面向行人跟踪的基于离线专家的在线残差学习
Online Residual Learning from Offline Experts for Pedestrian Tracking
AI总结：
针对在线行人轨迹预测，本文提出 ORL，将多个离线预测与递推最小二乘在线残差校正结合，并通过专家建议框架和自适应 softmax 聚合，实现遗憾保证与最佳性能。
摘要：
本文研究从数据中预测未知目标的问题。我们提出在线残差学习（ORL），这是一种将在线自适应与离线训练预测相结合的方法。在较低层级，我们采用在预测时域之前或开始时生成的多个离线预测。我们通过使用递推最小二乘算法在线学习每个离线预测相对于真实目标状态的各自残差误差，来增强每一个离线预测。在较高层级，我们将增强后的低层预测器视为专家，采用“带专家建议的预测”框架。我们利用自适应 softmax 加权方案形成聚合预测，并为 ORL 提供关于遗憾的保证。我们将 ORL 用于提升在线行人轨迹预测场景中的性能。基于 Stanford Drone Dataset 的数据，我们表明 ORL 能够展现出两全其美的性能。
英文摘要：
In this paper, we consider the problem of predicting unknown targets from data. We propose Online Residual Learning (ORL), a method that combines online adaptation with offline-trained predictions. At a lower level, we employ multiple offline predictions generated before or at the beginning of the prediction horizon. We augment every offline prediction by learning their respective residual error concerning the true target state online, using the recursive least squares algorithm. At a higher level, we treat the augmented lower-level predictors as experts, adopting the Prediction with Expert Advice framework. We utilize an adaptive softmax weighting scheme to form an aggregate prediction and provide guarantees for ORL in terms of regret. We employ ORL to boost performance in the setting of online pedestrian trajectory prediction. Based on data from the Stanford Drone Dataset, we show that ORL can demonstrate best-of-both-worlds performance.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
