AI Paper Atlas

Research map / Time series / Foundation models for time series

TS-ICL: A Flexible Time-Indexed Foundation Model for Time Series via In-Context Learning

Etienne Le Naour, Tahar Nabil, Adrien Petralia

NeurIPS 2026 · Time series › Foundation models for time series · approach: Time series foundation models · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Introduces TS-ICL, a 27M-parameter probabilistic in-context-learning encoder-regressor Transformer that unifies zero-shot forecasting and imputation, with covariates, trained on real and synthetic time series plus a DAG-based causal covariate prior.

Problem

placeholder

Approach

placeholder

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use TS-ICL time-indexed in-context regression instead of TabICLv2-TSimputation · fm-impute-bench univariateNMAE and CRPS: 17% and 15%Improved
use TS-ICL instead of TabPFNv2.5-TS with covariatesimputation · fm-impute-bench known covariatesNMAE and CRPS: 36% and 35%Improved
add covariates as inputsTS-ICL imputation · fm-impute-bench known covariatesNMAE and CRPS: 39% and 38%Improved
add covariates as inputsTS-ICL forecasting · fev-bench 30 covariate datasetsCRPSImproved
use TS-ICL instead of Chronos-2 with 30% look-back missingforecasting · fev-bench univariateMASE: 2.16 -> 1.77Improved
use TS-ICL instead of Chronos-2 with 50% look-back missingforecasting · fev-bench univariateMASE: 2.44 -> 1.89Improved
use TS-ICL instead of Chronos-2 with 70% look-back missingforecasting · fev-bench univariateMASE: 2.54 -> 2.16Improved
use TS-ICL instead of Chronos-2 with 90% look-back missingforecasting · fev-bench univariateMASE: 3.97 -> 3.63Improved
use TS-ICL instead of Chronos-2 with complete look-backforecasting · fev-bench univariateMASE: 1.62 -> 1.70Hurt
increase look-back missingness from 0% to 30% (Chronos-2)Chronos-2 · fev-bench univariateMASE: 1.62 -> 2.16Hurt
use TS-ICL instead of Chronos-2 for inference speedforecasting · fev-bench univariateinference time (s/window): 3.53 x 10^-3 -> 1.54 x 10^-2Hurt
use TS-ICL instead of TabICLv2-TS for inference speedimputation · fm-impute-bench univariateinference time (s/window): 3.07 x 10^-1 -> 6.51 x 10^-3Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy