Unravelling in Collaborative Learning

TL;DR

This paper investigates the challenges of collaborative learning with strategic agents having private data quality, revealing unravelling phenomena and proposing a probabilistic verification method to sustain cooperation without external transfers.

Contribution

It introduces a novel probabilistic verification approach to prevent unravelling in collaborative learning with private data quality information.

Findings

Unravelling can cause coalition collapse in strategic collaborative learning.

Probabilistic verification can ensure stable cooperation despite private data quality.

The proposed method makes the grand coalition a high-probability Nash equilibrium.

Abstract

Collaborative learning offers a promising avenue for leveraging decentralized data. However, collaboration in groups of strategic learners is not a given. In this work, we consider strategic agents who wish to train a model together but have sampling distributions of different quality. The collaboration is organized by a benevolent aggregator who gathers samples so as to maximize total welfare, but is unaware of data quality. This setting allows us to shed light on the deleterious effect of adverse selection in collaborative learning. More precisely, we demonstrate that when data quality indices are private, the coalition may undergo a phenomenon known as unravelling, wherein it shrinks up to the point that it becomes empty or solely comprised of the worst agent. We show how this issue can be addressed without making use of external transfers, by proposing a novel method inspired by probabilistic verification. This approach makes the grand coalition a Nash equilibrium with high probability despite information asymmetry, thereby breaking unravelling.