Hittauppsatser.se

Uppsats

Contrastive Learning For Molecular Representation

H

HANI, SALAM, LINDER, JONATHAN

Chalmers tekniska högskola / Institutionen för data och informationsteknik

Publicerad: 2025

Språk: Engelska

Nyckelord

klicka för att söka
⌕Artificial Intelligence⌕Machine Learning⌕AI⌕Deep-learning⌕benchmarking⌕Reinforcement Learning⌕Latent Space⌕t-SNE⌕transfer learning⌕self-supervised learning⌕LSTM⌕Contrastive learning⌕DL⌕Long Short-Term Memory⌕RNN⌕diversity⌕computer science⌕computer engineering⌕Representation learning⌕data augmentation⌕Embeddings⌕Generative models⌕recurrent neural networks⌕Neural architecture design⌕hyperparameter tuning⌕NT-Xent loss⌕Negative log-likelihood⌕pca⌕UMAP⌕drug discovery⌕De novo molecular generation⌕In silico screening⌕Molecular design⌕Molecular representations⌕Molecular fingerprints⌕SMILES⌕SMILES enumeration⌕SMILES randomization⌕Subgraph sampling⌕Canonicalization⌕Chemical space⌕Physicochemical properties⌕Tanimoto similarity⌕Fréchet ChemNet Distance⌕FCD⌕Synthetic accessibility⌕SA⌕Quantitative Estimate of Drug-likeness⌕QED⌕Stereoisomers⌕Stereochemistry⌕Stereocenters⌕Tautomerism⌕validity⌕Novelty⌕Internal diversity⌕IntDiv⌕ChEMBL35⌕MOSES benchmark⌕GuacaMol benchmark⌕Pharmaceutical AI⌕Cheminformatics⌕AstraZeneca⌕REINVENT⌕Bioactivity prediction

Sammanfattning

This thesis explores the integration of contrastive learning into REINVENT, AstraZeneca’s in-house generative model for molecular design, with the aim of improving the model’s understanding of chemical equivalence between different SMILESrepresentations of the same compound. To this end, a contrastive learning framework was developed, incorporating SMILES-based data augmentation techniques such as enumeration and subgraphing. The framework was evaluated on three datasets: a proprietary baseline derived from ChEMBL35, and the publicly available MOSES and GuacaMol datasets. To assess the impact of architectural design on performance, multiple model architectures were investigated, including a newly introduced intermediate architecture. Results indicate that the intermediate architecture consistently achieves higher validity across all datasets, but tends to reduce novelty. Furthermore, using multiple augmentation strategies improved the model’s ability to generate chemically diverse and novel compounds, as measured by metrics such as novelty and Fréchet ChemNet Distance (FCD). These findings suggest that contrastive learning can offer measurable benefits in de novo molecule generation, although its effectiveness may depend heavily on architecture and dataset-specific tuning.

Information

Författare
HANI, SALAM, LINDER, JONATHAN
Lärosäte / institution
Chalmers tekniska högskola / Institutionen för data och informationsteknik
Publiceringsdatum
2025
Uppsatstyp
H
Språk
Engelska
Nyckelord
⌕Artificial Intelligence⌕Machine Learning⌕AI⌕Deep-learning⌕benchmarking⌕Reinforcement Learning⌕Latent Space⌕t-SNE⌕transfer learning⌕self-supervised learning⌕LSTM⌕Contrastive learning⌕DL⌕Long Short-Term Memory⌕RNN⌕diversity⌕computer science⌕computer engineering⌕Representation learning⌕data augmentation⌕Embeddings⌕Generative models⌕recurrent neural networks⌕Neural architecture design⌕hyperparameter tuning⌕NT-Xent loss⌕Negative log-likelihood⌕pca⌕UMAP⌕drug discovery⌕De novo molecular generation⌕In silico screening⌕Molecular design⌕Molecular representations⌕Molecular fingerprints⌕SMILES⌕SMILES enumeration⌕SMILES randomization⌕Subgraph sampling⌕Canonicalization⌕Chemical space⌕Physicochemical properties⌕Tanimoto similarity⌕Fréchet ChemNet Distance⌕FCD⌕Synthetic accessibility⌕SA⌕Quantitative Estimate of Drug-likeness⌕QED⌕Stereoisomers⌕Stereochemistry⌕Stereocenters⌕Tautomerism⌕validity⌕Novelty⌕Internal diversity⌕IntDiv⌕ChEMBL35⌕MOSES benchmark⌕GuacaMol benchmark⌕Pharmaceutical AI⌕Cheminformatics⌕AstraZeneca⌕REINVENT⌕Bioactivity prediction

Utforska vidare

Liknande uppsatser

Uppsatser med liknande ämnen och nyckelord.