EasyDeepLearn

What is compound scaling in EfficientNet?

hard

Answer

  • Scale depth (d), width (w), and resolution (r) together by a single coefficient φ: d = α^φ, w = β^φ, r = γ^φ, with αβ2γ2{\alpha} \cdot {\beta}^{2} \cdot {\gamma}^{2} ≈ 2 to keep FLOPs proportional to 2^φ.
  • Better than scaling one dimension at a time.
  • Combined with a NAS-found base (EfficientNet-B0) and MBConv blocks (mobile inverted bottleneck), EfficientNet-B7 hit 84% ImageNet top-1 in 2019 at 8.4x fewer params than the previous SOTA.
Check yourself — multiple choice
  • Scale only depth
  • Jointly scale depth, width, and resolution with a compound coefficient — Pareto-optimal accuracy/FLOPs
  • EfficientNet uses no NAS
  • Only scales width

EfficientNet compound scaling: joint depth/width/resolution → Pareto-optimal.

#architectures#computer-vision

Practise Deep Learning

214 interview questions in this topic.

Related questions