# ALCo-FM: Adaptive Long-Context Foundation Model for Accident Prediction

arXiv: https://arxiv.org/abs/2507.08153 · HTML: https://mlsanity.com/paper/2507.08153/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Chain-of-thought reasoning for driving, one of 17 directions in LLM reasoning.
- Builds on 26 of its 60 references that are in the map.
- First posted on arXiv 10 Jul 2025.

## Placed in

- LLM reasoning / [Chain-of-thought reasoning for driving](https://mlsanity.com/lr/chain-of-thought-reasoning-for-driving/)

## Builds on (most cited first, 26 of 26)

- [XGBoost: A Scalable Tree Boosting System](https://mlsanity.com/paper/1603.02754/) (2016)
- [Billion-scale similarity search with GPUs](https://mlsanity.com/paper/1702.08734/) (2017)
- [Spatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting](https://mlsanity.com/paper/1709.04875/) (2017)
- [Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet](https://mlsanity.com/paper/2101.11986/) (2021)
- [Graph WaveNet for Deep Spatial-Temporal Graph Modeling](https://mlsanity.com/paper/1906.00121/) (2019)
- [Adaptive Computation Time for Recurrent Neural Networks](https://mlsanity.com/paper/1603.08983/) (2016)
- [Spatial-Temporal Fusion Graph Neural Networks for Traffic Flow Forecasting](https://mlsanity.com/paper/2012.09641/) (2020)
- [Spatial-Temporal Graph ODE Networks for Traffic Flow Forecasting](https://mlsanity.com/paper/2106.12931/) (2021)
- [SAITS: Self-Attention-based Imputation for Time Series](https://mlsanity.com/paper/2202.08516/) (2022)
- [Dynamic Graph Convolutional Recurrent Network for Traffic Prediction: Benchmark and Solution](https://mlsanity.com/paper/2104.14917/) (2021)
- [Adaptive Attention Span in Transformers](https://mlsanity.com/paper/1905.07799/) (2019)
- [Uncertainty-based Traffic Accident Anticipation with Spatio-Temporal Relational Learning](https://mlsanity.com/paper/2008.00334/) (2020)
- [ViT-CoMer: Vision Transformer with Convolutional Multi-scale Feature Interaction for Dense Predictions](https://mlsanity.com/paper/2403.07392/) (2024)
- [ContiFormer: Continuous-Time Transformer for Irregular Time Series Modeling](https://mlsanity.com/paper/2402.10635/) (2024)
- [Fully-Connected Spatial-Temporal Graph for Multivariate Time-Series Data](https://mlsanity.com/paper/2309.05305/) (2023)
- [MST-GAT: A Multimodal Spatial-Temporal Graph Attention Network for Time Series Anomaly Detection](https://mlsanity.com/paper/2310.11169/) (2023)
- [A Countrywide Traffic Accident Dataset](https://mlsanity.com/paper/1906.05409/) (2019)
- [Automated Dilated Spatio-Temporal Synchronous Graph Modeling for Traffic Prediction](https://mlsanity.com/paper/2207.10830/) (2022)
- [GTC: GNN-Transformer Co-contrastive Learning for Self-supervised Heterogeneous Graph Representation](https://mlsanity.com/paper/2403.15520/) (2024)
- [Urban Traffic Accident Risk Prediction Revisited: Regionality, Proximity, Similarity and Sparsity](https://mlsanity.com/paper/2407.19668/) (2024)
- [CrashFormer: A Multimodal Architecture to Predict the Risk of Crash](https://mlsanity.com/paper/2402.05151/) (2024)
- [Risk Prediction on Traffic Accidents using a Compact Neural Model for Multimodal Information Fusion over Urban Big Data](https://mlsanity.com/paper/2103.05107/) (2021)
- [Will there be a construction? Predicting road constructions based on heterogeneous spatiotemporal data](https://mlsanity.com/paper/2209.06813/) (2022)
- [Iterative missing value imputation based on feature importance](https://mlsanity.com/paper/2311.08005/) (2023)
- [SST-GCN: The Sequential based Spatio-Temporal Graph Convolutional networks for Minute-level and Road-level Traffic Accident Risk Prediction](https://mlsanity.com/paper/2405.18602/) (2024)
- [DocParseNet: Advanced Semantic Segmentation and OCR Embeddings for Efficient Scanned Document Annotation](https://mlsanity.com/paper/2406.17591/) (2024)

## Built on by (most cited first, 1 of 1)

- [DocVAL: Validated Chain-of-Thought Distillation for Grounded Document VQA](https://mlsanity.com/paper/2511.22521/) (2025)

## Abstract (arXiv)

Traffic accidents are rare, yet high-impact events that require long-context multimodal reasoning for accurate risk forecasting. In this paper, we introduce ALCo-FM, a unified adaptive long-context foundation model that computes a volatility pre-score to dynamically select context windows for input data and encodes and fuses these multimodal data via shallow cross attention. Following a local GAT layer and a BigBird-style sparse global transformer over H3 hexagonal grids, coupled with Monte Carlo dropout for confidence, the model yields superior, well-calibrated predictions. Trained on data from 15 US cities with a class-weighted loss to counter label imbalance, and fine-tuned with minimal data on held-out cities, ALCo-FM achieves 0.94 accuracy, 0.92 F1, and an ECE of 0.04, outperforming more than 20 state-of-the-art baselines in large-scale urban risk prediction. Code and dataset are available at: https://github.com/PinakiPrasad12/ALCo-FM