![[2602.19113] Learning from Complexity: Exploring Dynamic Sample Pruning of Spatio-Temporal Training](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.19113] Learning from Complexity: Exploring Dynamic Sample Pruning of Spatio-Temporal Training
Summary
The paper explores dynamic sample pruning techniques for spatio-temporal training, aiming to enhance training efficiency and model performance in deep learning applications.
Why It Matters
This research addresses a critical bottleneck in machine learning by optimizing the training process for spatio-temporal data, which is essential in various fields like transportation and climate science. By improving training efficiency, it can lead to faster model convergence and better resource utilization, making it highly relevant for practitioners and researchers in AI and machine learning.
Key Takeaways
- Dynamic sample pruning can significantly accelerate training speed.
- The ST-Prune method maintains or improves model performance.
- The approach is scalable and applicable to various real-world datasets.
- Optimizing training data usage can reduce computational costs.
- This technique addresses inefficiencies in traditional training methods.
Computer Science > Machine Learning arXiv:2602.19113 (cs) [Submitted on 22 Feb 2026] Title:Learning from Complexity: Exploring Dynamic Sample Pruning of Spatio-Temporal Training Authors:Wei Chen, Junle Chen, Yuqian Wu, Yuxuan Liang, Xiaofang Zhou View a PDF of the paper titled Learning from Complexity: Exploring Dynamic Sample Pruning of Spatio-Temporal Training, by Wei Chen and 4 other authors View PDF HTML (experimental) Abstract:Spatio-temporal forecasting is fundamental to intelligent systems in transportation, climate science, and urban planning. However, training deep learning models on the massive, often redundant, datasets from these domains presents a significant computational bottleneck. Existing solutions typically focus on optimizing model architectures or optimizers, while overlooking the inherent inefficiency of the training data itself. This conventional approach of iterating over the entire static dataset each epoch wastes considerable resources on easy-to-learn or repetitive samples. In this paper, we explore a novel training-efficiency techniques, namely learning from complexity with dynamic sample pruning, ST-Prune, for spatio-temporal forecasting. Through dynamic sample pruning, we aim to intelligently identify the most informative samples based on the model's real-time learning state, thereby accelerating convergence and improving training efficiency. Extensive experiments conducted on real-world spatio-temporal datasets show that ST-Prune significantl...
