Evolutionary Strategy for Optimizing Neural Models for Compound Similarity Estimation
Molecular similarity evaluation is a key aspect of bioinfor- matics and poses a significant challenge when dealing with compounds with unknown structures. In this context, graph neural networks have proven effective in generating representations based on the topology of chemical reactions. However,...
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Articulo |
| Lenguaje: | Inglés |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | http://sedici.unlp.edu.ar/handle/10915/177971 |
| Aporte de: |
| id |
I19-R120-10915-177971 |
|---|---|
| record_format |
dspace |
| spelling |
I19-R120-10915-1779712025-05-06T17:17:49Z http://sedici.unlp.edu.ar/handle/10915/177971 Evolutionary Strategy for Optimizing Neural Models for Compound Similarity Estimation Estrategia evolutiva para la optimización de modelos neuronales de estimación de similaridad entre compuestos Hermann, Tobías J. Vignolo, Leandro D. Gerard, Matias F. 2025-04 2025-04-04T15:22:05Z en Ciencias Informáticas graph neural networks evolutionary computation compound similarity metabolic pathways redes neuronales en grafo computación evolutiva similaridad entre compuesto vías metabólicas Molecular similarity evaluation is a key aspect of bioinfor- matics and poses a significant challenge when dealing with compounds with unknown structures. In this context, graph neural networks have proven effective in generating representations based on the topology of chemical reactions. However, designing these models and selecting their hyperparameters requires exploring a vast range of options. Evolutio- nary algorithms naturally arise as a solution for searching these exten- sive spaces, including the hyperparameter space of neural architectures. This study presents a comparison between a traditional hyperparameter search approach, based on expert knowledge, and a method leveraging evolutionary computation for the same task, specifically in compound similarity estimation. Using a predefined architecture, experiments are conducted to compare both approaches across different datasets. The re- sults indicate that the evolutionary computation-based method success- fully identifies suitable hyperparameters for the evaluated architecture, achieving a performance comparable to the expert-driven approach while eliminating the need for human intervention in the selection process. La evaluación de similaridad molecular es clave en el ámbito de la bioinformática, y representa un reto significativo cuando se trata de compuestos cuya estructura no se conoce. En este contexto, los modelos neuronales en grafos han demostrado ser efectivos para obtener representaciones a partir de la topología de reacciones químicas. No obstante, el diseño de estos modelos, así como la selección de sus hiperparámetros, requiere la evaluación de un extenso rango de opciones. Los algoritmos evolutivos se presentan como una solución natural para explorar estos amplios espacios de búsqueda, incluyendo el espacio de hiperparámetros de las arquitecturas neuronales. Este estudio propone una comparación entre un enfoque tradicional de búsqueda de hiperparámetros, basado en la experiencia del experto, y un método que utiliza computación evolutiva para la misma tarea, específicamente en la estimación de similaridad entre compuestos. Utilizando una arquitectura predefinida, se llevan a cabo experimentos para comparar ambos enfoques en distintos conjuntos de datos. Los resultados indican que el método basado en computación evolutiva logra identificar hiperparámetros adecuados para la arquitectura evaluada, alcanzando un desempeño comparable al del enfoque experto, pero sin requerir la intervención del conocimiento humano para dicha selección. Sociedad Argentina de Informática e Investigación Operativa Articulo Articulo http://creativecommons.org/licenses/by-nc-sa/4.0/ Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0) application/pdf |
| institution |
Universidad Nacional de La Plata |
| institution_str |
I-19 |
| repository_str |
R-120 |
| collection |
SEDICI (UNLP) |
| language |
Inglés |
| topic |
Ciencias Informáticas graph neural networks evolutionary computation compound similarity metabolic pathways redes neuronales en grafo computación evolutiva similaridad entre compuesto vías metabólicas |
| spellingShingle |
Ciencias Informáticas graph neural networks evolutionary computation compound similarity metabolic pathways redes neuronales en grafo computación evolutiva similaridad entre compuesto vías metabólicas Hermann, Tobías J. Vignolo, Leandro D. Gerard, Matias F. Evolutionary Strategy for Optimizing Neural Models for Compound Similarity Estimation |
| topic_facet |
Ciencias Informáticas graph neural networks evolutionary computation compound similarity metabolic pathways redes neuronales en grafo computación evolutiva similaridad entre compuesto vías metabólicas |
| description |
Molecular similarity evaluation is a key aspect of bioinfor- matics and poses a significant challenge when dealing with compounds with unknown structures. In this context, graph neural networks have proven effective in generating representations based on the topology of chemical reactions. However, designing these models and selecting their hyperparameters requires exploring a vast range of options. Evolutio- nary algorithms naturally arise as a solution for searching these exten- sive spaces, including the hyperparameter space of neural architectures.
This study presents a comparison between a traditional hyperparameter search approach, based on expert knowledge, and a method leveraging evolutionary computation for the same task, specifically in compound similarity estimation. Using a predefined architecture, experiments are conducted to compare both approaches across different datasets. The re- sults indicate that the evolutionary computation-based method success- fully identifies suitable hyperparameters for the evaluated architecture, achieving a performance comparable to the expert-driven approach while eliminating the need for human intervention in the selection process. |
| format |
Articulo Articulo |
| author |
Hermann, Tobías J. Vignolo, Leandro D. Gerard, Matias F. |
| author_facet |
Hermann, Tobías J. Vignolo, Leandro D. Gerard, Matias F. |
| author_sort |
Hermann, Tobías J. |
| title |
Evolutionary Strategy for Optimizing Neural Models for Compound Similarity Estimation |
| title_short |
Evolutionary Strategy for Optimizing Neural Models for Compound Similarity Estimation |
| title_full |
Evolutionary Strategy for Optimizing Neural Models for Compound Similarity Estimation |
| title_fullStr |
Evolutionary Strategy for Optimizing Neural Models for Compound Similarity Estimation |
| title_full_unstemmed |
Evolutionary Strategy for Optimizing Neural Models for Compound Similarity Estimation |
| title_sort |
evolutionary strategy for optimizing neural models for compound similarity estimation |
| publishDate |
2025 |
| url |
http://sedici.unlp.edu.ar/handle/10915/177971 |
| work_keys_str_mv |
AT hermanntobiasj evolutionarystrategyforoptimizingneuralmodelsforcompoundsimilarityestimation AT vignololeandrod evolutionarystrategyforoptimizingneuralmodelsforcompoundsimilarityestimation AT gerardmatiasf evolutionarystrategyforoptimizingneuralmodelsforcompoundsimilarityestimation AT hermanntobiasj estrategiaevolutivaparalaoptimizaciondemodelosneuronalesdeestimaciondesimilaridadentrecompuestos AT vignololeandrod estrategiaevolutivaparalaoptimizaciondemodelosneuronalesdeestimaciondesimilaridadentrecompuestos AT gerardmatiasf estrategiaevolutivaparalaoptimizaciondemodelosneuronalesdeestimaciondesimilaridadentrecompuestos |
| _version_ |
1852545711488368640 |