ENOS: Energy-Aware Network Operator Search for Hybrid Digital and Compute-in-Memory DNN Accelerators
ENOS: Energy-Aware Network Operator Search for Hybrid Digital and Compute-in-Memory DNN Accelerators
复制标题
ENOS：能源感知网络运营商寻找混合数字和内存计算 DNN 加速器
DOI：
--
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
A. Trivedi
中科院分区：
文献类型：
--
作者：
Shamma Nasrin;A. Shylendra;Yuti Kadakia;N. Iliev;Wilfred Gomes;Theja Tulabandhula;A. Trivedi
这项工作提出了一种新颖的能量感知网络算子搜索（ENOS）方法，以解决深度神经网络（DNN）加速器的能量 - 精度权衡问题。近年来，人们提出了新的推理算子来提高DNN的计算效率。除了这些算子，与之相应的新型计算模式也得到了探索。然而，DNN算子的简化总是以降低精度为代价的，特别是在复杂的处理任务上。我们提出的ENOS框架允许推理算子和计算模式进行最优的逐层集成，以实现能量和精度的理想平衡。ENOS中的搜索被表述为一个连续优化问题，可以使用典型的梯度下降方法求解，因此对于更大的DNN具有可扩展性，且训练成本的增加极小。我们在两种设置下对ENOS进行了描述。在第一种设置中，对于数字加速器，我们讨论了可重新配置为不同算子的乘累加（MAC）核上的ENOS。讨论并比较了具有单级和双层优化目标的ENOS训练方法。我们还讨论了ENOS中的一种顺序算子分配策略，该策略在一个训练步骤中只学习一层的分配，从而在收敛到最优算子分配方面具有更大的灵活性。此外，遵循贝叶斯原理，还提出了一种基于采样的ENOS变分模式。我们在CIFAR10和CIFAR100数据集上的流行DNN（ShuffleNet和SqueezeNet）上对ENOS进行了特性描述。
This work proposes a novel Energy-Aware Network Operator Search (ENOS) approach to address the energy-accuracy trade-offs of a deep neural network (DNN) accelerator. In recent years, novel inference operators have been proposed to improve the computational efficiency of a DNN. Augmenting the operators, their corresponding novel computing modes have also been explored. However, simplification of DNN operators invariably comes at the cost of lower accuracy, especially on complex processing tasks. Our proposed ENOS framework allows an optimal layer-wise integration of inference operators and computing modes to achieve the desired balance of energy and accuracy. The search in ENOS is formulated as a continuous optimization problem, solvable using typical gradient descent methods, thereby scalable to larger DNNs with minimal increase in training cost. We characterize ENOS under two settings. In the first setting, for digital accelerators, we discuss ENOS on multiply-accumulate (MAC) cores that can be reconfigured to different operators. ENOS training methods with single and bi-level optimization objectives are discussed and compared. We also discuss a sequential operator assignment strategy in ENOS that only learns the assignment for one layer in one training step, enabling greater flexibility in converging towards the optimal operator allocations. Furthermore, following Bayesian principles, a sampling-based variational mode of ENOS is also presented. ENOS is characterized on popular DNNs ShuffleNet and SqueezeNet on CIFAR10 and CIFAR100.
DOI：
10.1109/tcsi.2021.3064033
发表时间：
2021-01
期刊：
IEEE Transactions on Circuits and Systems I: Regular Papers
影响因子：
--
作者：
Shamma Nasrin;Diaa Badawi;A. Cetin;Wilfred Gomes;A. Trivedi
通讯作者：
Shamma Nasrin;Diaa Badawi;A. Cetin;Wilfred Gomes;A. Trivedi
DOI：
--
发表时间：
2016-11
期刊：
ArXiv
影响因子：
--
作者：
Barret Zoph;Quoc V. Le
通讯作者：
Barret Zoph;Quoc V. Le
内存计算颠倒：学习算子协同设计可扩展性的视角
DOI：
10.23919/date51398.2021.9474119
发表时间：
2021
期刊：
Automation & Test in Europe Conference & Exhibition (DATE
影响因子：
--
作者：
Nasrin, Shamma;Shukla, Priyesh;Jaisimha, Shruthi;Trivedi, Amit Ranjan
通讯作者：
Trivedi, Amit Ranjan