ESNet fuses element knowledge-graph embeddings with crystal graph features, achieving a reported band-gap MAE of 0.177 eV on the Materials Project, a small improvement over iComFormer's 0.193 eV.
Atomistic line graph neural network for improved materials property predictions
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Material Property Prediction with Element Attribute Knowledge Graphs and Multimodal Representation Learning
ESNet fuses element knowledge-graph embeddings with crystal graph features, achieving a reported band-gap MAE of 0.177 eV on the Materials Project, a small improvement over iComFormer's 0.193 eV.