RDDMPI introduces a residual denoising diffusion model for multivariate time series imputation. By decomposing the missing signal into a baseline reconstruction and a residual uncertainty component, the method conditions the diffusion process on both the completed signal and its latent representation, using a reliability-aware mechanism to balance baseline influence. Experiments on benchmark datasets show that this approach improves reconstruction accuracy and uncertainty quantification compared to prior diffusion-based methods.
By Ramiro Valdes Jara, David Chapman, Adam Meyers
The paper introduces the Masked Diffusion Time-series Imputation Model (MDTIM), which uses a masked diffusion training paradigm to directly predict original values for time series imputation. It separates missing and observed data via a MASK token and employs Stochastic Discretization to convert continuous values into ordinal-aware tokens, preserving temporal dynamics. Experiments on multiple benchmarks show that MDTIM outperforms existing deterministic and generative baselines in robustness and scalability across various missing data scenarios.
By Dongbin Kim, Seungyun Lee, Geonwoo Shin, Jaewook Lee
arXiv:2605.00126v2 Announce Type: replace
Abstract: Generative models for time-series imputation achieve strong reconstruction accuracy, yet provide no finite-sample reliability guarantees, a critica...
By Arnaud Zinflou
arXiv:2601.01480v3 Announce Type: replace-cross
Abstract: Traffic forecasting systems rely on fixed sensor networks that frequently exhibit contiguous blackouts. Such outages are usually treated as i...
By Aman Sunesh (New York University), Allan Ma (New York University), Siddarth Nilol (New York University)
The paper introduces a new evaluation protocol for deep multivariate imputation models on wearable device data, addressing the issue of structured missingness where sensor features drop out together. Using a Garmin smartwatch dataset from an epilepsy patient, the authors generate realistic block-missing patterns from training data and show that matching the training protocol to this distribution reduces BRITS’ mean absolute error by 43%. They also extend BRITS with time‑of‑day encoding and compare it to linear interpolation and SAITS, finding that no single model dominates and that model rankings vary with evaluation design.
By Skye Goodman, Roussel Desmond Nzoyem, Leandro Junges, Peter Kissack, Yasser Qureshi, Amberly Brigden, Jeff Clark, Nawid Keshtmand
arXiv:2609.37632v1 Announce Type: cross
Abstract: Time series imputation has progressed from statistical and deep learning approaches to diffusion-based models, which have shown strong recent perform...
By Fariza Rashid, Duc Van Le, Rahat Masood, Gustavo Batista, Aruna Seneviratne, Suranga Seneviratne
arXiv:2606. 06328v1 Announce Type: new Abstract: In healthcare, multimodal time series tasks often operate on incomplete observations in practice, for example when ECG segments are lost because electrodes detach or an entire respiratory channel is unavailable during overnight monitoring.
By Ziwen Kan, Wugeng Zheng, Tianlong Chen, Song Wang
The paper introduces Internal Dual-Wiener routing (Internal‑DW), a backward‑only method that weight‑balances internal gradient routes in autoregressive forecasting. By estimating bounded Wiener gains for identity and nonlinear paths, it suppresses unpredictable noise while preserving predictable learning signals, reducing forecast error by 5.2%–13.8% on four weak‑drive testbeds compared to full BPTT and outperforming gradient clipping, Jacobian regularization, and truncated BPTT in most cases. The approach shows that long‑horizon supervision can be effective without trusting every backward gradient equally.
By Junhao Zhao, David Michael Simberg, Jacob Kang, Colin Connor Kurniawan, Nan Xu
arXiv:2606. 27908v1 Announce Type: new Abstract: Long-term time series forecasting finds extensive applications in domains such as power demand, traffic flow, meteorological observation, and renewable energy dispatch.
By Wenchao Liu, Hongbing Wang, Youji Zhu, Xiaodong Liu, Xiangguang Xiong
Latent diffusion models achieve strong generative performance by operating in a compressed latent space produced by a variational autoencoder (VAE). However, it remains unclear whether all latent channels contribute equally to the diffusion process, or whether significant redundancy exists.
DynG-Diff is a new diffusion-based framework for probabilistic multivariate time‑series forecasting that addresses the challenge of information heterogeneity across variables. It uses a two‑stage training strategy with an unconditional diffusion backbone and introduces a lightweight state‑aware policy network that dynamically adjusts guidance strength based on real‑time variable reliability. The dynamic guidance is mathematically framed as local precision, allowing the model to focus on high‑confidence variables and suppress anomalous noise, leading to competitive performance and robustness on real‑world benchmarks.
By Zhente Zhang, Zhengwei Ni, Wei Fan
Probabilistic multivariate time series (MTS) forecasting is crucial for modeling complex dynamical systems. However, existing diffusion-based methods rely on task-specific conditional paradigms that l...