Learnable audio encryption for untrusted outsourcing machine learning services

Po Wen Chi, Pin Hsin Hsiao

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Applying machine learning to problems has become an avoidable trend. With the help of machine learning techniques, people can make predictions more accurately and can get more benefits. However, the machine learning technique relies on training from lots of data. That is, data should be open to the machine learning service provider. Considering the user privacy issue, the release of user data is not acceptable. In this paper, we propose an approach to take care both the machine learning feature and the data privacy. We focus on audio data and propose an audio encryption technique to keep audio data credential. In the meantime, we make the encrypted audio be able to be trained though machine learning service providers.

Original languageEnglish
Title of host publicationProceedings - 2019 14th Asia Joint Conference on Information Security, AsiaJCIS 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages151-156
Number of pages6
ISBN (Electronic)9781728125565
DOIs
Publication statusPublished - 2019 Aug
Event14th Annual Asia Joint Conference on Information Security, AsiaJCIS 2019 - Kobe, Japan
Duration: 2019 Aug 12019 Aug 2

Publication series

NameProceedings - 2019 14th Asia Joint Conference on Information Security, AsiaJCIS 2019

Conference

Conference14th Annual Asia Joint Conference on Information Security, AsiaJCIS 2019
Country/TerritoryJapan
CityKobe
Period2019/08/012019/08/02

Keywords

  • Audio encryption
  • Learnable encryption
  • Machine learning

ASJC Scopus subject areas

  • Software
  • Information Systems and Management
  • Computer Networks and Communications
  • Safety, Risk, Reliability and Quality

Fingerprint

Dive into the research topics of 'Learnable audio encryption for untrusted outsourcing machine learning services'. Together they form a unique fingerprint.

Cite this