Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Molecular Classification Using Hyperdimensional Graph Classification

Pere Verges, Igor Nunes, Mike Heddes, Tony Givargis, Alexandru Nicolau

TL;DR

The paper tackles efficient molecular graph classification by introducing a hyperdimensional computing approach that encodes graphs with star-subgraph patterns and emphasizes infrequent substructures via RefineHD. The method achieves competitive AUC compared to state-of-the-art graph models while delivering large training and inference speedups, enabling real-time analysis in chemoinformatics and cancer cell screening. It systematically compares against WL, GNNs, and previous HDC methods, showing robust performance across eleven anticancer datasets and demonstrating scalability up to tens of thousands of dimensions. The work highlights HDC as a promising, resource-efficient alternative for graph learning in biology, with practical impact for online learning and embedded applications.

Abstract

Our work introduces an innovative approach to graph learning by leveraging Hyperdimensional Computing. Graphs serve as a widely embraced method for conveying information, and their utilization in learning has gained significant attention. This is notable in the field of chemoinformatics, where learning from graph representations plays a pivotal role. An important application within this domain involves the identification of cancerous cells across diverse molecular structures. We propose an HDC-based model that demonstrates comparable Area Under the Curve results when compared to state-of-the-art models like Graph Neural Networks (GNNs) or the Weisfieler-Lehman graph kernel (WL). Moreover, it outperforms previously proposed hyperdimensional computing graph learning methods. Furthermore, it achieves noteworthy speed enhancements, boasting a 40x acceleration in the training phase and a 15x improvement in inference time compared to GNN and WL models. This not only underscores the efficacy of the HDC-based method, but also highlights its potential for expedited and resource-efficient graph learning.

Molecular Classification Using Hyperdimensional Graph Classification

TL;DR

The paper tackles efficient molecular graph classification by introducing a hyperdimensional computing approach that encodes graphs with star-subgraph patterns and emphasizes infrequent substructures via RefineHD. The method achieves competitive AUC compared to state-of-the-art graph models while delivering large training and inference speedups, enabling real-time analysis in chemoinformatics and cancer cell screening. It systematically compares against WL, GNNs, and previous HDC methods, showing robust performance across eleven anticancer datasets and demonstrating scalability up to tens of thousands of dimensions. The work highlights HDC as a promising, resource-efficient alternative for graph learning in biology, with practical impact for online learning and embedded applications.

Abstract

Our work introduces an innovative approach to graph learning by leveraging Hyperdimensional Computing. Graphs serve as a widely embraced method for conveying information, and their utilization in learning has gained significant attention. This is notable in the field of chemoinformatics, where learning from graph representations plays a pivotal role. An important application within this domain involves the identification of cancerous cells across diverse molecular structures. We propose an HDC-based model that demonstrates comparable Area Under the Curve results when compared to state-of-the-art models like Graph Neural Networks (GNNs) or the Weisfieler-Lehman graph kernel (WL). Moreover, it outperforms previously proposed hyperdimensional computing graph learning methods. Furthermore, it achieves noteworthy speed enhancements, boasting a 40x acceleration in the training phase and a 15x improvement in inference time compared to GNN and WL models. This not only underscores the efficacy of the HDC-based method, but also highlights its potential for expedited and resource-efficient graph learning.
Paper Structure (26 sections, 4 figures, 10 tables, 1 algorithm)

This paper contains 26 sections, 4 figures, 10 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Graph Learning Methods
  4. Kernel methods
  5. Graph Neural Network (GNN)
  6. Fingerprints
  7. Subgraph mining
  8. Hyperdimensional Computing (HDC)
  9. Information Representation
  10. Operations
  11. Similarity metric
  12. Superposition
  13. Binding
  14. Permutation
  15. Hyperdimensional Classification
  16. ...and 11 more sections

Figures (4)

  • Figure 1: Benzene to graph representation 4812459.
  • Figure 2: Proposed graph encoding.
  • Figure 3: Method comparison techniques AUC.
  • Figure 4: Method scalability.