Cut Inner Layers: A Structured Pruning Strategy for Efficient U-Net GANs
This work addresses the challenge of model efficiency for generative models, which is incremental as it adapts pruning techniques from discriminative models to U-Net GANs.
The study tackled the problem of compressing overparameterized generative models by applying structured pruning specifically to U-Net generators in conditional GANs, finding that many unnecessary filters in innermost layers can be pruned, and it outperformed global pruning baselines in tasks like image-to-image translation and speech-driven talking face generation.
Pruning effectively compresses overparameterized models. Despite the success of pruning methods for discriminative models, applying them for generative models has been relatively rarely approached. This study conducts structured pruning on U-Net generators of conditional GANs. A per-layer sensitivity analysis confirms that many unnecessary filters exist in the innermost layers near the bottleneck and can be substantially pruned. Based on this observation, we prune these filters from multiple inner layers or suggest alternative architectures by completely eliminating the layers. We evaluate our approach with Pix2Pix for image-to-image translation and Wav2Lip for speech-driven talking face generation. Our method outperforms global pruning baselines, demonstrating the importance of properly considering where to prune for U-Net generators.