Novel machine learning framework for thermal conductivity prediction by crystal graph convolution embedded ensemble
Problem
Predicting the lattice thermal conductivity of crystal materials is hard because usable datasets are scarce, limiting data-driven materials science.
Why it matters
Data scarcity is stated as the main obstacle to machine learning in materials science.
What the AI did
An ensemble framework embedding material crystal graphs into a graph convolutional neural network was built to train on and predict lattice thermal conductivities of crystal materials.
Results
not stated in source (only qualitative claim of superior learning and generalization ability)
Impact
Proposes a data-efficient ML framework for predicting crystal thermal conductivity.
Record
| Venue | SmartMat |
|---|---|
| Published | 2021 |
| Work type | article |
| Authors | Yajing Sun, Wenping Hu |
| Citations | 25 (OpenAlex, at import time) |
| References | 45 |
| Model or method | crystal graph convolution embedded ensemble (graph convolutional neural network) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1002/smm2.1074 |
| Open access | gold |
| Keywords | graph convolutional network, ensemble learning, crystal graph, thermal conductivity, materials informatics |
Importance rating
real but narrower advance
Data scarcity limits ML in materials science per abstract; framework addresses a genuine methodological need but no quantified results or scale stated.
More in Engineering & Materials
| # | Paper | Year | Citations | Importance |
|---|---|---|---|---|
| 1 | Scaling deep learning for materials discovery | 2023 | 1,471 |
8
|
| 2 | Application of Machine Learning in Material Synthesis and Property Prediction | 2023 | 165 |
7
|
| 3 | Machine learning-guided discovery of ionic polymer electrolytes for lithium metal batteries | 2023 | 115 |
7
|
| 4 | Machine Learning for Structural Health Monitoring of Aerospace Structures: A Review | 2025 | 44 |
7
|
| 5 | Power System Transient Stability Assessment Based on Machine Learning Algorithms and Grid Topology | 2023 | 42 |
7
|