新基准Labyrinth可精准测试模仿学习的泛化能力。
Quantifying Generalisation in Imitation Learning
- 构建可控环境，分离训练、评估与测试场景
- 支持可重复实验，验证不同泛化因素的影响
- 适合研究泛化性与鲁棒性的人工智能学者
模仿学习基准常因训练与评估设置差异不足，难以有效评估泛化性能。我们提出Labyrinth，一个可精确控制结构、起始与目标位置及任务复杂度的基准环境，实现训练、评估与测试设置的明确区分。该环境具备离散且完全可观测的状态空间，以及已知最优动作，支持可解释性与细粒度评估。其灵活架构可针对性测试泛化因素，包含部分可观测性、钥匙-门任务与冰面障碍等变体。通过支持受控、可复现的实验，Labyrinth推动了模仿学习泛化评估的发展，为开发更鲁棒智能体提供重要工具。
原文摘要 · Abstract (English)
Imitation learning benchmarks often lack sufficient variation between training and evaluation, limiting meaningful generalisation assessment. We introduce Labyrinth, a benchmarking environment designed to test generalisation with precise control over structure, start and goal positions, and task complexity. It enables verifiably distinct training, evaluation, and test settings. Labyrinth provides a discrete, fully observable state space and known optimal actions, supporting interpretability and fine-grained evaluation. Its flexible setup allows targeted testing of generalisation factors and includes variants like partial observability, key-and-door tasks, and ice-floor hazards. By enabling controlled, reproducible experiments, Labyrinth advances the evaluation of generalisation in imitation learning and provides a valuable tool for developing more robust agents.
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。