Skip to main navigation Skip to search Skip to main content

Plum: Exploration and Prioritization of Model Repair Strategies for Fixing Deep Learning Models

Research output: Chapters, Conference Papers, Creative and Literary WorksRGC 32 - Refereed conference paper (with host publication)peer-review

Abstract

The accuracy of DL models may not meet the user's expectations. To tackle this problem, existing work proposed diverse approaches, such as using more optimized training processes and training samples to evolve the model structure or parameters of such faulty DL models. In this paper, we present Plum, a novel hyperheuristic approach to fixing deep learning models. Plum generates a set of DL model candidates by applying low-level repair strategies. It then evaluates and prioritizes repair strategies based on their overall fixing effects exhibited by these model candidates and outputs a fixed DL model by applying the top-ranked repair strategy. We also formulate a novel repair strategy to show the compatibility of Plum in incorporating new repair strategies. The experiment on five DL models showed that Plum achieved improvements in test accuracy by 2.49% and 3.11% on the CIFAR-10 and CIFAR-100 datasets over the baselines and outperformed Apricot and MODE, two previous state-of-the-art deep learning repair techniques.
Original languageEnglish
Title of host publicationProceedings - 2021 8th International Conference on Dependable Systems and Their Applications, DSA 2021
PublisherIEEE
Pages140-151
ISBN (Electronic)978-1-6654-4391-3
ISBN (Print)978-1-6654-4392-0
DOIs
Publication statusPublished - 2021
Event8th International Conference on Dependable Systems and Their Applications, DSA 2021 - Yinchuan, China
Duration: 11 Sept 202112 Sept 2021

Publication series

NameProceedings - International Conference on Dependable Systems and Their Applications
ISSN (Print)2767-6676
ISSN (Electronic)2767-6684

Conference

Conference8th International Conference on Dependable Systems and Their Applications, DSA 2021
PlaceChina
CityYinchuan
Period11/09/2112/09/21

Research Keywords

  • Debugging
  • Deep neural networks
  • hyperheuristic
  • Model evolution
  • Model repair
  • Strategy prioritization

RGC Funding Information

  • RGC-funded

Fingerprint

Dive into the research topics of 'Plum: Exploration and Prioritization of Model Repair Strategies for Fixing Deep Learning Models'. Together they form a unique fingerprint.

Cite this