Robust Audio Adversarial Example for a Physical Attack

Authors: Hiromu Yakura, Jun Sakuma

IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In order to confirm the effectiveness of the proposed method, we conducted evaluation experiments. We played and recorded audio adversarial examples generated by the proposed method and verified whether these adversarial examples are recognized as target phrases.
Researcher Affiliation Academia Hiromu Yakura1,2 and Jun Sakuma1,2 1University of Tsukuba 2RIKEN Center for Advanced Intelligence Project
Pseudocode No The paper describes the generation process using mathematical equations but does not include structured pseudocode or algorithm blocks.
Open Source Code Yes Our full implementation is available at https://github.com/hiromu/robust audio ae
Open Datasets Yes For the input sample x, we prepared two different audio clips of four seconds cut from Cello Suite No. 1 by Bach and To The Sky by Owl City. The first clip is the same as the publicly released samples3 of [Carlini and Wagner, 2018]. The second clip is the same as the publicly released samples4 of [Yuan et al., 2018].
Dataset Splits No The paper describes the input audio clips and the collection of impulse responses, but it does not specify any explicit training, validation, or test dataset splits for these.
Hardware Specification Yes First, we played and recorded the adversarial examples using a speaker and a microphone (JBL CLIP2 / Sony ECM-PCV80U) in a meeting room... We also examined whether the generated adversarial examples could attack through the radio using Hack RF One7, a Software Defined Radio (SDR)... received the signal with a portable radio (Sony ICF-P36) in the same room, while the playback was recorded by a microphone (Sony ECM-PCV80U).
Software Dependencies No The paper mentions 'Tensor Flow' as the implementation framework but does not provide a specific version number for it or other software dependencies.
Experiment Setup Yes For optimization, we used Adam [Kingma and Ba, 2015] in the same manner as [Carlini and Wagner, 2018].