Feedback Control for Multi-Objective Graph Self-Supervision

TL;DR

This paper introduces ControlG, a control-theoretic framework that optimizes multi-objective graph self-supervised learning by dynamically allocating training resources, leading to improved performance and interpretability.

Contribution

It proposes a novel feedback control approach for temporal allocation of objectives in multi-task graph SSL, addressing interference and instability issues.

Findings

ControlG outperforms baselines on 9 datasets.

It provides an interpretable schedule of objective influence.

The method reduces training conflicts and improves stability.

Abstract

Can multi-task self-supervised learning on graphs be coordinated without the usual tug-of-war between objectives? Graph self-supervised learning (SSL) offers a growing toolbox of pretext objectives: mutual information, reconstruction, contrastive learning; yet combining them reliably remains a challenge due to objective interference and training instability. Most multi-pretext pipelines use per-update mixing, forcing every parameter update to be a compromise, leading to three failure modes: Disagreement (conflict-induced negative transfer), Drift (nonstationary objective utility), and Drought (hidden starvation of underserved objectives). We argue that coordination is fundamentally a temporal allocation problem: deciding when each objective receives optimization budget, not merely how to weigh them. We introduce ControlG, a control-theoretic framework that recasts multi-objective graph SSL as feedback-controlled temporal allocation by estimating per-objective difficulty and pairwise antagonism, planning target budgets via a Pareto-aware log-hypervolume planner, and scheduling with a Proportional-Integral-Derivative (PID) controller. Across 9 datasets, ControlG consistently outperforms state-of-the-art baselines, while producing an auditable schedule that reveals which objectives drove learning.