Low-bit Quantization for Deep Graph Neural Networks with Smoothness-aware Message Propagation

TL;DR

This paper introduces a low-bit quantization method for deep GNNs that maintains accuracy and mitigates oversmoothing, enabling efficient processing and significant speedups in resource-constrained environments.

Contribution

The paper proposes a novel quantization approach with a smoothness-aware message propagation mechanism, effectively reducing model size and addressing oversmoothing in deep GNNs.

Findings

Achieves high accuracy with INT2 quantization across all GNN stages.

Demonstrates over 5x speedup in inference with low-bit models.

Outperforms existing quantization and deep GNN methods.

Abstract

Graph Neural Network (GNN) training and inference involve significant challenges of scalability with respect to both model sizes and number of layers, resulting in degradation of efficiency and accuracy for large and deep GNNs. We present an end-to-end solution that aims to address these challenges for efficient GNNs in resource constrained environments while avoiding the oversmoothing problem in deep GNNs. We introduce a quantization based approach for all stages of GNNs, from message passing in training to node classification, compressing the model and enabling efficient processing. The proposed GNN quantizer learns quantization ranges and reduces the model size with comparable accuracy even under low-bit quantization. To scale with the number of layers, we devise a message propagation mechanism in training that controls layer-wise changes of similarities between neighboring nodes. This objective is incorporated into a Lagrangian function with constraints and a differential multiplier method is utilized to iteratively find optimal embeddings. This mitigates oversmoothing and suppresses the quantization error to a bound. Significant improvements are demonstrated over state-of-the-art quantization methods and deep GNN approaches in both full-precision and quantized models. The proposed quantizer demonstrates superior performance in INT2 configurations across all stages of GNN, achieving a notable level of accuracy. In contrast, existing quantization approaches fail to generate satisfactory accuracy levels. Finally, the inference with INT2 and INT4 representations exhibits a speedup of 5.11 $\times$ and 4.70 $\times$ compared to full precision counterparts, respectively.