A Generic Layer Pruning Method for Signal Modulation Recognition Deep Learning Models

TL;DR

This paper introduces a novel layer pruning technique for deep learning models in signal modulation recognition, significantly reducing model complexity while maintaining high accuracy, thus facilitating practical deployment in communication systems.

Contribution

A new generic layer pruning method that decomposes models into blocks, identifies essential layers, and fine-tunes compact models for improved efficiency in signal classification.

Findings

Outperforms state-of-the-art pruning baselines.

Reduces model size and computational complexity.

Maintains high classification accuracy across datasets.

Abstract

With the successful application of deep learning in communications systems, deep neural networks are becoming the preferred method for signal classification. Although these models yield impressive results, they often come with high computational complexity and large model sizes, which hinders their practical deployment in communication systems. To address this challenge, we propose a novel layer pruning method. Specifically, we decompose the model into several consecutive blocks, each containing consecutive layers with similar semantics. Then, we identify layers that need to be preserved within each block based on their contribution. Finally, we reassemble the pruned blocks and fine-tune the compact model. Extensive experiments on five datasets demonstrate the efficiency and effectiveness of our method over a variety of state-of-the-art baselines, including layer pruning and channel pruning methods.