Stochastic World Models for Verifying Vision-Based Neural Feedback Systems· 验证基于视觉的神经反馈系统的随机世界模型
Verifying a vision-based neural feedback system requires a model of the observations its controller acts upon. Such a model must capture the variation the sensor produces, while remaining tractable for closed-loop analysis. Generative adversarial networks (GANs) have served as perception surrogates, but they are large, reproduce complex scenes poorly, and are hard to verify. We explore stochastic world models as a richer class of perception surrogates. We train a world model with physically grounded latents, built from operations that standard verifiers bound. It reproduces held-out frames more faithfully than GAN surrogates with up to 130 times as many parameters. To verify these surrogates, we develop a procedure that combines falsification, adaptive refinement, symbolic, and backward an
利用随机世界模型改进基于视觉的神经反馈系统的验证方法。
- 核心方法
- 提出了使用物理基础潜在变量训练的随机世界模型，结合反例生成、自适应细化、符号化和逆向分析等技术进行验证。
- 适合谁读
- 适合研究者和工程师阅读，特别是关注系统验证和机器学习模型应用的读者。
- 要解决的问题
- 现有的基于视觉的神经反馈系统验证方法（如GANs）难以捕捉传感器变化且不易验证。
- 关键实验
- 实验显示，所提出的模型相比具有高达130倍参数量的GAN代理模型，在重现保留帧方面表现更好。
- 主要贡献
- 模型在参数量更少的情况下能更忠实地重现保留帧，且易于进行验证。
- 意义与局限
- 该方法提高了基于视觉的神经反馈系统验证的效率和准确性，但可能在复杂场景生成方面仍有限制。