts are taken at a single time point or layer, respectively. Bayesian inference applied to classification then predicts identical performance and capabilities for the two architectures. Numerically, we find that convergence towards the mean-field theory is typically slower for recurrent networks than for deep networks and the convergence speed depends non-trivially on the parameters of the weight prior as well as the depth or number of time steps, respectively. Our method exposes that Gaussian processes are but the lowest order of a systematic expansion in $1/n$ and we compute next-to-leading-order corrections which turn out to be architecture-specific. The formalism thus paves the way to investigate the fundamental differences between recurrent and deep architectures at finite widths $n$.