mlsanityData to 14 Oct 2026

Temporal Label Smoothing for Early Event Prediction

arXiv 2208.13764 ↗First posted 29 Aug 2022cs.LGcited by 12

Not on the map. Why?

Abstract (arXiv)

Models that can predict the occurrence of events ahead of time with low false-alarm rates are critical to the acceptance of decision support systems in the medical community. This challenging task is typically treated as a simple binary classification, ignoring temporal dependencies between samples, whereas we propose to exploit this structure. We first introduce a common theoretical framework unifying dynamic survival analysis and early event prediction. Following an analysis of objectives from both fields, we propose Temporal Label Smoothing (TLS), a simpler, yet best-performing method that preserves prediction monotonicity over time. By focusing the objective on areas with a stronger predictive signal, TLS improves performance over all baselines on two large-scale benchmark tasks. Gains are particularly notable along clinically relevant measures, such as event recall at low false-alarm rates. TLS reduces the number of missed events by up to a factor of two over previously used approaches in early event prediction.

Key facts. Cited by 12 papers in the map (data through 14 Oct 2026). Builds on 21 of its 55 references that are in the map. First posted on arXiv 29 Aug 2022.

Builds on

21 of 55 references.

References · 21

  1. 2017Attention Is All You Needcited by 60,361
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,305
  3. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,475
  4. 2014Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modelingcited by 2,367
  5. 2019When Does Label Smoothing Help?cited by 988
  6. 2017Multitask learning and benchmarking with clinical time series datacited by 290
  7. 2014cuDNN: Efficient Primitives for Deep Learningcited by 245
  8. 2021Multi-Time Attention Networks for Irregularly Sampled Time Seriescited by 238
  9. 2020Does label smoothing mitigate label noise?cited by 204
  10. 2021Delving into Deep Imbalanced Regressioncited by 158
  11. 2019Set Functions for Time Seriescited by 138
  12. 2022PolyLoss: A Polynomial Expansion Perspective of Classification Loss Functionscited by 63
9 more
  1. 2018Deep Recurrent Survival Analysiscited by 61
  2. 2017Learning to Detect Sepsis with a Multitask Gaussian Process RNN Classifiercited by 47
  3. 2020Generalized Entropy Regularization or: There's Nothing Special about Label Smoothingcited by 41
  4. 2021HiRID-ICU-Benchmark -- A Comprehensive Machine Learning Benchmark on High-resolution ICU Datacited by 39
  5. 2020Regularization via Structural Label Smoothingcited by 29
  6. 2019Early detection of sepsis utilizing deep learning on electronic health record event sequencescited by 15
  7. 2021Survival stacking: casting survival analysis as a classification problemcited by 13
  8. 2020Evaluating Progress on Machine Learning for Longitudinal Electronic Healthcare Datacited by 12
  9. 2022Disability prediction in multiple sclerosis using performance outcome measures and demographic datacited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 2; 2025: 3; 2026: 7.

12 papers.

  1. 2024Dynamic Survival Analysis for Early Event Predictioncited by 5
  2. 2025Incremental Sequence Classification with Temporal Consistencycited by 3
  3. 2026RiskProp: Collision-Anchored Self-Supervised Risk Propagation for Early Accident Anticipationcited by 2
  4. 2025Data-Driven Discovery of Feature Groups in Clinical Time Seriescited by 1
  5. 2024Towards Dynamic Feature Acquisition on Medical Time Series by Maximizing Conditional Mutual Informationcited by 0
  6. 2025Stable Prediction of Adverse Events in Medical Time-Series Datacited by 0
  7. 2026Early Warning of Intraoperative Adverse Events via Transformer-Driven Multi-Label Learningcited by 0
  8. 2026MATA-Former & SIICU: Semantic Aware Temporal Alignment for High-Fidelity ICU Risk Predictioncited by 0
  9. 2026ChunQiuTR: Time-Keyed Temporal Retrieval in Classical Chinese Annalscited by 0
  10. 2026Event-Based Early Warning of Vineyard Disease Risk from Environmental Time Seriescited by 0
  11. 2026Weighted Score-Oriented Losses for Temporally Localized Event Predictioncited by 0
  12. 2026Horizon-Aware Early Event Prediction for Tokamak Disruption Alarmscited by 0