Authors:Xiaorui Wang, Fanda Fan, Chenxi Wang, Yuxuan Yang, Rui Tang, Kuoyu Gao, Simiao Pang, Yuanfeng Shang, Zhipeng Liu, Wanling Gao, Lei Wang, Jianfeng Zhan
Abstract:Recent progress in time-series forecasting has led to rapidly increasing architectural complexity, yet many reported State-of-the-Art gains are statistically fragile or misattributed. We argue that progress requires a shift from model selection to modular attribution, identifying which components truly drive performance. We propose CombinationTS, a self-contained probabilistic evaluation framework that decomposes forecasting models into orthogonal modules--Input Transformation, Embedding, Encoder, Decoder, and Output Transformation--and evaluates them under a shared evaluation condition space. By quantifying each component via marginalized performance ($\mu$) and stability ($\sigma$), CombinationTS enables robust attribution beyond fragile point estimates. Through large-scale paired evaluation, we uncover the Identity Paradox: once the data view (Embedding) is well-designed, a parameter-free Identity Encoder often matches or outperforms complex backbones. We further show that explicit structural priors introduced via Input Transformations yield a more favorable performance-stability trade-off than increasing Encoder complexity, establishing a principled baseline for architectural necessity.
| Comments: | Accepted by ICML 2026 main track. Code available at this https URL |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2605.01231 [cs.LG] |
| (or arXiv:2605.01231v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2605.01231 arXiv-issued DOI via DataCite (pending registration) |
From: Fanda Fan [view email]
[v1]
Sat, 2 May 2026 04:08:51 UTC (879 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。