機器學習對抗攻擊是一顆“定時炸彈”

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"如果你一直在關注有關人工智能的新聞，你可能已經聽說過或看到過經修改的熊貓、海龜和停車標誌的圖像，這些圖像在人們眼中看似平凡，但卻能引起人工智能系統的異常行爲。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"這些被稱爲"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2018\/12\/27\/deep-learning-adversarial-attacks-ai-malware\/","title":"","type":null},"content":[{"type":"text","text":"“對抗樣本”或“對抗攻擊"}]},{"type":"text","text":"”圖像及其"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2019\/04\/29\/ai-audio-adversarial-examples\/","title":"","type":null},"content":[{"type":"text","text":"音頻"}]},{"type":"text","text":"和文本"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2019\/04\/02\/ai-nlp-paraphrasing-adversarial-attacks\/","title":"","type":null},"content":[{"type":"text","text":"對應物"}]},{"type":"text","text":"，在機器學習領域引起了越來越多的興趣和關注。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"然而，儘管針對"},{"type":"link","attrs":{"href":"https:\/\/bdtechtalks.com\/2020\/07\/15\/machine-learning-adversarial-examples\/","title":"","type":null},"content":[{"type":"text","text":"對抗機器學習"}]},{"type":"text","text":"的研究不斷增加，但數據表明，在解決現實世界應用中的對抗攻擊方面進展甚微。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"隨着機器學習的迅速普及，科技界必須制定一份保證人工智能系統免遭對抗攻擊的路線圖。否則的話，對抗機器學習將會是一場災難。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/59\/7a\/596da7623f1cfc13bbab6c80906b6c7a.jpg","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"人工智能研究人員發現，通過在停車標誌上貼上黑白小貼紙，就能使它們對計算機視覺算法不可見（來源：arxiv.org）。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"是什麼讓對抗攻擊有所不同"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"每種軟件都有其獨特的安全漏洞，並隨着軟件發展的新趨勢而產生新的威脅。比如， SQL 注入攻擊開始流行，因爲帶數據庫後臺的 Web 應用開始取代靜態網站。由於大量採用瀏覽器端腳本語言，導致了跨站腳本攻擊。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"緩衝區溢出攻擊利用 C 等編程語言處理內存分配的方式，覆蓋關鍵變量，在目標計算機上執行惡意代碼。反序列化攻擊利用了諸如 Java 和 Python 等編程語言在應用程序和進程之間傳輸信息的方式的缺陷。"}]}]}