Operation-Aware Soft Channel Pruning using Differentiable Masks
This work addresses the need for efficient model compression in deep learning, offering an incremental improvement by integrating operation-aware pruning into a differentiable framework without requiring fine-tuning.
The paper tackles the problem of compressing deep neural networks by proposing a data-driven channel pruning algorithm that jointly considers batch normalization and ReLU operations to prune channels with high deactivation probabilities, achieving outstanding performance in accuracy given the same resources compared to state-of-the-art methods.
We propose a simple but effective data-driven channel pruning algorithm, which compresses deep neural networks in a differentiable way by exploiting the characteristics of operations. The proposed approach makes a joint consideration of batch normalization (BN) and rectified linear unit (ReLU) for channel pruning; it estimates how likely the two successive operations deactivate each feature map and prunes the channels with high probabilities. To this end, we learn differentiable masks for individual channels and make soft decisions throughout the optimization procedure, which facilitates to explore larger search space and train more stable networks. The proposed framework enables us to identify compressed models via a joint learning of model parameters and channel pruning without an extra procedure of fine-tuning. We perform extensive experiments and achieve outstanding performance in terms of the accuracy of output networks given the same amount of resources when compared with the state-of-the-art methods.