arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1903.00520eess.SYcs.SY
收藏

一种用于验证具有深度神经网络控制器的动力系统的方法

A Reachability Method for Verifying Dynamical Systems with Deep Neural Network Controllers

Stanford University(斯坦福大学)

Kyle D. Julian, Mykel J. Kochenderfer

AI总结:

本文提出了一种结合可达性方法和开源神经网络验证工具的方法,用于在多个时间步上为深度神经网络控制器提供保证,通过限制系统动态和神经网络输出来过估计可达状态集,从而确保系统不会进入该集之外的状态。

摘要:

深度神经网络可以训练成为高效且有效的动力系统控制器;然而,深度神经网络的机制复杂且难以保证。本文提出了一种通用方法,通过结合可达性方法和开源神经网络验证工具,为深度神经网络控制器在多个时间步上提供保证。通过限制系统动态和神经网络输出,可以过估计可达状态集,从而提供保证系统不会进入该集之外的状态。该方法在山车问题以及飞机避障问题上得到了演示。结果表明,该方法可以在给定有界动态模型的情况下提供神经网络保证。

英文摘要:

Deep neural networks can be trained to be efficient and effective controllers for dynamical systems; however, the mechanics of deep neural networks are complex and difficult to guarantee. This work presents a general approach for providing guarantees for deep neural network controllers over multiple time steps using a combination of reachability methods and open source neural network verification tools. By bounding the system dynamics and neural network outputs, the set of reachable states can be over-approximated to provide a guarantee that the system will never reach states outside the set. The method is demonstrated on the mountain car problem as well as an aircraft collision avoidance problem. Results show that this approach can provide neural network guarantees given a bounded dynamic model.

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。

↑