The Copycat Perceptron: Smashing Barriers Through Collective Learning

TL;DR

This paper analyzes a model of coupled binary perceptrons with thermal noise, showing that collective learning via replicas smooths the landscape, enabling efficient learning and supporting the Bayes-optimality of replicated algorithms.

Contribution

It provides a detailed characterization of equilibrium properties of coupled perceptrons with thermal noise, demonstrating how replica coupling improves learning efficiency and landscape smoothness.

Findings

Replica coupling leads to a smoother free entropy landscape.

Replicated Simulated Annealing can efficiently reach the teacher solution.

Multiple students can learn faster and with fewer examples.

Abstract

We characterize the equilibrium properties of a model of $y$ coupled binary perceptrons in the teacher-student scenario, subject to a suitable cost function, with an explicit ferromagnetic coupling proportional to the Hamming distance between the students' weights. In contrast to recent works, we analyze a more general setting in which thermal noise is present that affects each student's generalization performance. In the nonzero temperature regime, we find that the coupling of replicas leads to a bend of the phase diagram towards smaller values of $\alpha$: This suggests that the free entropy landscape gets smoother around the solution with perfect generalization (i.e., the teacher) at a fixed fraction of examples, allowing standard thermal updating algorithms such as Simulated Annealing to easily reach the teacher solution and avoid getting trapped in metastable states as it happens in the unreplicated case, even in the computationally \textit{easy} regime of the inference phase diagram. These results provide additional analytic and numerical evidence for the recently conjectured Bayes-optimal property of Replicated Simulated Annealing (RSA) for a sufficient number of replicas. From a learning perspective, these results also suggest that multiple students working together (in this case reviewing the same data) are able to learn the same rule both significantly faster and with fewer examples, a property that could be exploited in the context of cooperative and federated learning.