# TrACT: A Training Dynamics Aware Contrastive Learning Framework for Long-tail Trajectory Prediction

arXiv: https://arxiv.org/abs/2404.12538 · HTML: https://mlsanity.com/paper/2404.12538/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 33 of its 44 references that are in the map.
- First posted on arXiv 18 Apr 2024.

## Builds on (most cited first, 30 of 33)

- [nuScenes: A multimodal dataset for autonomous driving](https://mlsanity.com/paper/1903.11027/) (2019)
- [Scalability in Perception for Autonomous Driving: Waymo Open Dataset](https://mlsanity.com/paper/1912.04838/) (2019)
- [Class-Balanced Loss Based on Effective Number of Samples](https://mlsanity.com/paper/1901.05555/) (2019)
- [Learning Imbalanced Datasets with Label-Distribution-Aware Margin Loss](https://mlsanity.com/paper/1906.07413/) (2019)
- [Argoverse: 3D Tracking and Forecasting with Rich Maps](https://mlsanity.com/paper/1911.02620/) (2019)
- [Prototypical Contrastive Learning of Unsupervised Representations](https://mlsanity.com/paper/2005.04966/) (2020)
- [Long-tail learning via logit adjustment](https://mlsanity.com/paper/2007.07314/) (2020)
- [Dataset Cartography: Mapping and Diagnosing Datasets with Training Dynamics](https://mlsanity.com/paper/2009.10795/) (2020)
- [Learning Lane Graph Representations for Motion Forecasting](https://mlsanity.com/paper/2007.13732/) (2020)
- [AgentFormer: Agent-Aware Transformers for Socio-Temporal Multi-Agent Forecasting](https://mlsanity.com/paper/2103.14023/) (2021)
- [Social-STGCNN: A Social Spatio-Temporal Graph Convolutional Neural Network for Human Trajectory Prediction](https://mlsanity.com/paper/2002.11927/) (2020)
- [MotionDiffuser: Controllable Multi-Agent Motion Prediction using Diffusion](https://mlsanity.com/paper/2306.03083/) (2023)
- [Delving into Deep Imbalanced Regression](https://mlsanity.com/paper/2102.09554/) (2021)
- [GOHOME: Graph-Oriented Heatmap Output for future Motion Estimation](https://mlsanity.com/paper/2109.01827/) (2021)
- [Leapfrog Diffusion Model for Stochastic Trajectory Prediction](https://mlsanity.com/paper/2303.10895/) (2023)
- [Relay Backpropagation for Effective Learning of Deep Convolutional Neural Networks](https://mlsanity.com/paper/1512.05830/) (2015)
- [HDGT: Heterogeneous Driving Graph Transformer for Multi-Agent Trajectory Prediction via Scene Encoding](https://mlsanity.com/paper/2205.09753/) (2022)
- [Spatially-Aware Graph Neural Networks for Relational Behavior Forecasting from Sensor Data](https://mlsanity.com/paper/1910.08233/) (2019)
- [TPCN: Temporal Point Cloud Networks for Motion Forecasting](https://mlsanity.com/paper/2103.03067/) (2021)
- [MUSE-VAE: Multi-Scale VAE for Environment-Aware Long Term Trajectory Prediction](https://mlsanity.com/paper/2201.07189/) (2022)
- [GANet: Goal Area Network for Motion Forecasting](https://mlsanity.com/paper/2209.09723/) (2022)
- [On Exposing the Challenging Long Tail in Future Prediction of Traffic Actors](https://mlsanity.com/paper/2103.12474/) (2021)
- [Vehicle trajectory prediction works, but not everywhere](https://mlsanity.com/paper/2112.03909/) (2021)
- [Context-Aware Trajectory Prediction](https://mlsanity.com/paper/1705.02503/) (2017)
- [Bifold and Semantic Reasoning for Pedestrian Behavior Prediction](https://mlsanity.com/paper/2012.03298/) (2020)
- [FEND: A Future Enhanced Distribution-Aware Contrastive Learning Framework for Long-tail Trajectory Prediction](https://mlsanity.com/paper/2303.16574/) (2023)
- [PedFormer: Pedestrian Behavior Prediction via Cross-Modal Attention Modulation and Gated Multitask Learning](https://mlsanity.com/paper/2210.07886/) (2022)
- [LatentFormer: Multi-Agent Transformer-Based Interaction Modeling and Trajectory Prediction](https://mlsanity.com/paper/2203.01880/) (2022)
- [IPCC-TP: Utilizing Incremental Pearson Correlation Coefficient for Joint Multi-Agent Trajectory Prediction](https://mlsanity.com/paper/2303.00575/) (2023)
- [A Novel Benchmarking Paradigm and a Scale- and Motion-Aware Model for Egocentric Pedestrian Trajectory Prediction](https://mlsanity.com/paper/2310.10424/) (2023)

## Built on by (most cited first, 7 of 7)

- [Curb Your Attention: Causal Attention Gating for Robust Trajectory Prediction in Autonomous Driving](https://mlsanity.com/paper/2410.07191/) (2024)
- [Beyond Imitation: Learning Safe End-to-End Autonomous Driving from Hard Negatives](https://mlsanity.com/paper/2605.19771/) (2026)
- [AMD: Adaptive Momentum and Decoupled Contrastive Learning Framework for Robust Long-Tail Trajectory Prediction](https://mlsanity.com/paper/2507.01801/) (2025)
- [Den-TP: A Density-Balanced Data Curation and Evaluation Framework for Trajectory Prediction](https://mlsanity.com/paper/2409.17385/) (2024)
- [Ensemble of Pre-Trained Models for Long-Tailed Trajectory Prediction](https://mlsanity.com/paper/2509.13914/) (2025)
- [AdaSports-Traj: Role- and Domain-Aware Adaptation for Multi-Agent Trajectory Modeling in Sports](https://mlsanity.com/paper/2509.16095/) (2025)
- [SAIL: Scene-aware Adaptive Iterative Learning for Long-Tail Trajectory Prediction in Autonomous Vehicles](https://mlsanity.com/paper/2604.04573/) (2026)

## Abstract (arXiv)

As a safety critical task, autonomous driving requires accurate predictions of road users' future trajectories for safe motion planning, particularly under challenging conditions. Yet, many recent deep learning methods suffer from a degraded performance on the challenging scenarios, mainly because these scenarios appear less frequently in the training data. To address such a long-tail issue, existing methods force challenging scenarios closer together in the feature space during training to trigger information sharing among them for more robust learning. These methods, however, primarily rely on the motion patterns to characterize scenarios, omitting more informative contextual information, such as interactions and scene layout. We argue that exploiting such information not only improves prediction accuracy but also scene compliance of the generated trajectories. In this paper, we propose to incorporate richer training dynamics information into a prototypical contrastive learning framework. More specifically, we propose a two-stage process. First, we generate rich contextual features using a baseline encoder-decoder framework. These features are split into clusters based on the model's output errors, using the training dynamics information, and a prototype is computed within each cluster. Second, we retrain the model using the prototypes in a contrastive learning framework. We conduct empirical evaluations of our approach using two large-scale naturalistic datasets and show that our method achieves state-of-the-art performance by improving accuracy and scene compliance on the long-tail samples. Furthermore, we perform experiments on a subset of the clusters to highlight the additional benefit of our approach in reducing training bias.