mlsanityData to 14 Oct 2026

Protocol before progress: leakage-aware evaluation of AIS trajectory prediction

arXiv 2609.25827 ↗First posted 22 Sep 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Reported gains in vessel-trajectory prediction from Automatic Identification System (AIS) data are credited to new architectures, but the evaluation protocol is rarely measured as a source of error reduction. We build a leakage-aware protocol with vessel-, time- and region-disjoint splits and apply it to two corpora with different traffic: 31 days of Danish national AIS traffic and 30 days of US Gulf coast traffic off Houston and Galveston. On both, we audit TrAISformer, GATransformer, and controlled AISFormer-inspired reconstructions. Three protocol effects appear in both corpora. First, TrAISformer's best-of-16 oracle decoder lowers error by a factor of 2.1-3.2 relative to greedy decoding. Second, a split that shares vessels lowers its greedy error by 23-25% at one hour, against 2% or less for a compact 0.43 M-parameter encoder. Third, a region-disjoint split raises TrAISformer's one-hour error from 2.2 to 24.6 km on the US corpus, because 99.9% of the test contexts fall in longitude bins never seen in training; the encoder built on local offsets is unaffected by this. Architectural mechanisms matter less: GATransformer's graph attention gives no measurable benefit on either corpus, while its waterway feature is worth 12-22%. The effect of a time-disjoint split is not stable across corpora (13% versus 2%). We release the splits and code.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 18 of its 32 references that are in the map. First posted on arXiv 22 Sep 2026.

Builds on

18 of 32 references.

References · 18

  1. 2022A Time Series is Worth 64 Words: Long-term Forecasting with Transformerscited by 2,039
  2. 2022FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecastingcited by 1,210
  3. 2017Deep Reinforcement Learning that Matterscited by 754
  4. 2019Argoverse: 3D Tracking and Forecasting with Rich Mapscited by 732
  5. 2018Social GAN: Socially Acceptable Trajectories with Generative Adversarial Networkscited by 690
  6. 2021CSDI: Conditional Score-based Diffusion Models for Probabilistic Time Series Imputationcited by 582
  7. 2016Recurrent Neural Networks for Multivariate Time Series with Missing Valuescited by 554
  8. 2020A Metric Learning Reality Checkcited by 321
  9. 2019Are We Really Making Much Progress? A Worrying Analysis of Recent Neural Recommendation Approachescited by 185
  10. 2022SAITS: Self-Attention-based Imputation for Time Seriescited by 169
  11. 2017On the State of the Art of Evaluation in Neural Language Modelscited by 104
  12. 2019Evaluating time series forecasting models: An empirical study on performance estimation methodscited by 53
6 more
  1. 2021Deep Learning Methods for Vessel Trajectory Prediction based on Recurrent Neural Networkscited by 32
  2. 2019Analyzing the Variety Loss in the Context of Probabilistic Trajectory Predictioncited by 31
  3. 2025AIS Data-Driven Maritime Monitoring Based on Transformer: A Comprehensive Reviewcited by 6
  4. 2024DiffuTraj: A Stochastic Vessel Trajectory Prediction Approach via Guided Diffusion Processcited by 1
  5. 2025STGDPM:Vessel Trajectory Prediction with Spatio-Temporal Graph Diffusion Probabilistic Modelcited by 1
  6. 2026EnvShip: A Unified Framework for Context-Aware and Cross-Region Vessel Trajectory Forecastingcited by 1

Built on by

Not cited yet.