Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

topology-aware network pruning

Topology-aware network pruning is a deep neural network compression method that eliminates redundant components, such as weights or channels, by evaluating the structural connectivity and relational dependencies across the entire network architecture. Unlike conventional pruning approaches that assess parameters in isolation or rely purely on localized importance metrics, topology-aware pruning models the network as an interconnected structural graph to capture global pathways, cross-layer interactions, and architecture-wide data flow. By leveraging this topological information to guide compression decisions, the technique selectively removes elements to reduce computational and memory overhead while preserving the essential structural integrity and predictive accuracy of the model.

1 item

Topology-Aware Network Pruning using Multi-stage Graph Embedding and Reinforcement Learning

Topology-Aware Network Pruning using Multi-stage Graph Embedding and Reinforcement Learning

Sixing Yu, Arya Mazaheri, Ali Jannesari

OrganizationsIowa State UniversityTechnische Universität Darmstadt

Why you should read this

Proposes a multi-stage graph embedding framework that captures hierarchical neural network topologies to automatically learn optimal, FLOP-constrained pruning policies via reinforcement learning across diverse architectures.

Model compression is an essential technique for deploying deep neural networks (DNNs) on power and memory-constrained resources. However, existing model-compression methods often rely on human expertise and focus on parameters’ local importance, ignoring the rich topology information within DNNs. In this paper, we propose a novel multi-stage graph embedding technique based on graph neural networks (GNNs) to identify DNN topologies and use reinforcement learning (RL) to find a suitable compression policy. We performed resource-constrained (i.e., FLOPs) channel pruning and compared our approach with state-of-the-art model compression methods. We evaluated our method on various models from typical to mobile-friendly networks, such as ResNet family, VGG-16, MobileNet-v1/v2, and ShuffleNet. Results show that our method can achieve higher compression ratios with a minimal fine-tuning cost yet yields outstanding and competitive performance. The code is open-sourced at https://github.com/yusx-swapp/GNN-RL-Model-Compression.

Added

2026-10-03