ASU 与马里兰大学提出 SEAL：面向异构 3D 存内计算加速器的安全感知层映射
Security-Aware Layer Mapping Reduces Fault Vulnerability In 3D Compute-In-Memory (ASU, U. of Maryland)
亚利桑那州立大学与马里兰大学研究人员发表 SEAL 框架，通过将关键神经网络层分配到更抗故障的器件与堆叠层组合，联合优化功耗、性能、面积与故障脆弱性。基于 ResNet-18 和 Qwen2.5-3B 的评估显示，相比安全盲映射，故障脆弱性分别最多降低 3.9 倍和 7.8 倍。
这条信号在站内保留事实摘要与战略研判。完整内容和上下文请以原文为准；融资、并购与技术效果以原始披露为准。
Researchers at Arizona State University and University of Maryland published “SEAL: Multitenant Security-Aware Layer Mapping for Heterogeneous 3-D CiM Accelerators.” Abstract: “Heterogeneous 3-D compute-in-memory architectures can combine different memory technologies across stacked tiers to accelerate deep neural network inference, but multitenant deployments introduce security risks through shared power delivery networks. The researchers developed SEAL, a security-aware... » read more The post Security-Aware Layer Mapping Reduces Fault Vulnerability In 3D Compute-In-Memory (ASU, U. of Maryland) appeared first on Semiconductor Engineering .
来源只提供摘要，完整内容请打开原文。
来源：Semiconductor Engineering · semiengineering.com