Graph kernels for chemoinformatics – a critical discussion
2011
Graph Kernels in Chemoinformatics
publication
Author Information
Author(s): Rupp Matthias
Primary Institution: Helmholtz Zentrum München, Deutsches Forschungszentrum für Umwelt & Gesundheit
Hypothesis
The study analyzes the use, advantages, and drawbacks of graph kernels in chemoinformatics.
Conclusion
Graph kernels provide state-of-the-art performance in chemoinformatics tasks but have limitations in terms of runtime and indirect solution expression.
Supporting Evidence
- Kernel-based machine learning is widely applied in chemoinformatics.
- Molecular graph kernels allow the adaptation of methods from graph theory to structure graphs.
- Graph kernels are defined directly on graphs and have strong properties that can be exploited.
Takeaway
This study looks at how graph kernels help in understanding molecules better, but they can be slow and complicated.
Methodology
The study compares kernel-based approaches with other methodologies and provides examples of applications.
Limitations
The solutions provided by graph kernels are expressed indirectly and can have high runtimes.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website