SCAN-Edge: Finding MobileNet-speed Hybrid Networks for Diverse Edge Devices via Hardware-Aware Evolutionary Search
Hung-Yueh Chiang, Diana Marculescu

TL;DR
SCAN-Edge introduces a hardware-aware evolutionary neural architecture search framework that designs hybrid networks optimized for diverse edge devices, effectively matching actual latency across CPUs, GPUs, and accelerators.
Contribution
It unifies NAS for various hardware by jointly searching for multiple operation types and employs a hardware-aware evolutionary algorithm to efficiently explore the large search space.
Findings
Hybrid networks match MobileNetV2 latency on multiple devices.
The framework effectively accommodates diverse hardware architectures.
Search process accelerates through improved search space quality.
Abstract
Designing low-latency and high-efficiency hybrid networks for a variety of low-cost commodity edge devices is both costly and tedious, leading to the adoption of hardware-aware neural architecture search (NAS) for finding optimal architectures. However, unifying NAS for a wide range of edge devices presents challenges due to the variety of hardware designs, supported operations, and compilation optimizations. Existing methods often fix the search space of architecture choices (e.g., activation, convolution, or self-attention) and estimate latency using hardware-agnostic proxies (e.g., FLOPs), which fail to achieve proclaimed latency across various edge devices. To address this issue, we propose SCAN-Edge, a unified NAS framework that jointly searches for self-attention, convolution, and activation to accommodate the wide variety of edge devices, including CPU-, GPU-, and hardware…
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
TopicsEvolutionary Algorithms and Applications · Energy Harvesting in Wireless Networks · Energy Efficient Wireless Sensor Networks
MethodsPointwise Convolution · Depthwise Convolution · Depthwise Separable Convolution · Convolution · Average Pooling · Batch Normalization · Inverted Residual Block · 1x1 Convolution
