2022
DOI: 10.48550/arxiv.2205.01938
|View full text |Cite
Preprint
|
Sign up to set email alerts
|

DeepFD: Automated Fault Diagnosis and Localization for Deep Learning Programs

Jialun Cao,
Meiziniu Li,
Xiao Chen
et al.

Abstract: As Deep Learning (DL) systems are widely deployed for missioncritical applications, debugging such systems becomes essential. Most existing works identify and repair suspicious neurons on the trained Deep Neural Network (DNN), which, unfortunately, might be a detour. Specifically, several existing studies have reported that many unsatisfactory behaviors are actually originated from the faults residing in DL programs. Besides, locating faulty neurons is not actionable for developers, while locating the faulty s… Show more

Help me understand this report
View published versions

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
13
0

Year Published

2024
2024
2024
2024

Publication Types

Select...
1
1

Relationship

0
2

Authors

Journals

citations
Cited by 2 publications
(13 citation statements)
references
References 39 publications
0
13
0
Order By: Relevance
“…In this direction, researchers have proposed several approaches centered on analyzing the runtime behavior during model training. Two examples are DeepFD [13] and Deep4Deep [12], both of which are learning-based frameworks designed for fault diagnosis and localization. These approaches involve the extraction of various runtime features from mutant models to train a machine-learning classifier for identifying fault types.…”
Section: Related Workmentioning
confidence: 99%
See 1 more Smart Citation
“…In this direction, researchers have proposed several approaches centered on analyzing the runtime behavior during model training. Two examples are DeepFD [13] and Deep4Deep [12], both of which are learning-based frameworks designed for fault diagnosis and localization. These approaches involve the extraction of various runtime features from mutant models to train a machine-learning classifier for identifying fault types.…”
Section: Related Workmentioning
confidence: 99%
“…Recently, numerous techniques for fault localization and repair in Deep Neural Networks (DNNs) have emerged. These techniques include UMLAUT [8], DeepLocalize [9], DeepDiagnosis [10], TheDeepChecker [11], Deep4Deep [12], and DeepFD [13]. They identify and localize bugs using various techniques such as static and dynamic analysis, along with machine learning techniques.…”
Section: Introductionmentioning
confidence: 99%
“…Cao et al [103] introduced DeepFD, a learning-based fault localization framework that identifies and diagnoses faults in DL programs. This framework utilizes a learning approach to determine fault types by observing runtime features during DNN training.…”
Section: Deep Learning Technique For Bug Detectionmentioning
confidence: 99%
“…D4D leverages a powerful representation of the deep learning model, allowing for the automatic learning of semantic features from both dynamic and static sources [83,103,172,204].…”
Section: Chapter 1 Introductionmentioning
confidence: 99%
See 1 more Smart Citation