🚀 SPENDE Improved

An Improved Graph-to-Sequence Model for Synthesis Recipe Prediction

Poster ID: C-MSE5

Arash Ahmadi, Danial Ebrahimzadeh, Sarah Sharif, Yaser M. Banad

✨ Key Improvement

Chemical Data Augmentation

Baseline Accuracy: 96.91%

Improved Accuracy: 97.23%

Improvement: +0.33%

Variance Reduction: -25%

Original Training: 617 samples

Augmented Training: 1,851 samples

Augmentation Factor: 3x

💡 Click and drag to rotate • Scroll to zoom • Tap nodes for details