-
Group Theory
- Describing Symmetry Transformations
- Invariant and Equivariant Maps
- Group Representations
- Irreducible Representations
- Illustration of Irreps via a Discrete Example
-
Tensor Product and Clebsh-Gordan Coefficients
-
SO(3) Group and Spherical Harmonics
- 3D Rotation and its Irreducible Representations for Functions on a Sphere
- Spherical Harmonics Projection and Equivariant Networks
- Connection with Angular Momentum
- Steerable CNNs
- 3D Steerable CNNs
- General E(2) Steerable CNNs
- Group Equivariant CNNs on Homogeous Spaces
- Wigner-Eckart Theorem for G-CNNs
- Steerable Partial Differential Operators
- escnn
- Coordinate Indepedent CNNs
-
Symmetry Breaking
-
Empirical Benefits and Expense of Equivariance versus Invariance
-
Universality of Equivariant Neural Architectures
-
Frame Averaging as an Alternative for Equivariance
-
Approximate Equivariance
- Restricted Boltzmann Machines
- Feed-Forward Neural Networks
- Convolutional Neural Networks
- Autoregressive and Recurrent Neural Networks
- Graph Neural Networks
- Molecules - Single Geometry
- Molecules - Multiple Geometries
- Superfluids and Homogeneous Electron Gas
- Invariant Networks, Data Augmentation
- Invariant Networks, Rotation with Wigner-D
- Equivariant Networks
- Exchange-Correlation Energy Density Functional Learning
- Kinetic Energy Density Functional Learning
- Invariant Methods (
$\ell=0$ Scalar Features) - Equivariant Methods (
$\ell=1$ Vector Features) - Equivariant Methods (
$\ell \geq 1$ Tensor Features)
- Learn the Distribution of Low-Energy Geometries
- Predict the Equilibrium Ground-State Geometry
- Generate Coordinate Matrices
- Generate SE(3)-Invariant Features
- ML Force Fields
- Enhanced Sampling
- Coarse Graining (CG)
- Represent Tetrahedral Chirality
- Represent Conformational Flexibility
- Two-Stage Learning
- End-To-End Learning
- Structure Representation: Coordinates
- Structure Representation: Frames
- Structure Representation: Internal Angles
- Material Representation: Multi-Edge Graphs
- Material Representation: Multi-Edge Graphs and Fully-Connected Graphs
- Material Representation: 3D Voxel Grids
- Material Representation: Fractional Coordinates
- Material Representation: Multi-Edge Graphs
- Material Representation: 1D Spectral Data
- Material Representation: Local Environment Descriptor
- Material Representation: Local Environment Descriptor
- Material Representation: Multi-Edge Graphs
- Predict Coordinates
- Predict Interatomic Distances
- Predict Rotation, Translation, and Torsions
- Autoregressively Generate Relative Position-Related Variables
- Generate Coordinates with Diffusion Models
- Invariant Methods
- Equivariant Methods
- Approximately Equivariant Methods
- Sequential Multiscale Processing
- Parallel Multiscale Processing
- Geometry Deformation
- Learned Adaptive Remeshing
- Data Augmentation
- Equivariant Architectures
- System Identification
- Tomography for Medical Imaging
- Fluid Assimilation
- History Matching
- Full Waveform Inversion for Geophysics
- Shape Design for Planes
- Ion Thruster Design
- Controlled Nuclear Fusion: 1,2
- Nanophotonics
- Battery Design: 1,2
- Chip Manufacturing
- Existing XAI Methods
- OOD in AI for Quantum Mechanics
- OOD in AI for Density Functional Theory
- OOD in AI for Molecular Science
- OOD in AI for Protein Science
- OOD in AI for Material Science
- OOD in AI for Chemical Interactions
- OOD in AI for Partial Differential Equations
- Self-Supervised Learning
- 2D Molecular Graph
- 3D Molecular Graph
- SSL for PDE Solvers
- Single-Modal Foundation Models
- Protein Geometry
- Molecule Topology, Geometry, String
- Natural Language Guided Scientific Discovery (LLMs for Science)
- Integrating Multi-Modalities
- Bi-Encoder, e.g., CLIP
- Text2Mol
- Mol-Instructions
- CLAMP
- BioTranslator
- ProteinDT
- Su et al.
- Joint-Representation, e.g., DALL-E
- KV-PLM
- MolT5
- Text+Chem T5
- MolXPT
- Smiles2Actions
- ProtNLM
- Adapting LLM to Scientific Domains
- Pre-Training, e.g., Galactica
- Fine-Tuning
- BioMedLM (SSL)
- Med-PaLM (SSL)
- DrugChat (SL)
- Prompting/In-Context Learning
- Chemistry Benchmark
- Jablonka et al.
- KEBLM
- ChatDrug
- BO-LIFT
- CancerGPT
- SynerGPT
- Integrating Multi-Modalities
- Uncertainty Quantification in Machine Learning
- Gaussian Process
- Bayesian Neural Network and Approximate Inference
- Other Techniques
- Uncertainty Quantification for Graph Learning
- Uncertainty Quantification in AI for Science
- UQ in Molecular Property Prediction
- UQ in Protein-Compound Binding Affinity
- UQ in Density Functional Theory
- UQ in Partial Differential Equations
- UQ Libraries