Methods for the Explanation of Machine Learning Processes and Results for Non-Experts
Machine learning systems such as artificial neural networks (ANNs) and support vector machines (SVMs) are commonplace in science, technology, business and health. However, the lack of an explanation capability is an impediment to the more wide-spread use of machine learning systems. For more than 20 years, the field of rule extraction from ANNs and SVMs has tried to generate explanations for machine learning based decisions in the form of propositional, probabilistic or first-order rules. However, there is no generally accepted technique that works for all machine learning systems. In addition, there is no method that is tailored to the needs of a user who may not be a domain expert and may not have a technology background. This paper introduces a number of techniques for the generation of multi-media clips that explain the behavior of a machine learning system to a non-expert user.
Diederich, J. (2018, April 3). Methods for the Explanation of Machine Learning Processes and Results for Non-Experts. https://doi.org/10.31234/osf.io/54eub
Sponsor: Discover men's top designer new arrivals on biffi.com