کنفرانس مهندسی برق ایران

صفحه اصلی / بیست و نهمین کنفرانس مهندسی برق ایران

Robust Object Detection Against Adversarial Perturbations with Gabor Filter

نویسندگان :

Mohammad Parsa Karimi¹ Abdollah Amirkhani² Shahriar B. Shokouhi³

1- دانشگاه علم و صنعت ایران 2- دانشگاه علم و صنعت ایران 3- دانشگاه علم و صنعت ایران

کلمات کلیدی :

adversarial attack, deep neural network, robustness, Gabor filter

چکیده :

Adversarial attacks are one of the most critical threats in the machine learning field, which raises doubts about the application of deep neural networks (DNNs). Despite the recent advances in DNNs, the adversarial robustness in DNNs has yet to reach an acceptable level, especially against different kinds of perturbations. In this paper, we aim to enhance the robustness of object detection against adversarial perturbations. To this end, we adversarially train YOLOv3 model with different backbones by means of parameterized Gabor convolutional layers. To assess the robustness of our trained models, we have adopted TOG vanishing, TOG fabrication, and TOG mislabeling adversarial attacks. These perturbations are crafted on PASCAL VOC and MSCOCO datasets to simulate three types of targeted specificity, including object-vanishing, object-fabrication, and object-mislabeling, respectively. Extensive evaluations demonstrate that our model equipped with the Gabor filters gain consideration adversarial robustness in addition to the high generalization performance on clean data.

لیست مقالات

لیست مقالات بایگانی شده

بهبود بازدهی انرژی در اینترنت اشیاء باند باریک با وفقی‌سازی لینک به کمک یادگیری عمیق

سمانه امیریان - محمدعلی سبقتی

Modeling of Photo-thermoelectric Current Effects in Phase Change Material based Optical Nano Dipole Antenna Energy Transducer

Daniyal Khosh Maram - Seyed Asad Amirhosseini

Weak GPS Signal Acquisition Based on Wavelet Transform Denoising and Deep Learning Method

Navid Moradi - Mohsen Nezhadshahbodaghi - Mohammad-Reza Mosavi

Non-contact Radar Technology and Machine Learning for Automated Sleep Apnea-Hypopnea Syndrome Detection

ُSaman Faridsoltani - Mohaddeseh Sadeghi - Zahra Rahmani - Somayyeh Chamaani

Establishment of a Virtual Power Plant in Grid for Maximizing Producers' Profits and Minimizing Pollutant Emissions and Investment Costs

Amir Hossein Gholami - Amir Abulfazl Suratgar - Mohammad Bagher Menhaj - Mohammad Reza Hesamzadeh

Outage Analysis of Distributed Relaying NOMA in Cognitive Radio Networks

Zahra Doorbash - Ali Jamshidi

Single-Channel Recursive Speech Separation with Unknown Speaker Count by Mask Estimation

Hadi Alizadeh - Rahil Mahdian Toroghi - Hassan Zareian

Image-Based Self-Localization Using Differential Observation Angle Based on Real-World Features

Seyed Mohammad Bagher Seyedin - Mahdi Goodarzi - Fereidoon Behnia

پیشنهاد یک ساختار جدید AC/DC مبتنی بر مبدل‌های SEPIC و CUK بهبودیافته برای کاربرد شارژر موتورسیکلت‌های الکتریکی

سجاد قابلی ثانی - رحیم عجبی فرشباف - میثم صادقی - محمد خدایاری

Integration of Deep Learning Techniques in Stock Market Forecasting: xLSTM-CNN with RevIN and Adaptive Wavelet Denoising

Alireza Mohammadi - Ali Doustmohammadi - Masoud Shafiee

بیشتر

ثمین همایش، سامانه مدیریت کنفرانس ها و جشنواره ها - نگارش 42.0.4