run_iterative_bridge_pipeline

Table of contents

  1. Overview
  2. Function Signature
  3. Example Usage

Overview

The run_iterative_bridge_pipeline function performs multiple rewiring rounds, progressively modifying the graph before training a final SelectiveGCN. It uses a fast SGC-based classifier for early iterations to speed up computation.

Function Signature 

from bridge.rewiring import run_iterative_bridge_pipeline

Refer to the Python docstring for full parameter details. Important options include n_rewire (number of rewiring iterations) and use_sgc to enable the fast SGC classifier during the iterative phase.

Example Usage 

results = run_iterative_bridge_pipeline(
    g=g,
    P_k=P_k,
    n_rewire=5,
    device="cuda",
)
print(results["selective"]["test_acc"])