Toward Pattern-Oriented Machine Learning Reliability Argumentation

Machine learning systems must be reliable. The Goal Structure Notation and Machine Learning Reliability Solution Patterns are often utilized to demonstrate the reliability of machine learning systems. However, there are instances where engineers fail to recognize the possibility of utilizing patterns because they lack prior knowledge and are not machine learning experts. To overcome these challenges, herein we propose the Pattern Application Support Framework. This framework should guide engineers and facilitate accurate pattern applications as it complements the lack of prior knowledge and guides engineers when applying Machine Learning Reliability Solution Patterns.