With the advent of autonomous systems and deep learning systems, safety pertaining to these systems has become a major concern. The existing failure analysis techniques are not enough to thoroughly analyze the safety in these systems. Moreover, because these systems are created to operate in various conditions, they are susceptible to unknown safety issues. Hence, we need mechanisms which can take into account the complexity of operational design domains, identify safety issues other than failures, and expose unknown safety issues. Moreover, existing safety analysis approaches require a lot of effort and time for analysis and do not consider machine learning (ML) safety. To address these limitations, in this dissertation, we discuss an artificial-intelligence driven model-based methodology that aids in identifying unknown safety issues and analyzing ML safety. Our methodology consists of 4 major tasks: 1) automated model generation, 2) automated analysis of component state transition model specification, 3) undesired states analysis, and 4) causal factor analysis. In our methodology we identify unknown safety issues by finding undesired combinations of components' states and environmental entities' states as well as causes resulting in these undesired combinations. In our methodology, we refer to the behaviors that occur because of undesired combinations as off-nominal behaviors (ONBs). To identify undesired combinations and ONBs that aid in exposing unknown safety issues with less effort and time we proposed various approaches for each of the task and performed corresponding empirical studies. We also discussed machine learning safety analysis from the perspective of machine learning engineers as well as system and software safety engineers. The results of studies conducted as part of our research shows that our proposed methodology helps in identifying unknown safety issues effectively. Our results also show that combinatorial methods are effective in reducing effort and time for analysis of off-nominal behaviors without overlooking any dependencies among components and environmental entities of a system. We also found that safety analysis of machine learning components is different from analysis of conventional software components and detail the aspects we need to consider for ML safety.
| Identifer | oai:union.ndltd.org:unt.edu/info:ark/67531/metadc1808442 |
| Date | 05 1900 |
| Creators | Madala, Kaushik |
| Contributors | Do, Hyunsook, Blanco, Eduardo, Ludi, Stephanie, Bryant, Barrett R |
| Publisher | University of North Texas |
| Source Sets | University of North Texas |
| Language | English |
| Detected Language | English |
| Type | Thesis or Dissertation |
| Format | xi, 150 pages, Text |
| Rights | Public, Madala, Kaushik, Copyright, Copyright is held by the author, unless otherwise noted. All rights Reserved. |
Page generated in 0.0128 seconds