XMA: a crossbar-aware multi-task adaption framework via shift-based mask learning method
XMA: a crossbar-aware multi-task adaption framework via shift-based mask learning method
复制标题
DOI：
10.1145/3489517.3530458
复制
发表时间：
2022-07
期刊：
影响因子：
--
通讯作者：
Fan Zhang;Li Yang;Jian Meng;Jae-sun Seo;Yu Cao;Deliang Fan
中科院分区：
文献类型：
--
作者：
Fan Zhang;Li Yang;Jian Meng;Jae-sun Seo;Yu Cao;Deliang Fan
ReRAM crossbar阵列作为一种高并行、快速、节能的结构，引起了人们的广泛关注，特别是在加速深度神经网络（DNN）对特定任务的推理方面。然而，由于权重重编程的高能耗和ReRAM单元的低耐久性问题，尚未很好地探索使交叉杆阵列适应多个任务。在本文中，我们首次提出了XMA，一种新的交叉杆感知的基于移位的掩码学习方法，用于ReRAM交叉杆DNN加速器中的多任务自适应。XMA利用流行的基于掩码的学习算法的优点来减轻灾难性遗忘，并为基于冻结主干模型的每个新任务学习特定于任务的、纵横列方式和基于移位的多级掩码，而不是最常用的元素方式的二进制掩码。通过我们的Crossbar感知设计创新，适应新任务所需的掩蔽操作可以在现有的基于Crossbar的卷积引擎中实现，具有最小的硬件/内存开销，更重要的是，不需要功耗高的单元重新编程，不像以前的工作。大量的实验结果表明，与最先进的多任务自适应Piggyback方法[1]相比，XMA平均提高了3.19%的准确率，同时节省了96.6%的内存开销。此外，通过消除单元重新编程，XMA的能效比Piggyback高出约4.3倍。
ReRAM crossbar array as a high-parallel fast and energy-efficient structure attracts much attention, especially on the acceleration of Deep Neural Network (DNN) inference on one specific task. However, due to the high energy consumption of weight re-programming and the ReRAM cells' low endurance problem, adapting the crossbar array for multiple tasks has not been well explored. In this paper, we propose XMA, a novel crossbar-aware shift-based mask learning method for multiple task adaption in the ReRAM crossbar DNN accelerator for the first time. XMA leverages the popular mask-based learning algorithm's benefit to mitigate catastrophic forgetting and learn a task-specific, crossbar column-wise, and shift-based multi-level mask, rather than the most commonly used element-wise binary mask, for each new task based on a frozen backbone model. With our crossbar-aware design innovation, the required masking operation to adapt for a new task could be implemented in an existing crossbar-based convolution engine with minimal hardware/memory overhead and, more importantly, no need for power-hungry cell re-programming, unlike prior works. The extensive experimental results show that, compared with state-of-the-art multiple task adaption Piggyback method [1], XMA achieves 3.19% higher accuracy on average, while saving 96.6% memory overhead. Moreover, by eliminating cell re-programming, XMA achieves ~4.3x higher energy efficiency than Piggyback.