Inductive transfer of knowledge: application of multi-task learning and feature net approaches to model tissue-air partition coefficients
| Autor: | Gilles Marcou, Igor V. Tetko, Alexandre Varnek, Igor I. Baskin, Anil Kumar Pandey, Cédric Gaudin |
|---|---|
| Přispěvatelé: | Chimie de la matière complexe (CMC), Université de Strasbourg (UNISTRA)-Institut de Chimie du CNRS (INC)-Centre National de la Recherche Scientifique (CNRS) |
| Rok vydání: | 2009 |
| Předmět: |
Quantitative structure–activity relationship
Informatics Databases Factual Property (programming) Computer science General Chemical Engineering Multi-task learning Quantitative Structure-Activity Relationship Library and Information Sciences Machine learning computer.software_genre 01 natural sciences Models Biological 03 medical and health sciences Inductive transfer Artificial Intelligence Feature (machine learning) Animals Humans Tissue Distribution Least-Squares Analysis Organic Chemicals 030304 developmental biology 0303 health sciences Artificial neural network business.industry Air Linear model General Chemistry Function (mathematics) 0104 chemical sciences Computer Science Applications Rats 010404 medicinal & biomolecular chemistry Linear Models Artificial intelligence Neural Networks Computer business computer [CHIM.CHEM]Chemical Sciences/Cheminformatics |
| Zdroj: | Journal of Chemical Information and Modeling Journal of Chemical Information and Modeling, American Chemical Society, 2009, 49 (1), pp.133-144. ⟨10.1021/ci8002914⟩ |
| ISSN: | 1549-9596 1549-960X |
| DOI: | 10.1021/ci8002914⟩ |
| Popis: | Two inductive knowledge transfer approaches - multitask learning (MTL) and Feature Net (FN) - have been used to build predictive neural networks (ASNN) and PLS models for 11 types of tissue-air partition coefficients (TAPC). Unlike conventional single-task learning (STL) modeling focused only on a single target property without any relations to other properties, in the framework of inductive transfer approach, the individual models are viewed as nodes in the network of interrelated models built in parallel (MTL) or sequentially (FN). It has been demonstrated that MTL and FN techniques are extremely useful in structure-property modeling on small and structurally diverse data sets, when conventional STL modeling is unable to produce any predictive model. The predictive STL individual models were obtained for 4 out of 11 TAPC, whereas application of inductive knowledge transfer techniques resulted in models for 9 TAPC. Differences in prediction performances of the models as a function of the machine-learning method, and of the number of properties simultaneously involved in the learning, has been discussed. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|