Unverified50% confidenceFactExact time
In 2012, AlexNet reduced the Top-5 error rate from 25.8% to 16.4% in the ImageNet competition (ILSVRC), trained using two NVIDIA GTX 580 GPUs.
1
Sources
50%
Confidence
Long-term
Relevance
9/10/2026
First Seen
Sources
Related Entities
Related Claims
Unverified2012年Hinton团队的AlexNet以top-5错误率从26%骤降至15%的成绩震惊学界70% similarUnverified残差连接支撑了152层超深网络的训练，并在2015年ImageNet竞赛中以低于4%的错误率首次超越人类水平（约5%）66% similarUnverified2012年ImageNet竞赛中CNN架构将错误率从26%降至15%66% similarUnverified2012年Hinton团队的AlexNet以top-5错误率从26%降至15%的成绩，ResNet（2015，微软研究院）通过残差连接解决深层网络梯度消失问题，EfficientNet（2019，Google Brain）通过复合缩放系数优化网络65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/887263API
curl https://kongchang.com/api/v1/knowledge/claims/887263MCP
get_claim(id=887263)