arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1708.03322cs.LG
收藏

多层神经网络的输出可达集估计与验证

Output Reachable Set Estimation and Verification for Multi-Layer Neural Networks

Vanderbilt University(范德堡大学)

Weiming Xiang, Hoang-Dung Tran, Taylor T. Johnson

摘要:

研究了多层感知器神经网络的输出可达估计和安全性验证问题。首先,引入最大灵敏度的概念,对于一类激活函数为单调函数的多层感知器,通过求解凸优化问题来计算最大灵敏度。然后,利用基于仿真的方法,将神经网络的输出可达集估计问题转化为一系列优化问题。最后,基于输出可达集估计结果开发了自动安全验证。最后以一个双关节机械臂模型的安全性验证为例,验证了所提方法的有效性。

英文摘要:

In this paper, the output reachable estimation and safety verification problems for multi-layer perceptron neural networks are addressed. First, a conception called maximum sensitivity in introduced and, for a class of multi-layer perceptrons whose activation functions are monotonic functions, the maximum sensitivity can be computed via solving convex optimization problems. Then, using a simulation-based method, the output reachable set estimation problem for neural networks is formulated into a chain of optimization problems. Finally, an automated safety verification is developed based on the output reachable set estimation result. An application to the safety verification for a robotic arm model with two joints is presented to show the effectiveness of proposed approaches.

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。

↑