arXiv Machine Learning

PAC-Bayesian Reconstruction Guarantees for Time Series Variational Autoencoders

The paper presents PAC‑Bayesian reconstruction guarantees for Variational Autoencoders applied to time‑series data. It extends existing bounds, which were limited to i.i.d. settings, to Markovian latent structures, allowing temporal dependencies to be captured without the bounds growing with trajectory length. The authors also provide an example framework showing that the required assumptions are not overly restrictive.

arXiv Machine Learning
Aug 31

Prequential posteriors

The paper introduces prequential posteriors, a Bayesian approach that uses a predictive‑sequential loss function to update deep generative forecasting models (DGFMs) when new data arrive. By adopting a consistency notion suitable for model misspecification, the authors prove that both the loss minimizer and the posterior concentrate on parameters with optimal predictive performance. Scalable inference is achieved with parallelisable waste‑free sequential Monte Carlo samplers that employ preconditioned gradient kernels, and the method is validated on synthetic and real meteorological time‑series data.

By Shreya Sinha-Roy, Richard G. Everitt, Christian P. Robert, Ritabrata Dutta
arXiv Machine Learning
Nov 6

Diffusion-Driven State Space Models

The paper introduces the Diffusion-Driven State Space Model (DDSSM), which replaces the Gaussian transition distribution in deep state space models with a diffusion model. This approach addresses the trade‑off between accurate forecasting and faithful capture of latent dynamics by enabling joint training of an autoencoder and a diffusion model on sequential data. Experiments show that DDSSM outperforms a state‑of‑the‑art deep SSM in fitting and forecasting a simulated time series with multimodal transitions.

By Jack Ruder, Michael Wojnowicz
arXiv Machine Learning
Nov 2

PCD: Predictive Coding Dynamics for Irregularly Sampled Time Series

The paper introduces Predictive Coding Dynamics (PCD), a residual feedback mechanism designed for irregularly sampled continuous-time series. PCD retains previous reconstructions, predicts observed entries, and uses masked signed residuals as an energy penalty to refine latent state recognition. Implemented with ODE-RNN and Latent ODE, it supports event-based hidden updates and improves forecasting and trajectory recovery under sparse, irregular sampling.

By Tianrui Gou, Jiayangmei Xu, Xingzhou Zhang, Lei Sha
arXiv Machine Learning
Oct 12

Scan-Parallel Variational Inference for event-driven Latent Jump Models

The paper introduces latent jump models (LJMs) for continuous-time latent state-space modeling, where the latent state evolves deterministically between Poisson-driven events and experiences learned instantaneous jumps at each event. It presents a tractable event-augmented ELBO and a novel Scan-Parallel Variational Inference (SPVI) method that samples the full latent event configuration before evaluating state recurrence, enabling logarithmic temporal depth and fast, scalable training. Experiments on synthetic and real irregular time series show that LJMs achieve competitive forecast accuracy with fewer latent updates and train one to two orders of magnitude faster than sequential baselines.

By Vansh Bansal, Min Chen, James G. Scott
arXiv Machine Learning
Jul 14

TimeSAE: Causal Sparse Decoding for Faithful Explanations of Black-Box Time Series Models

arXiv:2601. 09776v2 Announce Type: replace Abstract: As black box models and pretrained models gain traction in time series applications, understanding and explaining their predictions becomes increasingly vital, especially in high-stakes domains where interpretability and trust are essential.

By Khalid Oublal, Quentin Bouniot, Qi Gan, Stephan Cl\'emen\c{c}on, Zeynep Akata
arXiv Machine Learning
Jul 8

Temporal Variational Implicit Neural Representations

arXiv:2506. 01544v2 Announce Type: replace Abstract: We introduce Temporal Variational Implicit Neural Representations (TV-INRs), a probabilistic framework for modeling irregular multivariate time series that enables efficient and accurate individualized imputation and forecasting.

By Batuhan Koyuncu, Rachael DeVries, Ole Winther, Isabel Valera