Architectural Backdoors in Neural Networks · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/2206.07840/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMTArchitectural Backdoors in Neural Networks — mentioned by arnaudsmhttps://news.ycombinator.com/item?id=35315921urn:hackerpapers:mention:2206.07840:35315921Sun, 26 Mar 2023 16:00:35 GMT<p>Weights can be retrained but cannot be explained. Recent papers even demonstrated backdoors : https://arxiv.org/abs/2206.07840 Therefore a real open source model needs a public dataset and training code.</p><p><a href="https://hackerpaper.kaipfister.de/p/2206.07840/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.LG