Browse wiki
From MurrayWiki
Inverse Abstraction of Neural Networks Using Symbolic Interpolation |
Abstract |
Neural networks in real-world applications … Neural networks in real-world applications have to satisfy critical properties such as safety and reliability. The analysis of such properties typically involves extracting informa- tion through computing pre-images of neural networks, but it is well-known that explicit computation of pre-images is intractable. We introduce new methods for computing compact symbolic abstractions of pre-images. Our approach relies on computing approximations that provably overapproximate and underapproximate the pre-images at all layers. The abstraction of pre-images enables formal analysis and knowl- edge extraction without modifying standard learning algo- rithms. We show how to use inverse abstractions to automatically extract simple control laws and compact representations for pre-images corresponding to unsafe outputs. We illustrate that the extracted abstractions are often interpretable and can be used for analyzing complex properties. be used for analyzing complex properties. +
|
---|---|
Authors | Sumanth Dathathri, Sicun Gao, Richard M. Murray + |
Funding | VeHICaL: Verified Human Interfaces, Control, and Learning for Semi-Autonomous Systems + |
ID | 2018e + |
Source | To appear, 2019 AAAI Conference on Artificial Intelligence + |
Tag | dgm19-aiaa + |
Title | Inverse Abstraction of Neural Networks Using Symbolic Interpolation + |
Type | Conference paper + |
Categories | Papers |
Modification date This property is a special property in this wiki.
|
27 December 2018 05:21:34 + |
URL This property is a special property in this wiki.
|
http://www.cds.caltech.edu/~murray/preprints/dgm19-aiaa.pdf + |
hide properties that link here |
Inverse Abstraction of Neural Networks Using Symbolic Interpolation + | Title |
---|