JetBrains Research — наука, меняющая мир

Adversarial Examples Are Not Bugs, They Are Features

Adversarial examples have attracted significant attention in machine learning, but the reasons for their existence and pervasiveness remain unclear.

At the upcoming seminar we will:

- demonstrate that adversarial examples can be tied to the presence of non-robust features;
- establish their widespread existence in standard datasets;
- present a simple setting to observe a misalignment with the human notion of robustness;

Докладчик: Рауф Курбанов.

Ссылка на статью: https://arxiv.org/abs/1905.02175.

Язык доклада: русский.

Дата и время: 5-е ноября, 20:00-21:30.

Место: Таймс, аудитория 204.

Видео с предыдущих семинаров доступно по ссылке: http://bit.ly/MLJBSeminars