[ SLANG ][ AI CULTURE ][ AGENT SECURITY ]
无固定对应词
这个概念在中文里没有一个可以直接对应的固定说法，见下方说明。英文原词：Data Poisoning
对机器学习系统的一种攻击：故意污染训练数据，使模型出错更多或学到隐藏行为。后门攻击是其中一种，输入含特定触发条件时模型会按设定方式反应。
例句
"“论文显示，只要少量被投毒的文档就能在模型里埋下后门。”"
来源 & 用法
首次出现
Unknown (the concept predates current AI slang; no dated first use is documented here)
巅峰时期
2020s (AI security)
当前状态
主流俚语
MENTIONS OVER TIME
DOCUMENTED SINCE
Unknown (the concept predates current AI slang; no dated first use is documented here)
NOW
Radar observation data not yet available for this term.
想看完整的英文词条——分类标签、热度趋势、平台分布、社区投票等更多数据？前往 Data Poisoning 英文完整词条 →
AURA IMPACT ⓘ
+1 AURA
TREND STATUS
→ STEADY
EDITORIAL ESTIMATE
CULTURE CATEGORY
[ SLANG ][ AI CULTURE ][ AGENT SECURITY ]
TOP PLATFORMS
Not enough Radar data yet.
WHEN DID YOU FIRST HEAR THIS?
CLICK AN OPTION BELOW TO CAST YOUR VOTE.
[ VOTE TO REVEAL COMMUNITY RESULTS ]
OTHER LANGUAGES