T
traeai
Sign in

模型

ResNet-50

别名:ResNet50

经典的 50 层残差网络图像分类模型,广泛用于计算机视觉基准测试。

已跟踪 2 条高相关材料

TraeAI 观察

最近变化

2026-05-18 · 仅翻转ResNet 50中2个符号位可使ImageNet准确率下降99.8%

为什么值得关注

ResNet-50 被反复提及时,通常意味着它正在影响产品路线、开发者工作流或 AI 产业判断。这个页面把分散材料合并成一个可持续更新的观察入口。

对抗攻击AI安全ResNet位级攻击模型脆弱性

相关材料

已收录 2 条与 ResNet-50 相关的内容,按评分排序。

AI Models May Be More Fragile Than We Think

AI Models May Be More Fragile Than We Think

Last Week in AI182 字 (约 1 分钟)
82

Research shows that flipping just 2 sign bits in neural networks can cause a 99.8% drop in ResNet 50's ImageNet accuracy, revealing serious vulnerabilities in AI models.

入选理由:仅翻转ResNet 50中2个符号位可使ImageNet准确率下降99.8%

FeaturedVideo#AI Security#Neural Networks#Model Vulnerability#Adversarial Attacks英文
AI Models May Be More Fragile Than We Think

AI Models May Be More Fragile Than We Think

Last Week in AI182 字 (约 1 分钟)
75

Flipping just two signed bits in ResNet-50 can cause a 99.8% drop in ImageNet accuracy without any data access or training, revealing an extreme vulnerability called 'deep neural lesion' in neural networks.

入选理由:攻击仅需翻转 2 个参数的符号位,无需接触训练数据或进行优化

FeaturedVideo#Neural Network Security#Adversarial Attack#Model Robustness#Bit-level Attack#ResNet英文

跨材料问答 · ResNet-50

回答基于:ResNet-50 相关 2 条材料
    0 / 500

    AI may generate inaccurate information. Please verify important content.