Pith. sign in

REVIEW 4 cited by

SigDiffusions: Score-Based Diffusion Models for Time Series via Log-Signature Embeddings

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2406.10354 v2 pith:RIBRMRJ2 submitted 2024-06-14 cs.LG

Barbora Barancikova , Zhuoyue Huang , Cristopher Salvi This is my paper · ORCID
classification cs.LG
keywords log-signaturemodelsseriestimediffusiondataembeddingsformulae
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Score-based diffusion models have recently emerged as state-of-the-art generative models for a variety of data modalities. Nonetheless, it remains unclear how to adapt these models to generate long multivariate time series. Viewing a time series as the discretisation of an underlying continuous process, we introduce SigDiffusion, a novel diffusion model operating on log-signature embeddings of the data. The forward and backward processes gradually perturb and denoise log-signatures while preserving their algebraic structure. To recover a signal from its log-signature, we provide new closed-form inversion formulae expressing the coefficients obtained by expanding the signal in a given basis (e.g. Fourier or orthogonal polynomials) as explicit polynomial functions of the log-signature. Finally, we show that combining SigDiffusions with these inversion formulae results in high-quality long time series generation, competitive with the current state-of-the-art on various datasets of synthetic and real-world examples.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Signature Reconstruction from Randomized Signatures

    math.CA 2025-02 reject novelty 8.0 of 10

    Depth-two exponential randomized signatures are claimed to reconstruct up to d^(N+1) signature features from hidden dimension N, based on new linear independence results for tree-like vector fields.

  2. Sparse Signature Coefficient Recovery via Kernels

    math.NA 2024-12 conditional novelty 7.0 of 10

    A single signature coefficient at any depth can be recovered as a linear combination of signature kernel evaluations, each a Goursat PDE solution, with parallel time O(L).

  3. Rough kernel hedging

    math.FA 2025-01 conditional novelty 6.0 of 10

    A signature-kernel and operator-valued-kernel framework for hedging is proved to have a unique global minimizer with an explicit formula, and it approximates the delta hedge on a GBM example.

  4. Population Aware Diffusion for Time Series Generation

    cs.LG 2025-01 conditional novelty 6.0 of 10

    PaD-TS trains time series diffusion models with a maximum mean discrepancy penalty on cross-correlation distributions, improving population-level property preservation.

Pith tools