ANCHOR: Adaptive Network based on Cascaded Harmonic Offset Routing

TL;DR

ANCHOR is a versatile time-series modeling backbone that leverages frequency-guided modules to effectively handle non-stationary signals, outperforming existing methods in forecasting, anomaly detection, and classification.

Contribution

The paper introduces ANCHOR, a novel adaptive network utilizing cascaded harmonic offset routing and frequency priors for improved time-series analysis.

Findings

ANCHOR achieves state-of-the-art results in short-term forecasting.

It demonstrates strong performance in anomaly detection tasks.

The model is effective for time-series classification across various datasets.

Abstract

Time series analysis plays a foundational role in a wide range of real-world applications, yet accurately modeling complex non-stationary signals remains a shared challenge across downstream tasks. Existing methods attempt to extract features directly from one-dimensional sequences, making it difficult to handle the widely observed dynamic phase drift and discrete quantization error. To address this issue, we decouple temporal evolution into macroscopic physical periods and microscopic phase perturbations, and inject frequency-domain priors derived from the Real Fast Fourier Transform (RFFT) into the underlying spatial sampling process. Based on this idea, we propose a Frequency-Guided Deformable Module (FGDM) to adaptively compensate for microscopic phase deviations. Built upon FGDM, we further develop an Adaptive Network based on Cascaded Harmonic Offset Routing (ANCHOR) as a general-purpose backbone for time-series modeling. Through orthogonal channel partitioning and a progressive residual architecture, ANCHOR efficiently decouples multi-scale harmonic features while substantially suppressing the computational redundancy of multi-branch networks. Extensive experiments demonstrate that ANCHOR achieves the best performance in most short-term forecasting sub-tasks and exhibits strong competitiveness on several specific sub-tasks in anomaly detection and time-series classification, validating its effectiveness as a universal time-series foundation backbone.