> ML_ALGORITHM // GRAPH-ISOMORPHISM-NETWORKS-GIN_v1.0

Graph Isomorphism Network (GIN)

Provably maximally expressive graph neural network architecture achieving the theoretical upper bound of the 1-Weisfeiler-Lehman graph isomorphism test.

Expressive Graph Neural Networksgraph-relationalmoderate-posthocmedium (1k-100k)

Back to All Algorithms

Computational Complexity

Training Complexity:O(|E| * d + |V| * d^2)

Inference Complexity:O(|E| * d)

Hardware Profile

CPU Friendly:Yes

Requires GPU:Yes

Memory Footprint:low

Interpretability & Data

Interpretability Tier:moderate-posthoc

Training Data Needs:medium (1k-100k)

Interpretability Assessment

Theoretically proven to match the maximum distinguishing power of the classical 1-WL graph isomorphism test.

Suitable Tasks & Supported Modalities

Suitable Tasks:

graph classificationmolecular property prediction

Supported Modalities:

graph

Implementing Libraries

torch-geometric

Deep Graph Library (DGL)DMLC / AWS / NYU · v2.2.1

View Spec

Foundational Literature

How Powerful are Graph Neural Networks? (Graph Isomorphism Network - GIN)Keyulu Xu, Weihua Hu (2019) · International Conference on Learning Representations (ICLR)

Common Pitfalls & Warnings

Cannot count or distinguish simple non-isomorphic graphs containing identical 1-WL multiset trees (e.g., distinguishing a 6-cycle from two triangles)