Discovering symmetry invariants and conserved quantities by interpreting siamese neural networks
Sebastian J. Wetzel, Roger G. Melko, Joseph Scott, Maysum Panju, Vijay Ganesh
Abstract
The authors train a Siamese neural network to decide whether two different descriptions describe the same physical object. The neural network learns to identify the objects by calculating the underlying invariants and conserved quantities.
Topics & Concepts
Artificial neural networkComputer scienceArtificial intelligenceSymmetry (geometry)Conserved quantityTheoretical computer scienceSequence (biology)Invariant (physics)MathematicsAlgorithmInterpretation (philosophy)Conserved sequenceFeature (linguistics)GeneralizationDeep neural networksComputational Physics and Python ApplicationsQuantum many-body systemsModel Reduction and Neural Networks