[Submitted on 16 May 2025 (v1), last revised 13 Nov 2026 (this version, v4)]
Abstract:Hierarchical Gaussian Process (H-GP) models divide problems into different subtasks, allowing for different components to address each part, making them well-suited for problems with inherent compositional structure. However, existing H-GP frameworks typically employ one-way information sharing – either top-down or bottom-up – which limits sample efficiency and slows convergence. We propose Bidirectional Information Flow (BIF), which establishes continuous two-way communication. BIF retains the modular structure of hierarchical models – the parent conditions its own posterior on child summaries, treating them as structured priors – while introducing top-down feedback to softly decompose environment observations from the parent into sub-responses. This mutual exchange improves sample efficiency, enables robust training, and allows modular reuse of learned subtask models. We prove analytically the regret of a GP with a learned kernel scales linearly with the mismatch to the true kernel, tightening in the hierarchical case to the sum of child-level errors. Ablation shows removing the downward pathway collapses child $R^2$ by up to 58%. Across synthetic, neurostimulation, and HPO benchmarks, BIF achieves up to $4times$ higher parent $R^2$ and roughly 100% AUC improvement over vanilla GPBO, and outperforms or is the runner-up in 38/40 tests compared to the state-of-the-art baselines while supporting modular child transfer to novel composite tasks.
Submission history
From: Juan Guerra [view email]
[v1] Fri, 16 May 2025 14:26:03 UTC (5,530 KB)
[v2] Tue, 19 Aug 2025 17:38:23 UTC (5,252 KB)
[v3] Fri, 25 Sep 2026 18:41:01 UTC (7,008 KB)
[v4] Fri, 13 Nov 2026 18:55:03 UTC (4,364 KB)
