PulseAugur
EN
LIVE 06:45:43

MarathonTS benchmark compares synthetic time series generation methods

Researchers have introduced MarathonTS, the first open benchmark designed to compare methods for generating long synthetic time series. The benchmark evaluates two primary approaches: assembling series from shorter generated windows and using direct long-series generators. MarathonTS assesses fidelity to real data using 83 metrics, utility through downstream model training and testing, memorization of training data, and computational cost across 9 open hourly datasets. Findings indicate that the choice of windows significantly impacts series quality, more so than seam processing, and that no synthetic method currently matches the performance of real data for downstream tasks. AI

IMPACT Provides a standardized evaluation framework for synthetic time series generation, potentially improving model training and data augmentation techniques.

RANK_REASON The cluster contains an academic paper detailing a new benchmark for synthetic time series generation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

MarathonTS benchmark compares synthetic time series generation methods

How we ranked this

Signal score
40 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
The cluster contains an academic paper detailing a new benchmark for synthetic time series generation. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, other
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. arXiv cs.LG TIER_1 English(EN) · Ilyas Varshavskiy, Sergey Shulgin, Shuhrat Khalilbekov, Bonu Boboeva, Azizjon Azimi ·

    MarathonTS: a benchmark of long synthetic time series from window assembly and long-series generators

    arXiv:2611.18190v1 Announce Type: new Abstract: Many applications of synthetic time series require a series that is much longer than the windows on which a generative model is trained. Such a series can be assembled from a pool of short generated windows, which requires deciding …