DeepBooTS: Dual-Stream Residual Boosting for Drift-Resilient Time-Series Forecasting

2025-11-10

AI Summary

In time-series forecasting, concept drift-induced non-stationarity severely degrades model robustness. To address this, we propose DRD—an end-to-end dual-stream residual-decay enhancement framework—that pioneers a bias-variance trade-off perspective for modeling concept drift. DRD introduces a block-level residual correction mechanism and a multi-learner ensemble architecture to achieve input-target decoupling and progressive signal reconstruction. Technically, it integrates instance normalization, deep residual networks, and a dual-stream structure, augmented with block-level auxiliary output branches forming a high-speed pathway for layer-wise residual refinement and joint time-frequency domain signal separation. Evaluated on large-scale benchmarks, DRD achieves an average 15.8% improvement in forecasting accuracy, markedly enhancing adaptability to concept drift and model stability. This work establishes a novel paradigm for non-stationary time-series forecasting.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Data Mining & Knowledge Management: Data Stream Mining
• Computer Vision: Diffusion Models for Vision

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Time-Series (TS) exhibits pronounced non-stationarity. Consequently, most forecasting methods display compromised robustness to concept drift, despite the prevalent application of instance normalization. We tackle this challenge by first analysing concept drift through a bias-variance lens and proving that weighted ensemble reduces variance without increasing bias. These insights motivate DeepBooTS, a novel end-to-end dual-stream residual-decreasing boosting method that progressively reconstructs the intrinsic signal. In our design, each block of a deep model becomes an ensemble of learners with an auxiliary output branch forming a highway to the final prediction. The block-wise outputs correct the residuals of previous blocks, leading to a learning-driven decomposition of both inputs and targets. This method enhances versatility and interpretability while substantially improving robustness to concept drift. Extensive experiments, including those on large-scale datasets, show that the proposed method outperforms existing methods by a large margin, yielding an average performance improvement of 15.8% across various datasets, establishing a new benchmark for TS forecasting.

Problem

• Addressing concept drift in non-stationary time-series forecasting
• Reducing prediction residuals through dual-stream boosting architecture
• Enhancing robustness and interpretability of time-series models

Innovation

• Dual-stream residual boosting for time-series
• Ensemble learners with auxiliary output branches
• Progressive intrinsic signal reconstruction via residuals