Skip to Content
MethodsMethods

Methods

Machine learning methods commonly used in scientific applications.

Overview

Scientific ML draws from various methodologies:

  • Graph Neural Networks — For molecular and materials data
  • Transformers & LLMs — Language models for science
  • Generative Models — Designing new molecules and materials
  • Active Learning — Efficient experimental design
  • Uncertainty Quantification — Knowing what we don’t know

Quick Start

MethodDomain Examples
Graph Neural NetworksMolecular property prediction, materials
TransformersReaction prediction, property prediction
Generative ModelsMolecule design, materials discovery
Active LearningSelf-driving labs, experimental optimization

Resources by Method

Graph Neural Networks

ResourceFocus
DeepChem Tutorials GNNs for molecules
PyTorch Geometric General GNN framework

Transformers & LLMs

ResourceFocus
Transformers for Chemistry LLMs in chemistry/materials
awesome-scientific-language-models Scientific LLMs

Generative Models

ResourceFocus
AI4Chemistry Course Generative models for chemistry

Conferences

ConferenceFocusTiming
NeurIPSAI4Science workshopDecember
ICMLML4Science workshopsJuly
ICLRScience-focused papersMay