Abstract:Effective transfer learning for molecular property prediction has shown considerable strength in addressing insufficient labeled molecules. Many existing methods either disregard the quantitative relationship between source and target properties, risking negative transfer, or require intensive training on target tasks. To quantify transferability concerning task-relatedness, we propose Principal Gradient-based Measurement (PGM) for transferring molecular property prediction ability. First, we design an optimization-free scheme to calculate a principal gradient for approximating the direction of model optimization on a molecular property prediction dataset. We have analyzed the close connection between the principal gradient and model optimization through mathematical proof. PGM measures the transferability as the distance between the principal gradient obtained from the source dataset and that derived from the target dataset. Then, we perform PGM on various molecular property prediction datasets to build a quantitative transferability map for source dataset selection. Finally, we evaluate PGM on multiple combinations of transfer learning tasks across 12 benchmark molecular property prediction datasets and demonstrate that it can serve as fast and effective guidance to improve the performance of a target task. This work contributes to more efficient discovery of drugs, materials, and catalysts by offering a task-relatedness quantification prior to transfer learning and understanding the relationship between chemical properties. Transfer learning is known to enhance molecular property prediction in limited data sets, however, negative transfer due to insufficient relatedness between source and target tasks continues to be a major challenge. Here, the authors develop a principal gradient-based measurement to evaluate the quantitative transferability from the source property to the target property before applying transfer learning, significantly improving the transfer learning performance.

Enhancing Molecular Property Prediction through Task-Oriented Transfer Learning: Integrating Universal Structural Insights and Domain-Specific Knowledge

Pushing the Boundaries of Molecular Property Prediction for Drug Discovery with Multitask Learning BERT Enhanced by SMILES Enumeration

Improving molecular property prediction through a task similarity enhanced transfer learning strategy

Scalable Multi-Task Transfer Learning for Molecular Property Prediction

Enhancing drug property prediction with dual-channel transfer learning based on molecular fragment

Fast and Effective Molecular Property Prediction with Transferability Map

ADMET Property Prediction Via Multi-Task Graph Learning under Adaptive Auxiliary Task Selection

Advanced deep learning methods for molecular property prediction

Impact of Domain Knowledge and Multi-Modality on Intelligent Molecular Property Prediction: A Systematic Survey

MG-BERT: leveraging unsupervised atomic representation learning for molecular property prediction

From molecules to scaffolds to functional groups: building context-dependent molecular representation via multi-channel learning

Pre-training Transformers for Molecular Property Prediction Using Reaction Prediction

FG-BERT: a Generalized and Self-Supervised Functional Group-Based Molecular Representation Learning Framework for Properties Prediction.

Tetrahedral Molecular Pretraining for Enhanced Property Prediction

Molecular Graph Representation Learning Integrating Large Language Models with Domain-specific Small Models

Integrating Chemical Language and Molecular Graph in Multimodal Fused Deep Learning for Drug Property Prediction

Ensemble Model With Bert,Roberta and Xlnet For Molecular property prediction

CIMG-BERT: Pretraining Bidirectional Transformers with Chemistry Knowledge for Molecular Property Prediction

Domain-Adversarial Multi-Task Framework for Novel Therapeutic Property Prediction of Compounds

Knowledge-based BERT: a method to extract molecular features such as computational chemists

Knowledge-based BERT: a Method to Extract Molecular Features Like Computational Chemists