SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2408.17253 · ICML · 2025

VISIONTS: Visual Masked Autoencoders Are Free-Lunch Zero-Shot Time Series Forecasters

Mouxiang Chen, Chenghao Liu, Zhuo Li, Jianling Sun, Lefei Shen, Xiaoyun Wang

arXiv · PDF · Open in the Atlas

Code that ran

We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. The repositories linked to it are listed below.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The game began development in 2010, carrying over a large portion of the work... Time-Series Natural Images (no time-series training) No Pre-training Pre-training Dataset few-shot zero-shot 10% downstream training dataset no downstream training dataset Mean Square Error (MSE) Averaged on six long-term time series forecasting (LTSF) datasets and four prediction lengths GIFT-Eval Leaderboard Cut-off date: Nov 2024 Figure 1. Long-term forecasting (left) and GIFT-Eval (right) performance comparison. Our VISIONTS, without any training on time series data, outperforms the pure time series foundation models in the zero-shot setting.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2408.17253")
get_code_for_paper("2408.17253")
have("2408.17253")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2408.17253.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2408.17253.svg)](https://syntology.ai/paper/2408.17253)

Connect an agent — have() is free.