ENOS: Energy-Aware Network Operator Search for Hybrid Digital and Compute-in-Memory DNN Accelerators
Shamma Nasrin, Ahish Shylendra, Yuti Kadakia, Nick Iliev, Wilfred, Gomes, Theja Tulabandhula, and Amit Ranjan Trivedi

TL;DR
ENOS is a scalable, gradient-based framework that optimally integrates inference operators and computing modes in DNN accelerators to balance energy consumption and accuracy, demonstrated on digital and compute-in-memory architectures.
Contribution
ENOS introduces a continuous optimization approach for layer-wise operator selection, enabling energy-accuracy trade-offs in hybrid DNN accelerators.
Findings
ENOS effectively balances energy and accuracy in DNNs.
It is scalable to large models with minimal training overhead.
Demonstrated on ShuffleNet and SqueezeNet with CIFAR datasets.
Abstract
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.…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Neural Network Applications · Adversarial Robustness in Machine Learning · Domain Adaptation and Few-Shot Learning
MethodsXavier Initialization · Fire Module · Residual Connection · Pointwise Convolution · 1x1 Convolution · Channel Shuffle · Dropout · Convolution · Average Pooling · Dense Connections
