TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

How ML Model Data Poisoning Works in 5 Minutes

66 pointsby R41about 1 year ago

6 comments

fxtentacleabout 1 year ago
&quot;How ML Model Data Poisoning Works&quot;<p>It doesn&#x27;t. The mentioned Nightshade tool is useless. Does anyone have any example of successful model data poisoning?
评论 #39809053 未加载
Eisensteinabout 1 year ago
None of the cases of data poisoning it presented seemed effective in doing very much, except the MS case, and that was so flawed in implementation that it was a example of how not to deploy something.<p>&gt; Developers need to limit the public release of technical project details including data, algorithms, model architectures, and model checkpoints that are used in production.<p>Haven&#x27;t we learned that more eyes to find flaws is better than locking things down?
bee_riderabout 1 year ago
&gt; In 2016, Microsoft released their chatbot named Tay on Twitter to learn from human interactions by posting comments. But after the release, it started to act crazy.<p>&gt; It started using vulgar language and making hateful comments. This was one of the first incidents of data poisoning.<p>Is this true? I remember when this happened but I thought the story was that 4chan basically found an “echo” type debug command or something like that. The ML mode wasn’t being trained to say bad things, it was just being sent some kind of repeat-after-me command and then the things it was told to repeat were bad.<p>It seems odd that somebody would write a whole blog post without bothering to check that, though, so maybe I’m mis-remembering?
评论 #39809235 未加载
评论 #39809162 未加载
theszabout 1 year ago
I recently made a comment that neural models cannot provide chain of reason, while symbolic methods can: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=39759033">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=39759033</a><p>The vulnerability in the post is directly linked to that inability, in my opinion.
评论 #39812010 未加载
stanleykmabout 1 year ago
When these articles pop up on HN at least there seems to be a lot of focus on training poisoning. While intellectually interesting, it seems less useful or practical than defeating inference.
sonorous_subabout 1 year ago
how to train self-smashing looms