Adversarial machine learning
فێربوونی ئامێری دژبەرانە
Proposed Kurdish term — under reviewزاراوەی پێشنیارکراو — لە ژێر پێداچوونەوەدایە Definitionپێناسە
The study of attacks on systems and of defences against them. NIST groups the attacks into evasion (tricky inputs), poisoning (bad ), privacy attacks (stealing what a model has learned) and misuse.
لێکۆڵینەوەیە لە دژی سیستەمەکانی و بەرگری لێیان. NIST دەکاتە چوار گرووپ: خۆدزینەوە ( فێڵاوی)، ( خراپ)، هێرشی سەر (دزینی ئەوەی فێری بووە) و خراپ بەکارهێنان.
Picturesوێنەکان
The life of an AI system (collect data, train, finished model, use) with the attacks that hit each stage: poisoning, privacy attacks, evasion and misuse.ژیانی سیستەمێکی ژیریی دەستکرد (کۆکردنەوەی دراوە، ڕاهێنان، مۆدێلی تەواوکراو، بەکارهێنان) لەگەڵ ئەو هێرشانەی لە هەر قۆناغێکدا دەدەن: ژەهراویکردن، هێرشی سەر تایبەتێتی، خۆدزینەوە و خراپ بەکارهێنان.
Exampleنموونە
Testing a vision model with adversarial examples before it is used is part of adversarial machine learning.
تاقیکردنەوەی بە پێش بەکارهێنانی، بەشێکە لە .
Note on the Kurdish termتێبینی لەبارەی زاراوە کوردییەکە
Proposed: the registry's فێربوونی ئامێر (machine learning) + دژبەرانە (adversarial). The four groups follow NIST AI 100-2 E2025 (https://csrc.nist.gov/pubs/ai/100/2/e2025/final).