Multimodal Model Unifies Characterization Of CNT Films
Researchers propose an interpretable multimodal machine learning framework for end-to-end characterization of carbon nanotube (CNT) films, demonstrated in an arXiv preprint (Jan 31, 2026). They fuse SEM-derived morphology descriptors, Raman crystallinity indicators, gas adsorption surface area, and surface resistivity, and train regressors (XGBoost best under leave-one-out CV) to predict film properties. Feature-importances reveal physically meaningful drivers for resistivity and surface area.
Key Points
- 1Extracts SEM morphology features (curvature, orientation, intersection density, void geometry) and fuses Raman, adsorption, resistivity
- 2Demonstrates XGBoost best predictive accuracy under leave-one-out cross-validation for CNT film property regression
- 3Provides interpretable feature importances linking junction transport length, connectivity, and intersection density to properties
Scoring Rationale
Provides practical, interpretable multimodal methods for CNT characterization; limited novelty and based on a single arXiv preprint.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
