On the Importance of Pooling Layer Tuning for Profiling Side-Channel Analysis

Lichao Wu*, Guilherme Perin

*Corresponding author for this work

Research output: Chapter in Book/Conference proceedings/Edited volumeConference contributionScientificpeer-review

1 Citation (Scopus)

Abstract

In recent years, the advent of deep neural networks opened new perspectives for security evaluations with side-channel analysis. Profiling attacks now benefit from capabilities offered by convolutional neural networks, such as dimensionality reduction and the inherent ability to reduce the trace desynchronization effects. These neural networks contain at least three types of layers: convolutional, pooling, and dense layers. Although the definition of pooling layers causes a large impact on neural network performance, a study on pooling hyperparameters effect on side-channel analysis is still not provided in the academic community. This paper provides extensive experimental results to demonstrate how pooling layer types and pooling stride and size affect the profiling attack performance with convolutional neural networks. Additionally, we demonstrate that pooling hyperparameters can be larger than usually used in related works and still keep good performance for profiling attacks on specific datasets.

Original languageEnglish
Title of host publicationApplied Cryptography and Network Security Workshops - ACNS 2021 Satellite Workshops, AIBlock, AIHWS, AIoTS, CIMSS, Cloud S and P, SCI, SecMT, and SiMLA, 2021, Proceedings
EditorsJianying Zhou, Chuadhry Mujeeb Ahmed, Lejla Batina, Sudipta Chattopadhyay, Olga Gadyatskaya, Chenglu Jin, Jingqiang Lin, Eleonora Losiouk, Bo Luo, Suryadipta Majumdar, Mihalis Maniatakos, Daisuke Mashima, Weizhi Meng, Stjepan Picek, Masaki Shimaoka, Chunhua Su, Cong Wang
PublisherSpringer
Pages114-132
Number of pages19
ISBN (Print)9783030816445
DOIs
Publication statusPublished - 2021
Eventsatellite workshops held around the 19th International Conference on Applied Cryptography and Network Security, ACNS 2021, 3rd International Workshop on Application Intelligence and Blockchain Security, AIBlock 2021, 2nd International Workshop on Artificial Intelligence in Hardware Security, AIHWS 2021, 3rd International Workshop on Artificial Intelligence and Industrial IoT Security, AIoTS 2021, 1st International Workshop on Critical Infrastructure and Manufacturing System Security, CIMSS 2021, 3rd International Workshop on Cloud Security and Privacy, Cloud S and P 2021, 2nd International Workshop on Secure Cryptographic Implementation, SCI 2021, 2nd International Workshop on Security in Mobile Technologies, SecMT 2021, 3rd International Workshop on Security in Machine Learning and its Applications, SiMLA 2021 - Virtual, Online
Duration: 21 Jun 202124 Jun 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12809 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conferencesatellite workshops held around the 19th International Conference on Applied Cryptography and Network Security, ACNS 2021, 3rd International Workshop on Application Intelligence and Blockchain Security, AIBlock 2021, 2nd International Workshop on Artificial Intelligence in Hardware Security, AIHWS 2021, 3rd International Workshop on Artificial Intelligence and Industrial IoT Security, AIoTS 2021, 1st International Workshop on Critical Infrastructure and Manufacturing System Security, CIMSS 2021, 3rd International Workshop on Cloud Security and Privacy, Cloud S and P 2021, 2nd International Workshop on Secure Cryptographic Implementation, SCI 2021, 2nd International Workshop on Security in Mobile Technologies, SecMT 2021, 3rd International Workshop on Security in Machine Learning and its Applications, SiMLA 2021
CityVirtual, Online
Period21/06/2124/06/21

Keywords

  • Convolutional neural networks
  • Deep learning
  • Pooling
  • Side-channel analysis

Fingerprint

Dive into the research topics of 'On the Importance of Pooling Layer Tuning for Profiling Side-Channel Analysis'. Together they form a unique fingerprint.

Cite this