Evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles

In a safety-critical system like autonomous vehicles, it is essential to ensure that the observations shown are within the distribution of training data, otherwise they are called out-of-distribution (OOD). OOD detection is a fundamental problem that needs to be addressed to avoid errors in image re...

Full description

Saved in:
Bibliographic Details
Main Author: Dinh, Phuc Hung
Other Authors: Arvind Easwaran
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2023
Subjects:
Online Access:https://hdl.handle.net/10356/166524
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-166524
record_format dspace
spelling sg-ntu-dr.10356-1665242023-05-05T15:41:36Z Evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles Dinh, Phuc Hung Arvind Easwaran School of Computer Science and Engineering arvinde@ntu.edu.sg Engineering::Computer science and engineering In a safety-critical system like autonomous vehicles, it is essential to ensure that the observations shown are within the distribution of training data, otherwise they are called out-of-distribution (OOD). OOD detection is a fundamental problem that needs to be addressed to avoid errors in image recognition tasks, especially in real-time system. Variational autoencoder (VAE) has emerged as the most promising method to address this issue. Several modifications have been made to VAE to improve its performance, especially in terms of increasing disentanglement, yet no research has been done to evaluate its performance on OOD detection. In this research, four VAE variants were tested on a traffic dataset to see which one gives the best results. After which, the relationship between disentanglement and OOD detection is evaluated. Bachelor of Engineering (Computer Science) 2023-05-04T02:05:17Z 2023-05-04T02:05:17Z 2023 Final Year Project (FYP) Dinh, P. H. (2023). Evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/166524 https://hdl.handle.net/10356/166524 en application/pdf Nanyang Technological University
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Engineering::Computer science and engineering
spellingShingle Engineering::Computer science and engineering
Dinh, Phuc Hung
Evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles
description In a safety-critical system like autonomous vehicles, it is essential to ensure that the observations shown are within the distribution of training data, otherwise they are called out-of-distribution (OOD). OOD detection is a fundamental problem that needs to be addressed to avoid errors in image recognition tasks, especially in real-time system. Variational autoencoder (VAE) has emerged as the most promising method to address this issue. Several modifications have been made to VAE to improve its performance, especially in terms of increasing disentanglement, yet no research has been done to evaluate its performance on OOD detection. In this research, four VAE variants were tested on a traffic dataset to see which one gives the best results. After which, the relationship between disentanglement and OOD detection is evaluated.
author2 Arvind Easwaran
author_facet Arvind Easwaran
Dinh, Phuc Hung
format Final Year Project
author Dinh, Phuc Hung
author_sort Dinh, Phuc Hung
title Evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles
title_short Evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles
title_full Evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles
title_fullStr Evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles
title_full_unstemmed Evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles
title_sort evaluating variational autoencoder methods for out-of-distribution detection in autonomous vehicles
publisher Nanyang Technological University
publishDate 2023
url https://hdl.handle.net/10356/166524
_version_ 1765213826571042816