SPLICE: Latent Diffusion over JEPA Embeddings for Conformal Time-Series Inpainting

TL;DR

SPLICE introduces a novel framework combining latent diffusion models with conformal inference for reliable time-series imputation, achieving high accuracy and coverage guarantees across diverse datasets.

Contribution

The paper presents SPLICE, a modular latent diffusion-based method with adaptive conformal inference, providing both accurate imputation and finite-sample reliability guarantees in power system time-series data.

Findings

SPLICE achieves the lowest mean load-only MSE on thirteen datasets.

It produces coverage-guaranteed prediction bands with 93-95% empirical coverage.

The flow-matching variant speeds up inference by 5-10x while maintaining quality.

Abstract

Generative models for time-series imputation achieve strong reconstruction accuracy, yet provide no finite-sample reliability guarantees, a critical limitation in power systems where imputed values inform dispatch and planning. We introduce SPLICE (Self-supervised Predictive Latent Inpainting with Conformal Envelopes), a modular framework coupling latent generative imputation with distribution-free, online-adaptive prediction intervals. A JEPA encoder maps daily load segments into a 64-dimensional latent space; a conditional latent bridge with four sampling modes generates candidate gap trajectories; an hourly-conditioned decoder maps back to signal space; and Adaptive Conformal Inference (ACI) wraps the output with coverage-guaranteed prediction bands. The flow-matching variant achieves comparable quality to DDIM in 5--10 ODE steps (5-10x speedup). On thirteen load datasets (nine proprietary, three UCI Electricity, ETTh1), SPLICE achieves the lowest mean Load-only MSE (0.056), winning 9/12 non-degenerate datasets at 91-day gaps and 18/32 across all gap lengths vs. five established baselines, and produces the best CRPS (0.161, -18.3% vs. the strongest competitor). ACI delivers 93--95% empirical coverage, correcting under-coverage failures of up to 7.5 pp observed with static conformal prediction. A pooled JEPA encoder trained on nine feeds transfers to four unseen domains, matching or exceeding per-dataset oracles with only a quick bridge fine-tuning.