FLOC 2018: FEDERATED LOGIC CONFERENCE 2018
Policing Functions for Machine Learning Systems

Authors: Thai Son Hoang, Naoto Sato, Tomoyuki Myojin, Michael Butler, Yuichiroh Nakagawa and Hideto Ogawa

Paper Information

Title:Policing Functions for Machine Learning Systems
Authors:Thai Son Hoang, Naoto Sato, Tomoyuki Myojin, Michael Butler, Yuichiroh Nakagawa and Hideto Ogawa
Proceedings:VaVAS Proceedings
Editors: Clare Dixon, Brian Logan and Alessio Lomuscio
Keywords:Policing function, Metamorphic Testing, Formal Methods, Machine Learning Systems
Abstract:

ABSTRACT. Machine learning systems typically involve complex decision making mechanisms while lack clear and concise specifications. Demonstrating the quality of machine learning systems therefore is a challenging task. We propose an approach combining formal methods and metamorphic testing for improving the quality of machine learning systems. In particular, our framework enables the possibility of developing policing functions for runtime monitoring machine learning systems based on metamorphic relations.

Pages:10
Talk:Jul 18 11:30 (Session 126P: Distributed Systems II)
Paper: