提出新算法处理不规则时空任务，高效且更省资源。
Efficiently Scanning and Resampling Spatio-Temporal Tasks with Irregular Observations
- 交替使用二维隐状态与观测的交叉注意力和序列累积求和
- 在两个新任务上实现更高效率，参数少、训练推理更快
- 适合多智能体动态环境建模，如游戏策略分析
许多工作致力于结合循环模型的推理效率与多头注意力的训练并行性以实现序列建模。然而，多数方法聚焦于固定维度观测空间的任务，如语言建模中的单个词元或图像补全中的像素。为应对观测空间大小变化的情况，本文提出一种新算法：在二维隐状态与观测之间交替进行交叉注意力，并沿序列维度执行折扣累积和，以高效积累历史信息。我们发现该重采样循环对性能至关重要。为评估此领域下的高效序列建模，引入两个多智能体意图任务：模拟智能体追逐弹跳粒子，以及专业星露谷游戏（StarCraft II）中的微操分析。所提算法在参数量更少的前提下，实现了与现有方法相当的准确率，且训练与推理速度更快。
原文摘要 · Abstract (English)
Various works have aimed at combining the inference efficiency of recurrent models and training parallelism of multi-head attention for sequence modeling. However, most of these works focus on tasks with fixed-dimension observation spaces, such as individual tokens in language modeling or pixels in image completion. To handle an observation space of varying size, we propose a novel algorithm that alternates between cross-attention between a 2D latent state and observation, and a discounted cumulative sum over the sequence dimension to efficiently accumulate historical information. We find this resampling cycle is critical for performance. To evaluate efficient sequence modeling in this domain, we introduce two multi-agent intention tasks: simulated agents chasing bouncing particles and micromanagement analysis in professional StarCraft II games. Our algorithm achieves comparable accuracy with a lower parameter count, faster training and inference compared to existing methods.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。