Files in this item

FilesDescriptionFormat

application/pdf

application/pdfCUI-THESIS-2021.pdf (6MB)Restricted to U of Illinois
(no description provided)PDF

Description

Title:AI-driven methods for resiliency and security assessment: the case for autonomous driving system and HPC storage system
Author(s):Cui, Shengkun
Advisor(s):Kalbarczyk, Zbigniew T.
Department / Program:Electrical & Computer Eng
Discipline:Electrical & Computer Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:M.S.
Genre:Thesis
Subject(s):Autonomous Driving System
Autonomous Vehicle
HPC System
Artificial Intelligence
Machine Learning
Fault-tolerant System
Reliability
Security
Adversarial Attack
Fault Detection
Failure Localization
Failure Diagnosis
Abstract:Nowadays, computing systems are used extensively in mission-critical exploration, transportation, scientific study, and manufacturing. With the advances in computation technologies, computing systems have become ever more complex. Due to the system’s complexity, it is increasingly hard for humans operators to monitor, assess, and manage the system directly. Moreover, traditional model-based or rule-based assessment techniques cannot provide sufficient coverages because of the wide range of use cases and failure modes of complex systems. Recently artificial intelligence (AI)-driven methods are used for timely accurate and high-coverage assessments in complex systems because of their ability to learn from data without explicitly modeling the complex systems. This thesis discusses our work on AI-driven assessment methods—RoboTack, DiverseAV, and Kaleidoscope—in the domain of security (RoboTack) and reliability (DiverseAV, Kaleidoscope) assessment in two critical use cases: autonomous driving systems (ADS) and high-performance computing (HPC) storage systems. We show that by using artificial intelligence and machine learning-based techniques, we can perform high-accuracy, high-coverage security, or reliability assessments of large-scale, complex systems efficiently in real-time.
Issue Date:2021-04-14
Type:Thesis
URI:http://hdl.handle.net/2142/110655
Rights Information:Copyright 2021 Shengkun Cui
Date Available in IDEALS:2021-09-17
Date Deposited:2021-05


This item appears in the following Collection(s)

Item Statistics