RAID: Semantic Graph Diffusion for True Cold-Start and Cross-Lingual Forecasting

2026-06-15

AI Summary

This work addresses the challenges of genuine cold-start time series forecasting—where no historical observations are available—and cross-lingual generalization by proposing a novel approach based on semantic retrieval and graph-conditioned diffusion. The method constructs an inductive retrieval graph in a shared semantic space using frozen multilingual embeddings, first aggregating semantic neighbors to produce an initial forecast and then refining residual uncertainty through a gated diffusion module. It is the first to integrate semantic graph diffusion into cold-start prediction, enabling zero-shot cross-lingual transfer and supporting efficient non-autoregressive inference. Under strict cold-start evaluation protocols, the model significantly outperforms existing baselines in both point forecasting accuracy and prediction interval coverage, while reducing inference latency by an order of magnitude.

Abstract

Time-series foundation models show strong transfer performance when given a non-empty history window. However, true cold-start scenarios, where a new item has no prior observations, violate this assumption. We propose RAID (Retrieval-Augmented Iterative Diffusion) a framework, which replaces history-based correlation learning with metadata-driven semantic retrieval and graph-conditioned diffusion. RAID maps textual metadata into a shared semantic space using a frozen multilingual embedding model and constructs an inductive retrieval graph that extends naturally to unseen items. It first forms a base forecast by aggregating information from semantically related neighbors, then refines this forecast with a gated diffusion module to model residual uncertainty. Under a strict true cold-start protocol, RAID outperforms strong foundation models and competitive baselines on both forecasting accuracy and prediction interval coverage, while reducing inference latency by an order of magnitude through non-autoregressive decoding. The shared semantic space also enables zero-shot cross-lingual transfer, allowing a model trained on English descriptions to generalize to items described in other languages without direct supervision.

Problem

• cold-start
• time-series forecasting
• cross-lingual
• zero-shot transfer
• semantic retrieval

Innovation

• cold-start forecasting
• semantic graph
• diffusion model
• cross-lingual transfer
• retrieval-augmented generation