کد مقاله کد نشریه سال انتشار مقاله انگلیسی نسخه تمام متن
6863754 1439520 2018 10 صفحه PDF دانلود رایگان
عنوان انگلیسی مقاله ISI
Understanding adversarial training: Increasing local stability of supervised models through robust optimization
ترجمه فارسی عنوان
درک آموزش مشارکتی: افزایش پایداری محلی مدل های تحت نظارت از طریق بهینه سازی قوی
کلمات کلیدی
نمونه های مخالف، بهینه سازی قوی، مدل های تحت نظارت غیر پارامتری، یادگیری عمیق،
موضوعات مرتبط
مهندسی و علوم پایه مهندسی کامپیوتر هوش مصنوعی
چکیده انگلیسی
We show that adversarial training of supervised learning models is in fact a robust optimization procedure. To do this, we establish a general framework for increasing local stability of supervised learning models using robust optimization. The framework is general and broadly applicable to differentiable non-parametric models, e.g., Artificial Neural Networks (ANNs). Using an alternating minimization-maximization procedure, the loss of the model is minimized with respect to perturbed examples that are generated at each parameter update, rather than with respect to the original training data. Our proposed framework generalizes adversarial training, as well as previous approaches for increasing local stability of ANNs. Experimental results reveal that our approach increases the robustness of the network to existing adversarial examples, while making it harder to generate new ones. Furthermore, our algorithm improves the accuracy of the networks also on the original test data.
ناشر
Database: Elsevier - ScienceDirect (ساینس دایرکت)
Journal: Neurocomputing - Volume 307, 13 September 2018, Pages 195-204
نویسندگان
, , ,